Technical ReportPDF Available

A Re-interpretation of the 8.2ky BP Event

Authors:

Abstract

This report gathers together evidence in support of a new hypothesis concerning the 8.2ky BP event. The hypothesis is that the event was triggered not only by changes to the salinity of surface water in the north Atlantic but also by a dramatic oscillation of the Earth in response to the mass redistribution following the emptying of glacial lakes Agassiz and Ojibway, causing massive rises and falls in ocean level relative to the land surface. The report details planetary motion calculations and presents supporting evidence from geology (e.g. lakebed sediments, the 'Black Sea flood', Dead Sea brine), biology (e.g. species diversity patterns, known species spread events), archaeology (e.g. hiatuses in settlement patterns), linguistics (e.g. language spread events) and human prehistory (flood myths). The clear conclusion is that the balance of the evidence is strongly supportive of the hypothesis, and that this event has enormous implications for many branches of research.
A Re-interpretation of the
8.2ky BP Event
a report by
N.H. Thom
University of Nottingham
Faculty of Engineering
2015
i
Contents
1. Introduction
1.1 The 8.2ky BP Event
1.2 Lake Agassiz
1.3 The Archaeological Gap
1.4 Organisation of this Work
1.5 Terminology
2. Physics of the Earth
2.1 Properties of the Earth
2.1.1 Dimensions
2.1.2 Density
2.1.3 Stiffness Modulus
2.1.4 Viscosity
2.1.5 Conclusion
2.2 Isostatic Response to Emptying of Lake Agassiz
2.2.1 Elastic Response
2.2.2 Viscous Response – Mantle
2.2.3 Viscous Response – Core-Mantle Boundary
2.2.4 Centrifugal Force Imbalance
2.2.5 Conclusion
2.3 Planetary Response to Polar Shift
2.3.1 Changes in Mantle Level
2.3.2 Resistance due to Mantle Stiffness
2.3.3 Deduction based on the Chandler Wobble
2.3.4 Non-linearity Effects
2.3.5 Mantle Inhomogeneity
2.3.6 Suggested Planetary Motion
2.3.7 Recovery due to Viscous Flow
2.3.8 Energy Dissipation
2.3.9 Conclusion
2.4 Implications
2.4.1 Predicted Flooding
2.4.2 Local Corrections
2.4.3 Water Flows
2.4.4 Changes to the Land
ii
2.4.5 Possible Similar Episodes in Earlier Epochs
2.4.6 Conclusion
2.5 Summary
3. Geological Evidence
3.1 Earthquakes and Volcanism
3.1.1 Volcanic Eruptions
3.1.2 Changes in Sea Level
3.1.3 Conclusion
3.2 Oceans
3.3 Ice Sheets
3.3.1 George VI Ice Shelf
3.3.2 Amery Ice Shelf
3.3.3 Ross Ice Shelf
3.3.4 Conclusion
3.4 Lakes and Inland Seas
3.4.1 Overview
3.4.2 Total Organic Carbon (TOC)
3.4.3 Other Organic Sediment Indicators
3.4.4 Inorganic Sediment
3.4.5 Pollen
3.4.6 Other Parameters
3.4.7 The American Great Lakes
3.4.8 The African Great Lakes
3.4.9 Lake Chad
3.4.10 Conclusion
3.5 Salt Lakes and Playas
3.5.1 Saline Groundwater
3.5.2 Australian Salt Playas
3.5.3 Saharan Oases
3.5.4 The Tunisian Chotts
3.5.5 The Danakil Depression
3.5.6 The Salton Sea
3.5.7 The Dead Sea
3.5.8 Conclusion
3.6 The Black Sea Basin
3.6.1 Modelling Black Sea History
3.6.2 The Lacustrine-Marine Transition
3.6.3 The Bosphorus
3.6.4 The Dardanelles
3.6.5 Conclusion
3.7 The Aral-Caspian Basin
3.7.1 The Manych Spillway
iii
3.7.2 The Turgay Valley
3.7.3 The Chemistry of the Caspian Sea
3.7.4 The Aral Sea
3.7.5 Conclusion
3.8 Summary
4. Biological Evidence
4.1 The Effect of a Saltwater Flood
4.1.1 Flora
4.1.2 Marine Fauna
4.1.3 Freshwater Fauna
4.1.4 Land Mammals
4.1.5 Land Birds
4.1.6 Land Reptiles
4.1.7 Amphibians
4.1.8 Land Molluscs
4.1.9 Arthropods
4.1.10 Annelids
4.1.11 Conclusion
4.2 Flora
4.2.1 Colonisation Events
4.2.2 Accumulations of Organic Matter
4.2.3 Conclusion
4.3 Marine Fauna
4.3.1 Coral Reefs
4.3.2 Pacific-Atlantic Connection – Central America
4.3.3 Pacific-Atlantic Connection – Trans-Arctic
4.3.4 Red Sea-Mediterranean Connection
4.3.5 Saltwater Lakes on Islands
4.3.6 Conclusion
4.4 Continental Fresh and Brackish Water Fauna
4.4.1 Shell-rich Sediments
4.4.2 Freshwater Lake Ecosystems
4.4.3 The Caspian Sea Ecosystem
4.4.4 Species Dispersal – Caspian Sea
4.4.5 Species Dispersal – Lake Baikal
4.4.6 Species Dispersal – Western Siberia
4.4.7 Species Dispersal – Central Africa
4.4.8 Species Dispersal – West Africa
4.4.9 Species Dispersal – Southern Africa
4.4.10 Conclusion
4.5 Continental Land Fauna
4.5.1 Hotspots where Flooding is Predicted
4.5.2 Low-lying Hotspots above Predicted Flood Level
iv
4.5.3 Conclusion
4.6 Island Fauna
4.6.1 The Caribbean
4.6.2 The Atlantic Ocean
4.6.3 The Indian Ocean
4.6.4 The Malaysian and Indonesian Archipelagos
4.6.5 The Pacific Ocean
4.6.6 The Arctic Ocean
4.6.7 Conclusion
4.7 Summary
5. Archaeological Evidence
5.1 Africa
5.2 The Middle East
5.2.1 The Iraqi Plain
5.2.2 The Jordan Valley and Coastal Palestine
5.2.3 Conclusion
5.3 Europe
5.3.1 The Danube Basin
5.3.2 Southern Europe
5.3.3 North European Plain
5.3.4 European Russia
5.3.5 Conclusion
5.4 Northern Asia
5.4.1 Central Asia
5.4.2 Siberia
5.4.3 Conclusion
5.5 Southern Asia
5.5.1 The Indus Plain
5.5.2 The Ganges Plain
5.5.3 Conclusion
5.6 Eastern Asia
5.6.1 North China
5.6.2 Central China
5.6.3 Conclusion
5.7 South-East Asia
5.8 Australia
5.9 America
5.9.1 North America
5.9.2 South America
5.9.3 Conclusion
5.10 Summary
v
6. Linguistic Evidence
6.1 Africa
6.2 Northern Eurasia
6.2.1 Indo-European
6.2.2 Uralic
6.2.3 Altaic
6.2.4 North Caucasian
6.2.5 Conclusion
6.3 Southern Asia
6.3.1 Dravidian
6.3.2 Austro-Asiatic
6.3.3 Conclusion
6.4 Eastern Asia
6.4.1 Sino-Tibetan
6.4.2 Kadai
6.4.3 Chukchi-Kamchatkan
6.4.4 Conclusion
6.5 Australia
6.6 North America
6.6.1 Na Dené
6.6.2 Siouxan
6.6.3 Caddoan
6.6.4 Algonquian
6.6.5 Iroquoian
6.6.6 Muskogean
6.6.7 Conclusion
6.7 South America
6.8 Summary
7. Historical Evidence
7.1 Overview of Flood Stories
7.1.1 Distribution
7.1.2 Cause of Flooding
7.1.3 Flood Depth
7.1.4 Flood Duration
7.1.5 Means of Escape
7.1.6 Conclusion
7.2 Widespread Flood Stories
7.2.1 The Noah Stories
7.2.2 Brother-Sister Stories
7.2.3 The Old Man and the Muskrat
vi
7.2.4 Other Story Clusters
7.2.5 Conclusion
7.3 Comparison with Predictions
7.3.1 Africa
7.3.2 Europe
7.3.3 Northern Asia
7.3.4 Southern Asia
7.3.5 Eastern Asia
7.3.6 Mainland South-East Asia
7.3.7 Island South-East Asia and the Pacific
7.3.8 Australia
7.3.9 North America
7.3.10 Central America
7.3.11 South America
7.3.12 Conclusion
7.4 Summary
8. Concluding Discussion
8.1 Geology
8.2 Biology
8.3 Archaeology
8.4 Linguistics
8.5 History
8.6 Conclusion
References
Appendixes
A. Polar Shift Calculations
B. Lakebed Sediments
C. Salton Sea / Lake Cahuilla
D. The Dead Sea
E. The Black Sea – Caspian Sea System
F. Caspian Sea Salinity
CHAPTER 1 INTRODUCTION
1
1. Introduction
This report will set out a hypothesis which, if true, radically alters our understanding of
events in the early Holocene, the period during which mankind took their first steps from
a purely hunting and gathering economy to that of settled agriculture. It will propose that
a disaster of extraordinary magnitude struck the planet, including the human occupants of
the planet; yet it is a disaster that has hitherto been invisible to science. The task of this
report therefore is to describe the nature of that disaster and its physical causes in terms
of planetary science, detailing the assumptions, computations and deductions that
underlie the hypothesis, and then to investigate the geological, biological and
anthropological evidence relating to it.
It is intended that this work is sufficiently detailed for it to be evaluated as a stand-
alone document without the need to refer to numerous other texts, although references
will be given throughout. Important computational details are included in appendixes.
1.1 The 8.2ky BP Event
The key claim made in this report is that this well-known climatic event, approximately
8200 years ago, was actually much more catastrophic than has hitherto been recognised.
It is known as the most significant of the climatic fluctuations that have occurred during
the Holocene, the otherwise relatively stable period in the Earth’s history since the end of
the last ice age. The defining signal was discovered in ice cores from Greenland (Alley et
al, 1993; Alley et al, 1997), from which oxygen isotope analysis allowed evaluation of
regional temperatures to be carried out. The result is reproduced in Figure 1.1. Clearly the
moderate temperature reduction between about 8.2ky BP and 8.0ky BP is in no way
comparable to that of the last glacial maximum or the Younger Dryas period, but it is
nevertheless discernible and more significant than any subsequent fluctuations in the
Greenland area.
Since this ice signal was discovered, a large body of supporting evidence has been
produced from sources right across the planet. The Greenland evidence was paralleled in
one of the Antarctic cores (Jouzel et al, 1987); a large temperature reduction (7.8-10C)
has been inferred from a core though a glacier in Tibet (Wang et al, 2002) and cooling is
apparent from the aerosol content in the glacier at the summit of Mount Kilimanjaro in
East Africa (Thompson et al, 2002). Oceanic signals have been found in the Atlantic
(Duplessey et al, 1992; O’Reilly et al, 2004; de Menocal et al, 2000), the Caribbean
(Hughen et al, 1996), the North Sea (Klitgaard-Kristensen et al, 1998), the Norwegian
Sea (Calvo et al, 2002; Dolven et al, 2002; Birks and Koç, 2002), the South China Sea
(Wang et al, 1999) and the western Pacific (Gagan et al, 2002; Yu et al, 2008), all
indicating ocean cooling of around 1-2C and all relating to the same time period,
CHAPTER 1 INTRODUCTION
2
approximately 8.2ky BP. Speliothem evidence from Israel (Bar-Matthews et al, 1999)
and Oman (Fleitmann et al, 2003) is in full agreement; similarly lakebed deposits in
Germany (von Grafenstein et al, 1998) and snail-shell evidence in the UK (Rousseau et
al, 1998). The number of pieces of supporting evidence is now much too great to be listed
here in full; suffice it to say that the reality of the 8.2ky BP event is very widely accepted.
Figure 1.1. Smoothed data from the GISP2 ice core (Alley et al, 1993)
1.2 Lake Agassiz
In the opinion of most, the cause of the 8.2ky BP event is now known. The concentration
of climatic effects in the North Atlantic and the discovery of a band of red silt across the
bottom of the Hudson Strait (Kerwin, 1996), linking Hudson Bay in Canada with the
Labrador Sea, have led researchers straight to a particular and scientifically understood
occurrence, namely the final emptying of two coalesced glacial lakes, Lake Agassiz and
Lake Ojibway, at that time comprising the largest body of fresh water on the planet – see
Figure 1.2.
Lake Agassiz has been much studied by scientists over many decades. It left its
unmistakable mark right across southern Canada and into Minnesota and North Dakota.
The lake formed in the terminal phase of the Pleistocene Era as the Laurentide ice sheet
retreated northwards, trapping meltwater between the ice sheet and the Red River-
Minnesota River watershed on the South Dakota-Minnesota border. Teller (2003) has
described the progress of Lake Agassiz from its inception between 13ky BP and 12ky BP
to its maximum extent, conjoined with Lake Ojibway, in about 8.47ky BP. During this
period the lake level fell on numerous occasions as various elements of the ice sheet gave
way, and the outfall changed from the Red River to the Des Plaines River via Lake
-50
-45
-40
-35
-30
-25
-20
02468101214161820
Date (ky BP)
Air Temperature, Greenland
(degrees C)
Younger Dryas
8.2ky BP event
Glacial
Maximum
ICE AGE Little
Ice Age
CHAPTER 1 INTRODUCTION
3
Michigan and then to the Ottawa River. However these outflow bursts were relatively
insignificant compared to the event that brought Lake Agassiz to an end. Teller (2003)
estimated that the entire volume of the lake, around 163,000km3, emptied into Hudson
Bay in a very short space of time as the final ice barrier gave way.
Figure 1.2. Approximate location of Lake Agassiz-Ojibway in 8.47ky BP
The date most commonly quoted for this event is 8.47ky BP (Barber et al, 1999),
although there is a strong body of opinion that sees this as only the first stage of a two-
stage emptying process. One reason for this is that the Greenland ice core data does not
show any significant change until almost 8.2ky BP, while several other climatic
indicators show the downturn to commence prior to 8.4ky BP, for example sea surface
temperatures deduced by Kim et al (2007) from alkenone paleothermometry applied to a
sediment core from off the North African coast. Core records from the Labrador Sea
(Hillaire-Marcel et al, 2007), through which the escaping waters of Lake Agassiz would
have to have flowed, clearly show a sharp increase in sea ice cover at approximately
8.47ky BP based on dinocyst evidence, possibly impying the presence of an ultra-fresh
layer of water on the surface. Furthermore the detrital carbonate content, a measure that
has been shown to signify reworking of sediments from the Hudson Strait, suggesting
outflow from Hudson Bay, reveals a double-peak pattern, compatible with two distinct
maxima in outlow volume, while St. Onge and Lajeunesse (2007) deduce a two-pulse
release of water based on their study of Hudson Strait sediment cores. There is also
evidence from sediments in the lower Rhine (Hijma and Cohen, 2010) that there may
have been two emptying phases leading to two step changes in worldwide ocean level,
the first in around 8.47ky BP and the second (and possibly larger) at least 150 years later.
L. Agassiz
L. Ojibway
Great Lakes
Laurentide Ice Sheet
Hudson
Bay
Hudson Strait Labrador
Sea
Atlantic
Ocean
Pacific
Ocean
R. Ottawa
R. Mississippi
R. Red
CHAPTER 1 INTRODUCTION
4
With the current state of knowledge it therefore appears that the sequence of events
that led to the 8.2ky BP event was complex, but that significant outflow from Lake
Agassiz commenced in about 8.47ky BP, sufficient to alter the North Atlantic gyre
circulation (Born and Levermann, 2010) and so affect climate across much of the
northern hemisphere. Indeed, climatologists (e.g. Wiersma and Renssen, 2006) have
modelled this release of fresh water and generally predict climatic effects that match
those recorded reasonably closely. This is entirely logical and there is good reason to
accept that the broad mechanisms of climate change have been correctly identified.
It should be noted that it is not possible to be certain how long the emptying of Lake
Agassiz should have taken. Most assume it to be several years or decades, although
Clarke et al (2009) suggest no more than 6 months based on ice dam modelling. The
well-known (but much smaller) example of the Missoula ice dam and its repeated
formation and collapse (Benito and Connor, 2003) is indicative of the sort of event that is
most likely to have occurred. It is almost impossible for an ice dam to collapse gradually
because of the differential in density between ice and water; once the relative levels of ice
and water are sufficient to begin to float the ice, the laws of physics dictate that outflow
will be very rapid indeed – but dam re-closure is then possible as the lake water level
drops.
Turning to the hypothesis which is the subject of this report, it is proposed that there
were indeed two major phases of emptying of Lake Agassiz, although the entire period
was one of rapid ice sheet disintegration and so relatively high volumes of meltwater
would have flowed into Hudson Bay throughout. The first release of water in around
8.47ky BP was the smaller; the second, some time before 8.2ky BP, revealed by a sharp
and very short-lived spike in the oxygen isotope ratio in the Labrador Sea (Hillaire-
Marcel et al, 2007), was rather larger. It may also have been revealed by ultra-high
resolution analysis of one of the Greenland ice cores (Thomas et al, 2007) which shows a
spike in oxygen isotope ratio lasting no more than 3 years, equivalent to a temperature
reduction of about 3C compared to typical levels for preceding and following decades.
Thomas et al (2007) date this sudden occurrence to 8.19ky BP. The authors suggest that
this 3-year event should be discounted since it has not been observed in other ice cores;
whether this advice is well-founded or not, it is the contention in this report that a hitherto
unrecognised and exceptional phenomenon was precipitated by the second release of
water from Lake Agassiz.
1.3 The Archaeological Gap
Whilst it is reasonable to suggest that the cooling and, in some places, drying of climate
was sufficient to influence human settlement patterns, the degree to which this particular
climate change episode is being called upon to explain archaeological evidence is
surprising, and in many cases the evidence in question is actually absence of evidence.
In North America this approximate timeframe saw a major change in settlement
patterns in many parts of the continent. In the south-east the regional homogeneity that
had been apparent since the arrival of Clovis culture in about 14ky BP suddenly came to
CHAPTER 1 INTRODUCTION
5
an end (e.g. Walthall, 1990). There is a marked reduction in archaeological visibility, i.e.
a much reduced number of sites compared to the preceding Early Archaic phase.
Continuity of human culture is apparent in certain locations, but not in others. The only
explanation offered is climate change, leading to changes in vegetation and food
resources.
This slightly unsatisfactory picture in America is paralleled in Europe where a hiatus
in the archaeological evidence has frequently been found (e.g. Berger and Guilaine, 2009)
between the Mesolithic (late hunter-gatherer) and Neolithic (early farming) periods.
Mesolithic cultures appear to come to an end in many parts of the continent in
approximately 8.2ky BP and there is then a gap before the first Neolithic evidence. This
gap is seen from Spain to Russia. It is not ubiquitous but it is widespread – and once
again climate change, leading to changes in lifestyle, is most commonly held to be
responsible.
That this should be the case on two continents could be coincidence. However, when
a similar Mesolithic-Neolithic gap is seen in the Ganges plain of northern India (e.g.
Varma, 2008) at exactly the same time then it becomes reasonable to suggest that a
planet-wide effect may be to blame – and the only planet-wide effect known at this time
is the cooling and drying that is evident in Greenland ice cores and in many other records.
One could also mention the interruption in archaeological evidence across large parts of
Brazil and Argentina (Araujo et al, 2005) at this approximate date; similarly the sudden
and inexplicable demise of civilisations in South-East Asia and the Middle East. In every
case the only explanation on offer would appear to be climate change.
Whilst it is not the intention to deny that climate change took place, the question that
will be addressed here is whether climate change alone is a sufficiently potent force to
have caused such significant disruption to so many different human cultures and
civilisations. It will be suggested that the principal cause for these changes was in reality
a sudden catastrophe, initiated by the final release of water from Lake Agassiz, a
catastrophe that has left an archaeologically visible mark and led to orally transmitted
folk histories that continue to be told to this day.
1.4 Organisation of this Work
This report comprises a scientific investigation into a particular event. Thus the
arrangement into chapters is intended to follow a logical sequence rather than building
any unnecessary suspense. The phenomenon proposed is related primarily to planetary
science. The currently accepted impact of the emptying of Lake Agassiz is that sea levels
rose by about 0.5m, and that climatic changes were induced by the massive increase in
freshwater input to the North Atlantic. However it will be demonstrated in Chapter 2 that
the sudden loss of water mass from one part of the world, accompanied by isostatic
rebound of the Earth’s mantle, would have altered the balance of the centrifugal forces
induced by the planet’s daily rotation, and a change in centrifugal force balance would
have led to the initation of a phenomenon known as ‘true polar shift’ or ‘true polar
wander’, the possible extent and consequences of which will then be explored
CHAPTER 1 INTRODUCTION
6
mathematically. Reason will be given for believing that the extent of this polar shift may
have been massive, in which case the impact would have been felt principally in a
relative difference between the levels of the solid Earth and the oceans. If true then the
result would have been very extensive flooding in many parts of the world. It is this
flooding that comprises the ‘catastrophe’ referred to previously.
Chapters 3 to 7 will explore evidence from different fields, potentially related to the
catastrophe proposed in Chapter 2. The nature of this catastrophe is one of slow planetary
oscillation about equatorial axes, covering one complete cycle of movement of magnitude
±7and taking several years. Thus the flooding would not, in general, have been violent
but would rather have consisted of a very gradual rise and fall in sea level. Chapter 3 will
search for geological signs of this event, principally in lakes, inland seas and salt playas
that would suddenly have been transformed chemically by seawater ingress. It will also
investigate the effect on ice sheets – since sea level rise should have floated many,
potentially leading to their disintegration.
Chapter 4 will turn to the impact that such an event should have had on the biosphere,
seeking out whether any of the current distribution of flora and fauna may in reality have
been caused by the proposed flooding. Evidence will be presented that areas where
flooding is predicted to have taken place are notably species poor, suggesting that these
areas may still be recovering from a significant disaster. This will be found to apply to
continental regions and also to islands. It will also be suggested that the unexpected
movement of certain species to new habitats may have been facilitated by the proposed
event, for example allowing marine species to cross between bodies of water and, in
some cases, to colonise inland lakes.
Turning to the human world, Chapter 5 will investigate the archaeological evidence
available, concentrating on regions where flooding is predicted. It will be found that
locations where an archaeological gap exists datable to around 8.2ky BP correlate very
well with the proposed flooding. Instances will be found where human culture continued
uninterrupted on high ground but underwent dramatic change on low-lying ground. While
none of these instances will be able to prove the hypothesis of polar shift, no case will be
found that contradicts it.
Continuing with the impact of this event on humankind, Chapter 6 presents an
evaluation of the evidence to be gleaned from historical linguistics, i.e. whether inferred
language spread events support the annihilation of human beings from wide lowland
areas in about 8.2ky BP or not. Several examples will be presented which imply that a
dramatic occurrence must have taken place at that approximate time, sufficient to impel
certain language families to sudden expansion of territory – and in each case it will be
found that the spread is likely to have been from high ground across lower-lying plains.
The final source of evidence will be derived from human history. It is well known
that numerous ancient societies include orally transmitted stories of a ‘great flood’, the
story of Noah being the best known. Chapter 7 will investigate the evidence from 189
flood stories from cultures right across the world, comparing the content of each tale with
the flooding expected in the location where the tale is most likely to have originated. This
will reveal a pattern that is remarkably consistent with the distribution and local depth of
flooding proposed, inviting the conclusion that all 189 flood myths are in reality folk
memories of the same global catastrophe. Evidence will also be presented that pushes the
date of origin of at least two of the stories to around 8000 years ago.
CHAPTER 1 INTRODUCTION
7
The final chapter takes the form of a discussion. It will be argued that there is
overwhelming evidence that something dramatic occurred in around 8.2ky BP, something
with worldwide repercussions. A debate will then be presented, exploring whether it is
possible that the various strands of evidence could potentially be explained solely by
climate change or by other natural and accepted causes. In certain instances it will be
found that this is possible – though unlikely; however attention will be drawn to a hard
core of evidence that would appear to demand something much more catastrophic in
nature.
1.5 Terminology
Although this is a scientific work, it is a work that cuts across several disciplines. The
view has therefore been taken that it would be counter-productive to use terminology that
is so specialised that it cannot be understood by those from other fields. A balance has
therefore been sought such that the text is clear and not subject to alternative meanings
but that it is also expressed using relatively non-specialist language. Acronyms have
generally been avoided except in a few instances where a term is used on a large number
of occasions, in which case it is clearly explained on first usage.
Expression of date is in terms of the BP (before present) system, calibrated in the case
of 14C dates, as used by geologists and palaeobiologists, and the unit used is thousands of
years (or kiloyears – ky), occasionally millions of years (or megayears – My). ‘Present’ is
defined as AD1950, following the current convention. In some cases reference is made to
the AD/BC system, as more commonly used by archaeologists, historical linguists and
historians, but a conversion to BP is generally given.
CHAPTER 2 PHYSICS OF THE EARTH
8
2. Physics of the Earth
The immediate impact of the sudden emptying of Lake Agassiz in about 8.2ky BP was
that a mass of water estimated at 1.63 1015kg (Teller, 2003) or 1.51 1015kg (Clarke et
al, 2004) was lost from one part of the world, Canada, and redistributed across the
oceans. The volume would have been sufficient to cause about 0.5m of sea level rise
worldwide. That this event took place is now very widely acknowledged.
However it is the contention here that researchers have yet to appreciate the possible
consequences of this event with regard to its effect on the motion of the Earth. From one
point of view this is reasonable since the mass involved is small when set against the total
mass of the ice sheets that have come and gone across northern parts of the planet on
several occasions over the course of the last two million years. An argument could be
made that, if the Earth remained stable under these very substantial changes in mass
distribution, then it should logically have remained stable under the much smaller mass
redistribution caused when Lake Agassiz emptied. Nevertheless this argument misses one
crucial point; the suddenness involved. An ice sheet takes thousands of years to form and
thousands more to melt, during which time the various layers within the Earth might be
expected to have time to respond and to counterbalance the added (or subtracted) mass.
The emptying of Lake Agassiz may have taken no more than 6 months (Clarke et al,
2009). It is this point which will be central to the argument presented in this chapter.
2.1 Properties of the Earth
Before describing the calculations that have been carried out to simulate the likely motion
of the Earth following the emptying of Lake Agassiz, it is first necessary to detail the
assumptions made regarding the properties and dimensions of the planet’s various
components, and to acknowledge the unknowns involved.
2.1.1 Dimensions
The following values have been used in calculations:
Average radius of the Earth = 6368km
Difference between radii at equator and poles = 21.3km
Average thickness of the mantle-crust layer = 2883km
The difference between radii at the equator and at the poles is well known and it can be
shown that it has arisen as an inevitable consequence of the centrifugal forces generated
by the spin of the Earth. A similar though reduced centrifugal force effect applies at the
core-mantle boundary, and if core elipticity is considered to be due purely to centrifugal
CHAPTER 2 PHYSICS OF THE EARTH
9
forces at the radius of the core-mantle boundary and the gravitational attraction operating
at the core-mantle boundary is as determined in the next subsection, then the difference
between mantle thicknesses at the equator and the poles is deduced here to be around
15.5km. Other researchers (e.g. Matthews et al, 2002) suggest a greater elipticity for the
core-mantle boundary and therefore a reduced mantle thickness variation, typically
around 12km. Souriau and Souriau (2007) quote a range of hypothesised elipticities
equivalent to mantle thickness variations between 1.5km and 14.5km. Unfortunately,
accurate measurement of core-mantle boundary elipticity does not appear to be possible
since there is in reality a blurred transition zone of thickness 10-40km (Rost and
Revenaugh, 2003) between the liquid iron of the outer core and the solid material of the
lower mantle. Thus mantle thickness variation, which will be found to have a
considerable effect on the predicted response of the planet to the emptying of Lake
Agassiz, has to be acknowledged as an unknown.
2.1.2 Density
The following values have been assumed based on Robertson (1966), typical of those
commonly adopted by researchers:
Density at the surface = 2700kg/m3
Density at the crust-mantle transition (100km depth assumed here) = 3200kg/m3
Density at the base of the upper mantle (670km depth) = 3900kg/m3
Density at the top of the lower mantle (670km depth) = 4500kg/m3
Density at the base of the lower mantle (2883km depth) = 5600kg/m3
Density at the top of the outer core (2883km depth) = 9700kg/m3
It is acknowledged that these are approximate
figures. The crust-mantle transition is in most
places shallower than 100km; the discontinuity at
670km depth is relatively well defined although
quoted depths vary slightly. Linear variation has
been assumed between the values given except in
the case of the upper mantle for which density has
been assumed to increase more rapidly at depths
greater than 300km. These assumptions are
plotted in Figure 2.1.
Having adopted assumptions for density,
gravitational attraction can then be calculated at
each depth. At any given level it can be shown
that material above that level has no influence on
the gravitational attraction felt at that point so
long as the planet can be assumed to be a sphere
and to be compositionally uniform at a given
radius. The attraction is therefore due solely to
material at greater depth. Applying Newton’s
equation for gravitational attraction, the result
shown in Figure 2.2 has been derived. The
0
500
1000
1500
2000
2500
0 2000 4000 6000
Density (kg/m3)
Depth below Surface (km)
Figure 2.1. Assumed mantle
density distribution
CHAPTER 2 PHYSICS OF THE EARTH
10
increased gravitational attraction at the core-
mantle boundary is slightly counter-intuitive since
the mass of material in the core is only some 33%
that of the planet as a whole. However the much
increased density of the core relative to the mantle
leads to a high gravitational attraction.
2.1.3 Stiffness Modulus
The property required here is a value for the
stiffness modulus (or Young’s modulus) of the
mantle-crust layer as a whole. Planetary physicists
generally assume a value of 300-600GPa based on
laboratory testing of synthesized perovskite at
very high temperature and pressure (e.g. Andrault
et al, 2001; Deschamps and Trampert, 2004;
Sinelnikov et al, 1998), perovskite being the
principal component of the lower mantle. It is
unstable at lower pressure – hence the
discontinuity in mantle density at 670km depth,
coinciding with the uppermost perovskite layer.
The principal mineral in the upper mantle is
peridotite. There are inevitable uncertainties that
are still the subject of fierce debate regarding the
actual temperature profile through the mantle and this gives rise to significant uncertainty
in modulus. Nevertheless it would be hard to argue for a value less than about 300GPa
based on laboratory testing, supported by theoretical analysis of perovskite’s molecular
structure (Oganov et al, 2001).
The problem however is that both the testing and the theoretical prediction relate to
continuous material; they take no account of discontinuities that might be present in a
large material mass. And the question has to be asked as to whether an ‘element’ modulus
of this type is appropriate to describe the behaviour of the whole 2883km-thick mantle
layer in flexure. There are known to be very significant weak zones, notably plumes of
molten magma extending throughout its thickness, such as that underlying Hawaii.
Indeed, an unexpectedly low measurement of the effective stiffness modulus of the
mantle-crust layer immediately underlying a lake in Chile (Bevis et al, 2004) was
obtained by measuring the land level change as the lake level rose and fell. The inferred
value, about 18GPa, was put down to the likelihood of magma lenses in a tectonically
active region.
The phenomenon of plate tectonics is considered critical here. The Earth’s crust
comprises several plates of varied size (Figure 2.3) that are in a continuous state of
dynamic flux. Features such as the mid-Atlantic ridge represent locations where magma
is being driven up from the mantle, forcing plates to separate; in subduction zones such as
that off the coast of Chile one plate rides up over another; elsewhere the principal
movement is one of shear as one plate moves laterally relative to another. The issue is to
what extent the pattern seen at the surface is reflected in deep discontinuities within the
mantle itself. The majority of theories current at present (e.g. Okamoto et al, 2005) see
Figure 2.2. Variation in
gravitational attraction
0
500
1000
1500
2000
2500
9 10 11
Gravitational attraction (m/s2)
Depth below Surface (km)
CHAPTER 2 PHYSICS OF THE EARTH
11
magma plumes as initiating right at the base of the mantle, driven by heat from the core,
while Bovolo (2005) refers to seismic evidence that subduction zones also penetrate deep
into the lower mantle. One might therefore suspect that the mantle beneath the mid-
Atlantic ridge contains a near-continuous ‘wall’ of low-stiffness, low-viscosity material
and that its resistance to flexure would be very considerably less than that of intact
perovskite or peridotite. The same would be expected in the regions of raised temperature
within subduction zones.
Figure 2.3. Crustal plate map
Although the information available on the structure of the mantle will probably never
be sufficient to enable a confident estimate of flexural stiffness modulus to be made, it is
instructive to consider what the effect of major discontinuities would be. Thom (2008)
has presented a means of predicting the effective flexural stiffness modulus of a fractured
layer, depending on the element (laboratory-scale) modulus of the material, the spacing
of fractures and the shear stiffness across a fracture, applying it to the well-understood
case of a concrete road pavement. If this approximate equation is applied to the mantle as
a whole, taking 300GPa as the element modulus and assuming fractures at an average
spacing of 5000km (at the surface), infinitely stiff in shear, the predicted effective whole-
mantle modulus in flexure is around 20GPa.
One might reasonably argue that this estimate may be unrealistically low since the
discontinuities formed by rising magma plumes and subduction zones are less complete
than discrete fractures in concrete; on the other hand the distance between major faults is
less than 5000km in many parts of the world, and several lesser faults also exist.
Thus it is quite impossible to argue with confidence based on the nature of the mantle
alone because of the inevitable paucity of data, although it is logical to suggest that the
true effective modulus value in flexure should be significantly less than that of small
intact specimens. Mantle stiffness modulus therefore stands as a further unknown.
2.1.4 Viscosity
Viscosity is the other potentially relevant mantle property, a property that is deducible
from the continuing isostatic rebound of continental areas previously subject to ice sheet
CHAPTER 2 PHYSICS OF THE EARTH
12
loading, notably in North America and Scandinavia. For purposes of this report the
estimates given by Kaufmann and Lambeck (2000) have been adopted. They suggest
ranges of 5-10 1020 Pascal seconds for the upper mantle and 300-800 1020 Pascal
seconds for the lower mantle. If these ranges are correct then it is clear that almost all the
viscous flow taking place is in the peridotite of the upper mantle.
Other properties such as the viscosity of the liquid iron forming the outer core are not
directly relevant to the calculations presented here. In effect it will be assumed that the
outer core flows as required with negligible resistance.
2.1.5 Conclusion
Three key properties of the Earth relevant to the calculations made in this report are not
accurately known. These are the variation in mantle thickness, with limiting estimates of
1.5km and 15.5km, mantle viscosity, with a range of about 5-800 1020 Pascal seconds,
and mantle stiffness modulus in flexure, with estimates varying from 300-600GPa for
intact continuous material down to 20GPa when considered as a discontinuous layer.
2.2 Isostatic Response to Emptying of Lake Agassiz
Since the emptying of Lake Agassiz is being suggested here as the catalyst for a much
larger planetary catstrophe, it is clearly essential that the immediate isostatic response is
taken into account properly. In general two types of response are activated whenever
there is any redistribution of mass across the surface of the planet. The first is elastic and
occurs practically instantaneously; the second is viscous and the effect may continue for
thousands of years. This is true of the demise of the great ice sheets of the past; it would
have been equally true in the aftermath of the disappearance of Lake Agassiz.
2.2.1 Elastic Response
The mantle of the Earth is, in mechanical terms, a thick flexible layer, and it is supported
on a dense liquid foundation, namely the liquid iron outer core. Thus, when the load on
the mantle reduced by an estimated 1.6 1018 Newtons following the loss of Lake
Agassiz the entire layer would have risen locally due to its ability to extend and bend
elastically. The calculation is similar to that in several situations in engineering practice,
once adjustments are made to take account of the curvature of the Earth and the changes
in gravitational attraction with depth, although the result is clearly dependent on the
assumption made for the stiffness modulus of the mantle. For example, Figure 2.4
presents approximate magnitudes of immediate elastic level change assuming either a
300GPa or a 150GPa mantle stiffness modulus, calculated using multi-layer linear elastic
analysis.
2.2.2 Viscous Response – Mantle
Viscous flow within the mantle has an effect equivalent to reducing the long-term
stiffness modulus. For example, if a weighted average viscosity of 1021 Pascal seconds is
assumed to apply, and if the elastic modulus of the mantle is taken to be 300GPa, then the
CHAPTER 2 PHYSICS OF THE EARTH
13
rebound shape after about 100 years would be equivalent to that calculated using a
modulus of only 150GPa. The rate of level increase would reduce as the long-term
equilibrium condition was approached. This is the type of viscous response that took
place after the retreat of the ice sheets and it continues today in some parts of the world.
However it is too slow an effect to have had any appreciable influence on the short-term
response of the planet to the emptying of Lake Agassiz.
Figure 2.4. Elastic rebound due to emptying of Lake Agassiz
2.2.3 Viscous Response – Core-Mantle Boundary
It has been known for many decades that the transition from the solid perovskite of the
lower mantle to the liquid iron of the outer core is not a sharp one. Sound theoretical
support for the lowermost 250km of the mantle, known as the D” layer, having a different
structure has emerged (Murakami et al, 2004) and the material has become known as
‘post-perovskite’. Furthermore, seismic observations (e.g. Rost and Revenaugh, 2003)
have identified what is termed an ‘ultra-low velocity zone’ (ULVZ) immediately adjacent
to the core-mantle boundary. Via (2008) has carried out a theoretical simulation of
seismic studies and proposes that the viscosity actually decreases dramatically through
the D” layer, being no more than about 1014 Pascal seconds 3km from the core-mantle
boundary, suggesting that the reduction should theoretically continue through the
remaining 3km due to partial melt.
These observations and the deductions that stem from them imply that material in the
ULVZ would flow to fill the slight dome at the core-mantle boundary created by the
immediate elastic response illustrated in Figure 2.4. If for example the viscosity of a
3km-thick layer is assumed to be 1014 Pascal seconds and the density difference between
the ULVZ material and the core is taken to be 4000kg/m3, then the dome shown in Figure
2.4 for a 300GPa mantle modulus is predicted to half-fill with ULVZ material in about 6
years, the rate decreasing logarithmically. This is therefore a sufficiently rapid effect to
influence immediate planetary behaviour. It suggests that any planetary imbalance
created by the draining of Lake Agassiz would probably have dissipated substantially
within a few years – although the actual rapidity cannot be known with confidence.
0
2
4
6
8
10
12
14
16
18
-6000 -4000 -2000 0 2000 4000 6000
Distance from Lake Agassiz (km)
Elastic Rebound (m)
Surface (300GPa Mantle)
Core-Mantle Boundary (300GPa Mantle)
Surface (150GPa Mantle)
Core-Mantle Boundary (150GPa Mantle)
CHAPTER 2 PHYSICS OF THE EARTH
14
2.2.4 Centrifugal Force Imbalance
The nature of the imbalance that would have occurred relates to the centrifugal force (due
to diurnal rotation) distribution around the planet. Without the isostatic rebound, the loss
of water mass from Lake Agassiz would have represented a large loss of centrifugal force
attributable to Canada; however once the immediate elastic rebound is accounted for,
raising the level of the relatively dense mantle and the even denser liquid iron outer core,
the loss of centrifugal force is at least partially cancelled out – and the degree to which
this would have occurred depends on the value assumed for the modulus of the mantle.
However, whatever the magnitude, it is certain that there would have been an overall
change in the centrifugal force attributable to the North America region.
In response to this change in centrifugal force balance the Earth would inevitably
have started to tilt. The centrifugal force imbalance would have created an overturning
‘moment’; this would have led to an angular acceleration of the planet about an equatorial
diameter with magnitude dependent on the moment of inertia of the planet. For example,
with a mantle modulus of 300GPa, the overturning moment is calculated to be
approximately -7.5 1021 Newton metres, the negative sign representing a tendency for
Canada to be pulled north. The moment of inertia of the Earth, excluding the core since it
is debatable whether much of it would have been affected by this overturning motion, is
calculated to be 7.06 1037 kgm2, which means that the initial angular acceleration of the
Earth would have been about 10-16radian.s-2.
These statements and calculations are
fundamentally non-controversial. The type
of rotation induced is termed ‘true polar
shift’. It should be understood that this
involves no significant change to the axis of
diurnal rotation, nor to the rate of that
rotation. In effect it superimposes a
secondary rotation of the solid parts of the
planet about an axis at right angles to the
polar axis. Figure 2.5 illustrates the concept.
Thus, while the positions of the north and
south poles relative to the rest of the solar
system remain unchanged, the actual surface
of the planet would shift relative to those
poles. If the motion continued for long
enough then it would theoretically be
possible for Canada to move sufficiently to
cover the North Pole while Siberia drifted
steadily south.
2.2.5 Conclusion
It is absolutely certain that the emptying of Lake Agassiz, followed by isostatic rebound,
would have initiated polar shift of the planet due to an induced centrifugal force
imbalance.
North Pole
South Pole
Point which
used to lie
on the North
Pole
True Polar Shift
Point which
used to lie
on the
South Pole
Equator
Line which
used to be
the equator
Figure 2.5. The principle of true
polar shift
CHAPTER 2 PHYSICS OF THE EARTH
15
2.3 Planetary Response to Polar Shift
The inate tendency of the Earth – or any planet – to develop oscillatory polar shift was
first recognised by Leonard Euler over 200 years ago. He calculated that, in the event that
any impulse led to initiation of polar shift motion of the Earth, an oscillation with a
period of 305 days would develop. This took full account of the 21.3km bulge at the
equator compared to the poles as well as the so-called ‘Coriolis effect’, in which a
spinning object rotated about a secondary axis induces a further rotation about a third
orthogonal axis, which in turn induces a moment resisting the motion about the secondary
axis. However he did not take into account the fact that the Earth is of finite stiffness, nor
that it comprises a solid mantle overlying a liquid outer core. When an actual polar shift
oscillation of the planet was discovered by Seth Chandler in 1891 (known as the
Chandler Wobble), having a variable amplitude of a few metres at the surface and a
principal period of about 433 days, it was deduced that the discrepancy between this
observation and Euler’s prediction was due to the Earth’s finite stiffness and liquid outer
core.
Thus the concept of polar shift is non-controversial, as is the statement that polar shift
motion of an unusually large amplitude would have been induced when Lake Agassiz
drained. However the suggestion that will be made in the following subsections regarding
the possible extent of that shift will be highly contentious, and a key factor affecting
prediction will be the flexibility of the system.
2.3.1 Changes in Mantle Level
It was stated above that the 21.3km equatorial bulge has formed as a consequence of the
centrifugal force distribution throughout the body of the Earth. The effect of polar shift is
to change the effective latitude of each point on the surface of the planet relative to the
unchanged axis of spin, with consequent changes to the centrifugal forces applying at
each point and therefore to equilibrium distances from the centre of the Earth. For
example, if a point on the equator on the longitude of Lake Agassiz were to have moved
north through 1 degree, it would have found itself about 6.4m too far from the centre of
the Earth for that latitude; it would no longer have been in equilibrium. In response, the
mantle in that region would have sunk back down into the liquid iron outer core. In fact,
while the mantle in two quadrants of the Earth would have tended to sink, in the other
two it would have tended to rise by an equal amount – Figure 2.6 illustrates – although
the desired sinking and rising of different parts of the planet’s mantle relative to the core
would have been resisted by the need for the mantle to flex to accommodate this effect.
Ignoring for the moment the constraint due to the stiffness of the mantle in resisting
flexure, it is possible to calculate the amount by which the surface would have either
risen or sunk if it were unconstrained. If the variation in mantle thickness with latitude
were also to be discounted, then the surface would simply have attempted to reach the
level appropriate to its new latitude. In the case of the point originally on the equator
referred to earlier, it would therefore have sunk by about 6.4m if it had moved 1 degree
north. Every point on the longitude of Lake Agassiz north of the equator would have
sunk; points south of the equator would have risen.
CHAPTER 2 PHYSICS OF THE EARTH
16
However the variation in mantle
thickness must also be taken into account.
For example, adopting a 12km mantle
thickness difference between the poles and
the equator as recommended by Matthews et
al (2002), the thickness of the mantle on the
equator is theoretically about 3.7m greater
than the equilibrium thickness for 1 degree
north latitude, which means that the land
surface would have sunk rather less than
6.4m. It is a similar calculation to that for
isostatic rebound in that the variation in
gravity through the mantle has to be
considered as well as the curvature of the
Earth and the consequent reduction in
effective area at depth. The result is that the
surface is calculated to have sunk by around
6.0m.
The importance of this detail is seen
when a comparison is made between the
behaviour of the solid Earth and that of the
oceans. Since the axis of diurnal spin would have remained unchanged, the oceans would
have remained unchanged in their levels relative to that axis. Thus at a point 1 degree
north of the equator the surface of the ocean would have remained about 6.4m nearer to
the centre of the Earth than on the equator. If, on the other hand, the land surface was just
6.0m nearer to the centre of the Earth than when it was on the equator, the effect would
be that the land would have risen 0.4m relative to the ocean. Furthermore the equator is
the least affected part of the planet because of the sinusoidal nature of mantle thickness
variation; a point on latitude 10North would have experienced about twenty times as
much radial displacement and an 8m uplift of the land relative to the ocean. Conversely, a
point on latitude 10South would have experienced an 8m sea level rise. These
magnitudes are also dependent on the assumed variation in mantle thickness; for example
if the thickness difference between the equator and the poles had been taken as 15km
then the 8m sea level rise would become 10m.
Nevertheless, these calculations still take no account of the resistance provided by the
mantle’s flexural stiffness. In effect, the mantle is being required to bend, most sharply in
middle latitudes, something that would clearly be resisted to a degree that depends on the
value assumed for its stiffness modulus.
2.3.2 Resistance due to Mantle Stiffness
To model the effect of mantle bending, calculations have been carried out (see Appendix
A) assuming that each segment of the mantle behaves as a beam on a dense liquid
foundation. A relatively crude adjustment has then been made to take account of the
secondary effect of latitudinal bending. This approach is acknowledged to be imprecise;
however it is justified on the grounds that the uncertainty in choice of stiffness modulus
is, by comparison, a much more significant factor affecting the result.
0
30
30
60
60
90
0
30
30
60
60
North Pole
South Pole
Mantle rising andsinking during tilt
Areas
which sink
most
Areas
most
Equator
Figure 2.6. Rising and sinking of
the mantle during polar shift
CHAPTER 2 PHYSICS OF THE EARTH
17
Bending calculations as described have then been inserted into a prediction of polar
shift (see Appendix A). After each increment of motion the mantle shape is re-calculated.
The sum of all the moments due to centrifugal force is then computed and added to (or
subtracted from) that due to the emptying of Lake Agassiz. A similar calculation is
carried out for mantle bending and moment in the orthogonal direction as a secondary
rotation is induced by the Coriolis effect. The calculated moments then induce angular
acceleration or deceleration of the planet and this in turn affects the rate and extent of
polar shift during the next increment.
Thus it is possible to predict the progress of polar shift motion as the resistance of the
planet changes due to the combination of changes in mantle thickness (at a given latitude
relative to the axis of diurnal rotation) and mantle flexure. Unfortunately, as noted above,
both mantle thickness variation and mantle flexural modulus are imprecisely known.
2.3.3 Deduction based on the Chandler Wobble
The Chandler Wobble is an ongoing irregular oscillatory polar shift of the Earth with an
amplitude of a few metres at most and an approximate return period of 433 days. There
has been much debate in recent years (e.g. Wilson and Chen, 2005) as to the driving force
behind the Chandler Wobble, the most confident proposal being that put forward by the
US Jet Propulsion Laboratory (Gross, 2005) that it is due to seasonal changes in deep
ocean pressure. It is known that the cause is related to an annual effect because the
motion of the Chandler Wobble actually has two dominant frequencies, the most
significant being the 433-day period mentioned above, but the secondary frequency
having a period of roughly 365 days.
The next logical step therefore is to apply the calculation approach described in the
previous subsection to simulation of the Chandler Wobble. The result (see Appendix A)
is that, with a mantle thickness variation of 12km, and applying a sufficiently large input
moment to induce an oscillatory movement with a maximum amplitude of about ±6m, it
is necessary to assume a mantle modulus of about 360GPa in order to match the observed
motion and the 433-day period. With a different assumed mantle thickness variation, e.g.
10km, then the mantle modulus would have to decrease to about 330GPa. The required
input moment in either case (applied as a sinusoidal input of period 365 days) is about
1% of that calculated to have been delivered when Lake Agassiz drained.
This simulation would therefore appear to vindicate the choice of a relatively high
mantle stiffness modulus – 360GPa is nearly twice the modulus of steel under Earth
surface conditions. Indeed, in the calculations that follow, a mantle thickness variation of
12km and a modulus of 360GPa will be adopted as reasonable estimates.
2.3.4 Non-linearity Effects
The deduction of a flexural stiffness modulus of 360GPa based on simulation of the
Chandler Wobble represents a validated starting point. Adopting this value and applying
the impulse generated by the disappearance of Lake Agassiz, it is predicted (see
Appendix A) that an oscillation of magnitude ±0.001 degrees (approximately ±100m on
the ground) would have been induced, insufficient to generate more than a very slight
perturbation in ocean levels.
However the difference in magnitude between the moment required to generate the
Chandler Wobble and that due to the draining of Lake Agassiz is a factor of about 100,
CHAPTER 2 PHYSICS OF THE EARTH
18
which implies that stresses and strains within the mantle are also likely to have been
much larger in the latter case. And while one might expect intact perovskite to maintain
approximately the same elastic stiffness modulus over a wide range of stress and strain
conditions, this is not the case for a discontinuous layer. A discontinuous material
introduces the possibility of slippage between intact blocks of solid matter, controlled by
interlock between irregular faces and by frictional resistance, and the nature of such a
system is that the initial resistance that has to be overcome in order for slip to commence
is greater than the value applying once it is in progress. Thus the resistance of a
discontinuous system to deformation is generally much greater at small strain than at
large strain, an effect that is well recognised in the case of granular construction materials
and soils. For example Von Quintus et al (1994) quote data for both soils and gravels
showing factors as high as 10 between small-strain moduli and those at strains 100 times
as great. Clearly it is not possible to know the extent of so-called ‘strain softening’ that is
likely to apply in the case of the Earth’s mantle but, at strains considerably greater than
those caused by the Chandler Wobble, it is logical to expect the effective modulus of the
mantle to decrease significantly from a low-strain value of 360GPa.
Many different models have been put forward to describe the quasi-elastic behaviour
of frictional materials, some highly complex (e.g. Thom, 1988). However, bearing in
mind the uncertainty present in this case it is considered that a simple formulation is
justified, as follows:
Modulus = 360
(k
tilt)-n GPa
where kis constrained to give a modulus of 360GPa at a tilt of 0.00006(the approximate
maximum amplitude of the Chandler Wobble) or less and nis a variable describing the
extent of strain softening present.
There is also a secondary effect applying to discontinuous materials at large strains,
namely ‘strain hardening’, an increase in modulus as the material begins to ‘lock up’,
which would in practice limit the stiffness reduction at large values of strain. Clearly the
magnitude of strain hardening is no more possible to assign with confidence than is the n
value for strain softening; thus it will be included here in a somewhat arbitrary manner as
a multiplier on stiffness modulus linearly dependent on the magnitude of polar shift.
Both strain softening and strain hardening represent logical, reasonable and necessary
adjustments to the rather simplistic assumption of linear elasticity; unfortunately neither
is directly quantifiable and so both will require estimation. However it is worth noting
that if strain softening were ever to reduce the effective mantle modulus to as low as
50GPa, flexure of the mantle would then be sufficient to reduce the resisting moment due
to centrifugal force to zero. With any modulus lower than this the overturning moment
due to centrifugal force would actually increase as the angle of tilt increased.
Nevertheless, no matter how small the modulus becomes due to the effects of strain
softening, the predicted oscillatory motion is still not dramatic, being simply a little larger
and slower than would be the case with a uniform 360GPa modulus, and this is due to the
Coriolis effect which generates a restoring moment in the primary tilt direction due to
angular rotation in the secondary direction (see Appendix A). For example, if the strain-
softening parameter nis chosen to be 0.2 the predicted amplitude of oscillation increases
from ±0.001° to ±0.002°, with a period of just over 2 years; at n= 0.4 this increases to
CHAPTER 2 PHYSICS OF THE EARTH
19
±0.014° – still quite insufficient to cause any planetary disaster despite the computed
mantle modulus reducing to about 38GPa at maximum tilt.
2.3.5 Mantle Inhomogeneity
Thus it is a mathematical fact that, with an Earth model that displays the same properties
on all lines of longitude, predicted oscillation will always be relatively small; in fact
±0.014° is the approximate maximum that could be predicted using the model described
in Appendix A. However, a key element in the hypothesis being put forward in this report
is that the effective modulus of the mantle in bending is actually different on different
lines of longitude. This is a function of the distribution of tectonic plate boundaries –
shown in Figure 2.3. Near to the longitudes of primary tilt, approximately 90° East and
West, there are significant numbers of faults. These lie through Central America, the
eastern Pacific, northern India and the eastern Indian Ocean. In contrast on the longitudes
of secondary tilt, induced through Coriolis effects, there are many fewer faults, the
Pacific and African plates comprising large areas of relatively fault-free mantle.
Self-evidently it is not possible to translate these observations into numbers with any
certainty. Nevertheless, the logical consequence is that the larger number of faults on the
longitudes of primary tilt should be reflected in a greater strain-softening effect, i.e. a
larger nvalue. For example, if nwere to be 0.4 in the primary tilt direction but only 0.2 in
the secondary direction then the prediction would be for a tilt of ±5.8°, over 400 times the
prediction if nwas 0.4 in all directions. And when strain hardening is also applied then
the effect of a discrepancy in nbetween the two tilt directions becomes even more
pronounced. The key finding here is therefore that such a discrepancy, combined with a
significant strain softening brought about by discontinuities within the mantle, has the
potential to induce a catastrophic tilt in response to an impulse of the magnitude
generated when Lake Agassiz disappeared.
2.3.6 Suggested Planetary Motion
The proposal that will be put forward in this subsection is not subject to any direct check
other than that it meets appropriate criteria of reasonableness. It is acknowledged that the
principal sources of evidence will be the indirect signals left in the geology, biology and
human history of the Earth, and these are the subjects of Chapters 3 to 7 of this report. It
is this evidence that has led to the view that the actual magnitude of tilt in the primary
direction, 90° East and West, was around ±7° with a rather smaller tilt in the secondary
direction. This is a movement that could potentially be achieved with various
combinations of strain-softening and strain-hardening parameters. The motion shown in
Figure 2.7 was computed assuming nvalues of 0.34 and 0.17 to describe strain softening
in the primary and secondary tilt directions respectively, and corresponding strain-
hardening multipliers of [1 + 56 × (tilt angle in radians)] and [1 + 28 × (tilt angle in
radians)]. The effect of these non-linearity parameters is that the effective modulus of the
mantle in the primary tilt direction decreases from a low-strain value of 360GPa to a
minimum of 25GPa at about 0.5of tilt due to strain softening (having reached a strain
10,000 times as great as that induced by the Chandler Wobble) and increases back to
53GPa at maximum tilt (7). In the secondary direction the minimum modulus value is
96GPa.
CHAPTER 2 PHYSICS OF THE EARTH
20
Figure 2.7. Proposed polar shift motion
Based on the known distribution of tectonic plates it would be hard to argue that the
proposed difference between mantle properties in the two tilt directions is unreasonable.
A greater level of doubt should probably be reserved for the absolute magnitude of strain
softening proposed. Nevertheless, the point has been made that, were the mantle to be
considered as a fractured layer with major faults avery 5000km, then an effective layer
modulus of 25GPa meets expectations – and this takes no account of lesser faults and
associated zones of raised temperature and therefore reduced stiffness. It is therefore
considered that this proposal passes the broad test of reasonableness concerning mantle
flexural modulus – although it is recognised that this does not mean that it is necessarily
correct.
At this stage it is also worth noting the timescale involved. It would have taken about
8 years before the motion of the planet became significant and 13.5 years before the first
peak was reached. The prediction is then for a return to near-normal conditions before a
second phase of tilting motion, in the opposite direction, reaching its peak about 27 years
after Lake Agassiz had emptied. The reason for this second phase is discussed in the next
subsection.
2.3.7 Recovery due to Viscous Flow
In a system in which there was no energy loss and no change in the magnitude of the
force that initiated the motion, the result would be a continuing oscillation between the
point of maximum tilt and the start point. Because of the complex combination of factors
that contribute towards the planet’s resistance to polar shift, the motion would not be
sinusoidal, but it would nevertheless repeat. However neither of these two assumptions is
valid. Energy loss would certainly occur – discussed in the next subsection – and the
magnitude of the driving force would also change.
The force to which the planet was subject when Lake Agassiz drained was a function
of the immediate elastic rebound response, leading to an imbalance in centrifugal forces.
-8
-6
-4
-2
0
2
4
6
8
0 5 10 15 20 25 30
Years following emptying of Lake Agassiz
Tilt (degrees)
Primary Tilt
Secondary Tilt
CHAPTER 2 PHYSICS OF THE EARTH
21
However, this initiating force would rapidly have started to dissipate due to viscous
effects, as discussed earlier. Material in the ultra-low velocity zone (ULVZ) at the base of
the mantle would have flowed to fill the domed shape created at the core-mantle
boundary due to rebound and, based on the example calculation presented in Section
2.2.3, a half-life of 6 years has been assumed to apply to the centrifugal force imbalance
as a consequence. This is acknowledged to be no better than an informed estimate but it is
not believed that anything substantially better than this is possible. It is clear that the
planet has been able to maintain stability under the less sudden changes that took place
during the deglaciation phase of the terminal Pleistocene, when the ice sheets of North
America lost around 1000km3of ice per year on average, representing a mass equivalent
to the water lost from Lake Agassiz every 150 years. Yet this caused no significant
perturbation of planetary motion, which implies that 150 years gave sufficient time for
the imbalance to be corrected by viscous flow effects. Bearing this in mind, together with
inferred ULVZ viscosity values (Via, 2008), the proposed 6-year half-life is considered
reasonable. It is also of note that the prediction is relatively insensitive to this parameter.
A reducing centrifugal force imbalance and therefore a reducing excitation moment
means that the work done (moment × tilt angle) during the initial tilt would have been
greater than the energy recovered during the return to a zero tilt angle. Neglecting energy
dissipation at this stage, any non-recovered energy would have to have been stored in the
form of kinetic energy, which means the planet would still have been rotating as it
regained zero tilt; hence the second phase of motion shown in Figure 2.7.
2.3.8 Energy Dissipation
Kinetic energy dissipation (in the form of heat) would have occurred for several reasons.
Strain energy loss in the mantle is implied by the assumption of strain softening since this
is a function of frictional effects within the mantle at discontinuities. To this frictional
energy loss must be added effects due to flow within liquid phases (liquid iron in the
outer core, magma within the mantle, oceans at the surface). Losses would also have
occurred due to viscous flow within the ULVZ. Clearly the current state of knowledge of
mantle and ULVZ properties does not allow anything other than discussion and
hypothesis. However, the fact of energy dissipation means that the oscillatory movement
shown in Figure 2.7 could not have continued indefinitely.
As implied above, with the proposed strain-dependency laws the planet is in a
‘metastable’ state with regard to polar shift. Once the angle of tilt exceeds a certain value
– less than one tenth of a degree – the moment due to centrifugal force distribution ceases
to resist the tilting motion and actually begins to contribute towards it. Thus, so long as
there were sufficient momentum available as the oscillation passed through zero to reach
0.1negative tilt, a second major polar shift episode would have occurred – as shown in
Figure 2.7 – and, as argued in the previous subsection, this momentum would almost
certainly have existed due to the steady reduction in the overturning moment attributable
to Lake Agassiz. However, further reduction of this overturning moment during the
second tilt episode would then have acted as a brake on the motion which, when
combined with further energy losses, would logically have meant that there was
insufficient momentum to cause a third major tilt. This is conjectural – but it is also
reasonable.
CHAPTER 2 PHYSICS OF THE EARTH
22
2.3.9 Conclusion
The discussion in this section has of necessity been speculative. While reference to the
Chandler Wobble suggests a small-strain mantle modulus of about 360GPa, and good
reason has been given to believe that the effective flexural modulus of the Earth’s mantle
is likely to be non-linear, i.e. strain-dependent, the proposal illustrated in Figure 2.7 has
no intrinsic weight behind it; it is simply one of several reasonable scenarios that could
have been put forward. A minor change, for example in the degree of difference between
mantle properties in the primary and secondary tilt directions, would dramatically change
the prediction, and it is therefore quite reasonable to believe that there never was any
catastrophic polar shift episode of the sort depicted in Figure 2.7. Nevertheless, based on
the totality of the evidence presented in Chapters 3 to 7, Figure 2.7 represents the
hypothesis put forward in this report. It is that the planet tilted once in each direction by
about 7° over a period of about 30 years following the sudden emptying of Lake Agassiz.
2.4 Implications
It is the contention here that the proposal set out in the previous section is in accord with
the demands of both logic and reasonable calculation. This section will now set out the
effects that such planetary motion would have had on the Earth and its inhabitants.
2.4.1 Predicted Flooding
The proposed motion applies to the solid Earth. The predicted changes in surface level do
not apply to the liquid water on the planet’s surface. The result would therefore be a
change in relative level between the land and the ocean. Because the Earth is predicted to
have tilted once in each direction, each quadrant of the planet would therefore have
experienced flooding at some stage. It would have been deepest in middle latitudes and
on the longitudes of primary tilt, approximately 90West (the longitude of Lake Agassiz)
and 90East, reducing to a minimum on those lines of longitude exactly 90distant, 0
and 180. Figure 2.8 gives values of predicted peak sea level rise, showing an absolute
maximum of around +440m in middle latitudes.
During the first phase of motion, the majority of which would have occurred over a
period of about 8 years, the flooded quadrants would have been those including Europe,
Asia and northern Africa in the northern hemisphere and most of South America in the
southern hemisphere. These floods would have accompanied movement toward the line
of the equator. Meanwhile the other two quadrants of the Earth, including North and
Central America, Australia and southern Africa, would have moved toward the poles and
the land would have risen relative to the ocean. If 7genuinely represents the magnitude
of polar shift then the coast of Ellesmere Island in the Canadian Arctic would have come
to cover the North Pole.
In the second phase, which would have had similar duration, it would have been
North America, Australia and southern Africa that saw flooding while Europe, Asia,
northern Africa and South America rose relative to the sea. If the deductions made above
are correct then the sea level rise would have been similar to that of the first phase. Figure
CHAPTER 2 PHYSICS OF THE EARTH
23
2.9 is a map showing the approximate maximum extent of predicted flooding worldwide
– regardless of when that maximum was reached.
Figure 2.8. Proposed maximum sea level rise
Figure 2.9. Approximate maximum extent of predicted flooding
Regarding rate of increase (or decrease) of ocean level, middle latitudes within about
45of the longitude of primary tilt would have experienced rates of no more than about
0.5m per day; other locations would have seen even lower rates of water level rise. Thus
there would have been sufficient opportunity for escape, by humans or animals, so long
as that escape was to ground of sufficient maximum height. However, if an inappropriate
choice of high ground was made – difficult to avoid across undulating lowland regions –
escape may then have become impossible without water transport.
0
50
100
150
200
250
300
350
400
450
500
0 10 20 30 40 50 60 70 80 90
Latitude (degrees)
Maximum Sea Level Rise (m)
90 degrees longitude
60 degrees longitude
30 degrees longitude
0 degrees longitude
Equator Pole
CHAPTER 2 PHYSICS OF THE EARTH
24
2.4.2 Local Corrections
The calculated distortion of the planet presented thus far has taken account of the
redistribution of the waters of the oceans and the consequent effect on gravitational
pressures, but only as an average effect spread evenly over the surface of the planet; this
is self-evidently inaccurate. Extensive continental regions would clearly have
experienced reduced effect from water redistribution, while in regions of continuous
ocean the full effect would have been felt. The primary consequence of this would be that
sea level rise would have been up to about 10% lower than that presented in Figure 2.8
across wide areas of ocean, but slightly higher than in Figure 2.8 around some of the
continents.
A secondary consequence would be to offset the line of zero flood from the equator,
pushing it toward the hemisphere with less continental land mass. Considering the
western hemisphere, the difference between regions north and south of the equator is
relatively slight, although the line of zero flood would have been displaced slightly, south
of the equator in the eastern Pacific and north of the equator in the western Atlantic, but
by one or two degrees at most. However the difference is much more pronounced in the
eastern hemisphere. Using a weighting related to the relative extent of sea level change in
different areas, the northern half of the eastern hemisphere is approximately 80%
continental, while the southern half is nearly 90% oceanic. The effect of this would be to
displace the line of zero flood to the south by several degrees of latitude and the impact
would have been most significant for the islands of the Indian Ocean. This point will
become relevant in Chapter 4 with regard to species survival on individual islands.
Other relatively minor effects would have been a southward displacement in the
Atlantic off the West African coast and a slight northward displacement in the Pacific
north of New Guinea.
It is also the case that the sea level rise shown in Figure 2.8 is based on the mantle
having uniform stiffness on any given line of longitude; yet the point has been made that
the principal reason for the relatively low effective stiffness modulus applying at high
strain is the faulted nature of the mantle layer. The actual distorted shape of the mantle
surface would not therefore have been smooth but rather quasi-polygonal, with
significant angular distortion at boundaries between tectonic plates and reduced bending
of the plates themselves. Whilst the irregularities of the tectonic plates and a total lack of
detailed knowledge of other faults within the mantle make confident computation of the
magnitude of this effect impossible, it is considered likely that locations close to major
faults may have experienced several metres less sea level rise than would otherwise have
been the case.
2.4.3 Water Flows
In general this event would not have seen significant induced oceanic currents. A rise (or
fall) of up to 0.5m per day is less than the expected tidal range in most parts of the world,
and the effects across the majority of the world’s oceans in terms of current flow rate
would have been negligible. However certain exceptions should be noted.
One region of interest is the Arctic Ocean. There are only two access routes, from the
North Atlantic via the Norwegian and Greenland Seas and from the Pacific via the Bering
Strait. During both tilt phases the Arctic would have had to draw large quantities of water
from the other oceans because of the topography of the basin and surrounding flood
CHAPTER 2 PHYSICS OF THE EARTH
25
plains. In the first phase the Bering Strait would have been flooded during inflow to the
Arctic but practically dry during outflow; during the second phase it would have been
shallow during inflow but much enlarged during outflow (Appendix A, Figure A8). None
of this would have resulted in dramatic currents – although it is reasonable to infer
relatively strong flows through the Bering Strait during certain phases. However,
substantial volumes of water would have been transferred between the Arctic and both
the Pacific and the Atlantic Oceans, in which case it is likely that water may have passed
quite rapidly from the Pacific to the Atlantic or vice versa, something that would happen
only very gradually during normal circumstances. It is also logical to expect the Arctic
Ocean to have warmed significantly during this event.
A much more dramatically affected region however would have been the
Mediterranean-Black Sea-Caspian Sea system. The Strait of Gibraltar is situated nearly
90from the longitude of Lake Agassiz and would have remained open throughout; it is
also of sufficient width and depth to accommodate the required flow volumes. Also,
during the first phase of motion, as the Mediterranean rose relative to the land, water
would have begun to flow across the Suez Canal zone from the Red Sea. Considerable
water volumes would have moved from the Red Sea into the Mediterranean and back,
although flow velocities would not have been high.
However, the route to the Black Sea is much more constricted. If these predictions are
correct then one inescapable conclusion is that very large flows must have occurred
through the Dardanelles and Bosphorus Straits, both as the water rose relative to the land
and then as it fell. A similar violent flow would have to have occurred between the Black
Sea and the Caspian, through the ancient channel known as the Manych spillway.
Furthermore, as the floods rose across Siberia from the Arctic, they would have set up a
second point of access to the Aral-Caspian basin, namely through the Turgay valley,
colloquially known as the Turgay spillway. Here too a violent flow is predicted to have
occurred in each direction. If real, these are significant events that one might expect to be
geologically recognisable.
Naturally there would have been other locally high flows through constricted valleys.
The Iron Gates, where the Danube flows out from the Hungarian plain, would be one
such location; the Yangtze gorges would be another. Nevertheless, across the vast
majority of the planet this would have been a non-violent event, bringing gradual changes
and unexceptional water flow rates. The rate of level increase would however have been
much too high to permit any recognisable wave-induced erosion to have occurred in the
vast majority of locations.
2.4.4 Changes to the Land
Two points should be noted here. The first is that the mantle-crust layer would have
undergone significant straining during these episodes of polar shift, straining that would
have been concentrated at faults and discontinuities within the layer. The effect of this is
impossible to prove but it would be reasonable to expect earthquakes and volcanism as a
result. Earthquakes would be a logical accompaniment to polar shift, possibly continuing
for several centuries thereafter as stress relief within the mantle continued to occur;
volcanism might be expected as a consequence of forced flow within magma lenses.
A more immediate consequence of the predicted straining would have been temporary
changes to the slope of the land relative to the ocean and, equally importantly, relative to
CHAPTER 2 PHYSICS OF THE EARTH
26
the Earth’s gravitational field. The maximum magnitude of these slope changes is
predicted to have been approximately 1.5 10-4, whereas many of the world’s river
systems tend to have falls in the region of 1 10-4. There would therefore have been the
potential for river flows to be reversed in some locations or for rivers to take alternative
courses across the land. In many parts of the world this is hardly relevant since the river
basins themselves would have been substantially flooded. However it is believed to be
particularly relevant in Africa, where little flooding is predicted and where relatively flat
land occurs across the watersheds between several major river systems.
In other respects however, it is not expected that any significant changes would have
occurred. As was the case for the oceans, this was an almost entirely non-violent event
and thus, in most non-flooded regions, there would have been no observable effect on the
land. In such locations, the only matters that a human observer might have noted would
be changes in the elevation of the sun due to temporary changes in effective latitude, and
associated disruption to weather patterns.
2.4.5 Possible Similar Episodes in Earlier Epochs
The deductions made so far refer specifically to the known sudden draining of Lake
Agassiz in the second half of the 9th Millennium BP. No other event of similar magnitude
is known from the most recent phase of glaciation and, if the assumptions made in
generating Figure 2.7 were correct, then anything less than 97% of the Lake Agassiz
outflow would have done no more than induce a fraction of a degree of tilt. In North
America several earlier sudden outflow events have been deduced, from Lake Agassiz
and its partner lake to the east, Lake Ojibway; however they were all of much smaller
magnitude. The glacial lakes of northern Siberia and southern Scandinavia during the
terminal phases of glaciation are less well mapped, but it is not thought that any similarly
large sudden outflow event occurred.
It would however be naïve to expect that none of the previous glacial periods known
to have occurred during the Pleistocene Era was accompanied by any such sudden
outflow. It is thought, for example, that a massive glacial lake covered most of western
Siberia during the Weichselian glaciation in about 90ky BP (Mangerud et al, 2001).
Similarly, previous North American glaciations would have produced lakes of similar
extent to Lake Agassiz in the past. There would always have been a tendency for sudden
releases of water to have occurred towards the end of all previous glacial periods simply
due to the physics of ice dam collapse. If the analysis in foregoing sections is correct then
the logical implication is that this sort of large-scale polar shift, together with its
accompanying floods, almost certainly took place on more than one occasion during the
Pleistocene Era. It may not have occurred during every deglaciation phase but it is likely
to have happened during some.
2.4.6 Conclusion
If the hypothesis set out in the previous section is true then the principal effect would
have been a slow steady rise in ocean level and a non-violent drowning of vast swathes of
continental lowland. Only in certain specific locations where water flow was constricted
would violent flows have occurred, the Dardanelles and Bosphorus Straits being notable
examples. On land not reached by the ocean the only potentially significant effects would
have been changes to the gradients of rivers, possibly inducing diversions of flow.
CHAPTER 2 PHYSICS OF THE EARTH
27
The point was also made that major polar shift episodes may have occurred several
times during the Pleistocene Era, precipitated by the emptying of earlier glacial lakes at
the close of earlier glacial periods.
2.5 Summary
This chapter has argued, based on the known properties of the Earth, that the emptying of
Lake Agassiz would inevitably have initiated polar shift due to an imbalance in
centrifugal force. The magnitude of this motion has been shown to be critically dependent
on the flexural stiffness modulus assumed for the mantle-crust layer. It was then proposed
that the ongoing small-amplitude oscillation of the Earth known as the Chandler Wobble
can also be understood using the same form of analysis but with a much smaller system
excitation, and that in order to match the observed frequency of the Chandler Wobble it is
necessary to assume that the effective modulus of the mantle as a whole in flexure is
around 360GPa. However, it was suggested that this modulus should be expected to
‘strain-soften’ since discontinuities within the mantle layer would have a significant
effect, and that the degree of strain softening should be dependent on the distance
between tectonic plate boundaries. This led to the suggestion that different strain-
softening parameters would have applied along different lines of longitude. Under these
conditions it was found that the emptying of Lake Agassiz could potentially have induced
several degrees of tilt. It was then argued that dissipation of the centrifugal force
imbalance would have occurred, principally due to movement of material at the core-
mantle boundary, and that this would have allowed a second phase of polar shift, in the
opposite direction. However it was considered likely that the rate of energy loss in the
system would have been sufficient to prevent a third major tilting phase.
These calculations and deductions are logical and well founded. They are not
however reliable since they depend on several debatable assumptions, notably the extent
of strain softening and the properties in the ultra-low velocity zone (ULVZ) at the base of
the mantle. Thus the proposal made here can only be seen as one possible scenario,
selected from a range of alternatives. Whether this proposal is accepted therefore depends
almost entirely on evidence obtained from other fields. Reason has been presented for
believing that the proposed event could have occurred – but a small change in the
assumptions made would have resulted in quite different predictions. It has been
suggested that the actual extent of polar shift was around 7in each direction and a map
of consequent flooding has been presented, but this proposal now has to be tested against
the evidence of other disciplines. The next several chapters will assess whether the
predicted consequences of this proposed event are compatible with the information
available from geology, biology, archaeology, linguistics and history. Only if this
exercise provides sufficiently robust support can the hypothesis be upgraded from
‘possible’ to ‘likely’ – or even ‘practically certain’.
CHAPTER 3 GEOLOGICAL EVIDENCE
28
3. Geological Evidence
The point has been made that the event proposed in this report was not, in general, one of
violent upheaval of the sort that might be expected to leave obvious geological evidence.
The tilting of the planet in itself would have done nothing more significant than to change
the elevation of the sun for a short period of time. The flexure induced within the mantle
layer may possibly have been sufficient to cause earthquakes and volcanic eruptions, as
occur on Jupiter’s moon Io due to the ferocious gravitational effects of Jupiter itself. The
surface of Io is thought to change its elevation by around 100m during each of the
moon’s rotations (1.76 Earth days), which is less than is predicted to have taken place
during the proposed polar shift episode. Taking the relative size of Io into account – its
diameter is just over a quarter that of the Earth – the strains induced within the Earth’s
mantle would still have been slightly larger than those within Io. However the rate of
strain, and therefore the rate at which energy could have been transformed into heat,
would have been three orders of magnitude less than on Io. Thus it is debatable whether
the proposed event could have transferred sufficient energy into the mantle to cause any
measurable increase in volcanism.
An obvious short-term effect would have been seen in the oceans. The changes in
level relative to the continents would have forced short-term changes to current flow
patterns worldwide. The Arctic Ocean would have increased its volume by around 10%
over a period of about 4 years, drawing in warmer water through the Bering Strait and via
the Greenland Sea, inevitably affecting currents in the Pacific and North Atlantic. Across
the tropical and subtropical belts the changing location of the equator would have led to
significant changes in atmospheric circulation patterns relative to the land, leading to
shifts in the major currents of the Pacific and Atlantic. When the planetary motion came
to an end the oceans would have been left in an altered state in terms of temperature and
salinity and it is therefore possible that this could have triggered permanent changes in
oceanic circulation – although it would be difficult to prove that polar shift was the cause
of such a change.
One important effect of the suggested sea level rise should have been to break up
many of the ice sheets around the northern and southern polar regions. Since ice is lighter
than water a rise in sea level would float ground-bearing ice shelves, disconnecting them
from the adjacent land and potentially breaking them into bergs. Evidence may therefore
reasonably be sought in the form of an increase in ice-rafted debris.
Clear geological signals might also be expected to stem from the effects of the ocean
invading the land. This is unlikely to be revealed in terrestrial deposits since the average
rate of accumulation of sediment on the sea floor is typically less than 1mm per year.
Similarly, it is unlikely that reworking of soils in the wave zone would have left any
clearly identifiable signal today. However, geologically visible effects might reasonably
be expected in lakes and inland seas that had been subjected to seawater invasion and a
sudden change in salinity. Particularly in large open bodies of water, where decades or
even centuries may have been required to restore freshwater status, there would have
CHAPTER 3 GEOLOGICAL EVIDENCE
29
been a significant period during which both the water chemistry and the flora and fauna
present were markedly altered and this may in some instances have endured long enough
to provide a notable chemical and/or physical change in the sediment record. Closed
lakes, including ephemeral lakes and playas, would also have seen changed conditions for
a significant period.
Finally, there are specific locations where the rise and fall of the oceans relative to the
land should have produced a more violent and therefore more geologically visible effect.
Much the most significant of these are the channels and valleys separating the Black Sea
and Caspian Sea from the world’s oceans, namely the Bosphorus and Dardanelles, the
Manych spillway between the Black Sea and the Caspian, and the Turgay valley between
Aral-Caspian basin and western Siberia. In these specific locations very high flow rates
are predicted and massive erosion might reasonably be expected.
It is the task of this chapter to investigate each potential source of geological evidence
in turn. The first two topics ‘earthquakes and volcanism’ and ‘oceans’ are included for
completeness although, as intimated above, direct supportive evidence is not anticipated.
In other areas however it is certain that the proposed event, if real, would have left
distinct signals, and the remaining sections of this chapter will demonstrate that these
signals are indeed present and strongly supportive of the hypothesis set out in Chapter 2.
3.1 Earthquakes and Volcanism
Earthquakes are regular occurrences at all the major tectonic plate boundaries and so it is
not expected that any temporal concentration of slip events between tectonic plates 8200
years ago would be detectable today. The only types of earthquake that might be
discerned are therefore those distant from plate boundaries, where no subsequent
earthquake has occurred, and it is certainly the case that there is no particular
concentration of such earthquake evidence in around 8.2ky BP. It is true that the Parvie
and Landsjarv faults in northern Sweden date from this approximate time and that both
represent evidence for a massive earthquake in a location where earthquakes do not
normally occur. On the other hand it is quite reasonable to attribute them to the isostatic
rebound of the mantle following deglaciation, leading to differential uplift between
different locations.
Another interesting coincidence is the so-called ‘Storegga slide’, a large under-sea
landslip that occurred off the coast of Norway and which is dated to around 8.1ky BP
(e.g. Weninger et al, 2008). Dating accuracy is insufficient to distinguish this from the
8.2ky BP event and it is certainly logical that the warping of the mantle-crust layer during
polar shift could have triggered such an event.
Nevertheless it has to be conceded that earthquake-related evidence is insufficient
either to support the proposed polar shift event or to contradict it.
3.1.1 Volcanic Eruptions
Evidence for past volcanism is much more readily detectable. Zielinski et al (1994) report
that the period from 9ky BP to 7ky BP saw a concentration of volcanic activity as
CHAPTER 3 GEOLOGICAL EVIDENCE
30
deduced from Greenland ice core evidence. They make the point that not only were
volcanic eruptions approximately three times as frequent during this period as during the
most recent two thousand years but the magnitude of those eruptions was, on average,
significantly larger than has been the case in recent millennia. Admittedly this date range
includes a few hundred years before the proposed polar shift event and this may
legitimately be seen as evidence that the cause was something quite different, perhaps the
same rebound following deglaciation that may have been responsible for the Swedish
earthquakes. On the other hand it is also possible that the evidence relates in part to
straining during and after the proposed event.
Considering the strain distribution expected within the mantle under the effects of
polar shift, maximum values of bending are predicted to have occurred at latitudes 35-65
North and South of the equator. The volcanoes that might have been most affected are
therefore those in New Zealand, the southern Andes, the Rockies, the Mediterranean,
Japan and Kamchatka. In that regard it is noted that the Global Volcanism Program
(2010) lists 17 very large eruptions between 8.5ky BP and 8.0ky BP (adopting the median
of the date ranges quoted), including one in New Zealand, three in Japan and eight in
Kamchatka and the neighbouring Aleutian Islands; the other five occurred in the tropics.
In comparison, the previous 500-year period saw 13 very large eruptions, comprising
three in Kamchatka, two in Japan, two in the southern Andes, two in the Mediterranean
and four in the tropics. The subsequent 500 years also saw 13 very large eruptions, five in
Kamchatka, one in the Kuril Islands just to the south, one in Japan, two in the central
Rockies and four in the tropics. On this basis therefore it is hard to argue that there was
any statistically significant concentration either in the latitudes where maximum mantle
flexure is predicted to have occurred or in around 8.2ky BP. While it is possible to pick
out specific documented eruptions such as that of the now-dormant Hasan Dağ volcano in
southern Turkey, recorded in the form of a wall painting at the ancient settlement of Çatal
Höyük (Mellaart, 1967) and dating to the 9th Millennium BP, the evidence as a whole is
insufficient to prove anything other than the general upturn in volcanism worldwide
during the 9-7ky BP range.
3.1.2 Changes in Sea Level
One possible further signal of tectonic activity is sudden change in sea level relative to
the land. The Indian Ocean earthquake of 26th December 2004 for example, variously
estimated at between 9.1 and 9.3 on the Richter scale, caused significant raising and
lowering of shorelines around Sumatra and adjacent island groups.
An important point however is that it is not usually possible to be positive regarding
the actual degree of suddenness of geologically ‘sudden’ events. Hijma and Cohen (2010)
discuss an apparently sudden 3.0m jump in sea level in the Rhine delta region in the 8.4-
8.2ky BP time range, suggesting that it may have been caused by the combined water
volume of Lake Agassiz and large parts of the Laurentide Ice Sheet, although this
magnitude is larger than more usual estimates of this combined effect, typically 1.0-1.5m.
Törnqvist et al (2004) report evidence from the Mississippi delta that a local rise in sea
level of no more than 1.2m occurred in about 8.2ky BP, although Roblyn et al (2008)
point out that interpretation of this evidence is complicated by the need to take into
account the isostatic rebound following Lake Agassiz’s draining. Cronin et al (2007)
CHAPTER 3 GEOLOGICAL EVIDENCE
31
suggest that a steady rise of about 6.0m occurred between 8.2ky BP and 7.6ky BP in
Chesapeake Bay on the Atlantic coast.
Sharp sea level rises are also reported from Asia. Liu et al (2007) suggest that an
abrupt rise of a few metres would explain the sudden changes in sedimentation around
the Shandong Peninsula in north-east China that occurred from about 8.2ky BP. Hori and
Saito (2007) report evidence from three valleys, in China, Vietnam and Japan, that a
sudden sea level rise was responsible for the initiation of estuarine formation, whereas
subsequent deltaic development implies much more gradual sea level increase. In
Singapore Bird et al (2007) determined that levels rose very rapidly until about 8ky BP
and were then approximately constant (and only 3m lower than today) until 7.4ky BP.
Similar episodes of zero sea level increase are reported from Hawaii (Engels et al, 2004;
8.0-6.9ky BP), Australia (Larcombe and Carter, 1998; 8.2-7.5ky BP) and southern Spain
(Fernadez-Salas et al, 2001; 8.0-7.0ky BP). In some areas it is reported that sea levels
around 8.0ky BP were actually higher than today. Li et al (2009) found evidence in the
form of marine micro-organisms that Taihu lake, just south of the lower Yangtze, had
been inundated by the sea some time between 9.5ky BP and 7.4ky BP. Around the coasts
of Australia it is reported that a marine transgression occurred in the early Holocene. For
example Cann et al (1999) suggest that this state prevailed from 7.5ky BP in South
Australia, while Rowe (2007) reports a period of saltwater swamp development between
6.0ky BP and 3.0ky BP at a currently freshwater site on the Torres Strait islands. Thom
and Roy (1985) suggest that the Australian evidence overall points to a marine
transgression that developed into a period of almost zero change from about 6.5ky BP.
On the opposite side of the planet, Blanchon and Shaw (1995) deduced from the
sudden cessation of a coral reef off the Cayman Islands that a rapid 6.5m rise had taken
place some time between 8.1ky BP and 7.6ky BP, followed by a long period of almost
constant level, apparent from an incised shoreline feature – although this finding is not
reflected by evidence reported from reefs in the Pacific (Bard et al, 1996; Montaggioni et
al, 1997).
In southern Sweden Yu et al (2007a) report evidence for very sudden (within a few
years at most) and synchronous flooding of several river basins in 7.6ky BP with a sea
level rise of 4.5m. They too attribute this to collapse of parts of the Laurentide Ice Sheet
in Labrador, although Mörner (2004) suggested that the cause may have been a seismic
event of some sort. Behre (2007) has presented parallel evidence from northern Germany
of a 4.0m sea level rise in the 7.9-7.6ky BP period, while Lemke (2004) found that a
6.0m rise had occurred between 8.0ky BP and 7.7ky BP in the south-west Baltic.
To summarise, the picture is inconsistent. It is possible that some of these
observations may relate to adjustments within the Earth’s mantle following straining
during polar shift – but it is equally possible that they relate to a combination of the
acknowledged rise in sea level due to the release of water from Lake Agassiz and tectonic
events unrelated to and possibly later than the proposed polar shift episode.
3.1.3 Conclusion
There is no hard evidence of tectonic activity specifically related to the 8.2ky BP event,
despite the date lying within a period of heightened volcanism and irregular sea level
change. However, since it is debatable whether the predicted straining would have been
CHAPTER 3 GEOLOGICAL EVIDENCE
32
sufficient to induce such an effect, this finding merely constitutes absence of support
rather than evidence against the polar shift hypothesis.
3.2 Oceans
Whatever the true mechanism or mechanisms at work during the 8.2ky BP event, it is
acknowledged that it represents a period of altered climate and that it was intimately
connected with changes that took place in the North Atlantic. It would therefore be
expected that signals of this episode would be found in sediment cores from the North
Atlantic and possibly further afield. The question is whether any of these signals
represent evidence for the massive but short-lived re-arrangement of the oceans described
in this report.
The classic double-peak pattern of the 8.2ky BP event is illustrated by Hillaire-Marcel
et al (2007) in sediment cores from the Labrador Sea where concentrations of detrital
carbonate are believed to represent periods of accelerated meltwater release from the
Laurentide ice sheet. A very similar form is seen in a (smoothed) plot of the relative
abundance of the cold-loving foraminifer Neogloboquadrina pachyderma sinistral from
the Norwegian Sea (Ellison et al, 2006). Both are shown in Figure 3.1 in the form of
variations from a median value. In many records it is not possible to distinguish this
pattern with clarity, often because of the resolution of the data, but it is commonly
possible to discern a 200-300 year period of change. For example Keigwin et al (2005)
measured a minimum in oxygen isotope δ18O (from samples of Neogloboquadrina
pachyderma sinistral obtained from sediments off Newfoundland) lasting from about
8.3ky BP to 8.0ky BP. Andrews et al (2003) found peaks and minima in magnetic
susceptibility parameters from sediments recovered off the north coast of Iceland. Came
et al (2007) deduced a period of decreased temperature in the mid North Atlantic based
on the magnesium/calcium ratio in foraminifers. Yet these are simply indicators of a
complex period of climate change and not of any sudden event.
However the fact that a sudden event did indeed take place is shown by the δ18O
values (again from Neogloboquadrina pachyderma sinistral) plotted in Figure 3.1 from
one of Hillaire-Marcel et al (2007)’s Labrador Sea cores. Assuming that the whole event
spanned a period of about 300 years, as was indicated by the Greenland ice core
evidence, the period of very low δ18O shown in Figure 3.1 can have lasted no more than
about 20 years – possibly less according to Hillaire-Marcel et al – and it occurred late in
the event as a whole. It is logical to identify this as the time of the final outburst of Lake
Agassiz water, flooding into the Labrador Sea, and one might then assume that earlier
stages of the event were activated by significant earlier releases of glacial water, keeping
the sea temperature low.
A similar sudden event is also shown by Risebrobakken et al (2003) from the
Norwegian Sea. Their measurements of the percentage of Neogloboquadrina pachyderma
sinistral in the foraminiferal assemblage mirrors the double-peak form from Ellison et al
(2006); however they also present total concentrations of both Neogloboquadrina
pachyderma sinistral and dextral, and both show an extreme and very short-lived spike
CHAPTER 3 GEOLOGICAL EVIDENCE
33
during the second peak, plausibly correlating with the δ18O spike from the Labrador Sea.
Figure 3.1 illustrates, including adjustments of up to 150 years in dating to bring the
different records into synchronisation and to bring the spikes in plots (d) and (e) to
coincide with the 3-4 year period of ultra-low δ18O from the GRIP ice-core in Greenland,
referred to in Chapter 1 – Thomas et al (2007). These adjustments are acknowledged to
have no intrinsic justification but lie within the uncertainty band inherent in all 14C
dating. In the case of Hillaire-Marcel et al (2007) the dates are particularly problematic
because some of the sediments had clearly been reworked. A key question however is
whether the Norwegian Sea spike in Neogloboquadrina pachyderma could have been
caused solely by the sudden escape of Lake Agassiz’s water or not. In the context of the
hypothesis being put forward here, this extreme spike in a cold water indicator could also
reflect the mass liberation of icebergs from continental ice sheets due to a massive
temporary sea level increase.
Figure 3.1: Calcium carbonate (a) and δ18O (d) from the Labrador Sea (Hillaire-
Marcel et al, 2007; authors’ dates adjusted by approximately -150 years) and
Neogloboquadrina pachyderma sinistral from the Norwegian Sea ((b)Ellison et al,
2006; authors’ dates adjusted by -100 years; (c) and (e) Risebrobakken et al, 2003;
authors’ dates adjusted by approximately +140 years)
Moving away from the far north, Çağatay et al (2002) reported a short-lived
anomalous change in sediment from the Carolina shelf in the western Atlantic, with
abrupt changes in the relative proportions of different clay minerals. Although the dating
is not well constrained and the resolution is poor, the authors suggest that these changes
probably occurred due to climatic fluctuations during the 8.2ky BP event. In the Aegean,
Marino et al (2009) measured a spike in the proportion of the dinocyst species
Spiniferites elongatus dated almost exactly to 8.2ky BP, implying a very short period of
cooler water, while several authors (e.g. Roussakis et al, 2004) have reported a brief
hiatus in the organic sapropel deposition stage known as ‘S1’ in the eastern
Mediterranean and Aegean, dating approximately to 8.2ky BP. Kim et al (2007) reported
0
50
100
150
200
250
300
8 8.1 8.2 8.3 8.4 8.5 8.6
Date (ky BP)
% of median: 8.0-8.6ky BP
a) Calciumcarbonate [% - 20]
b) N. pachyderma(sin) [%]
c) N. pachyderma (sin) [%]
d) Oxygen isotope [‰ - 2]
e) N. pachyderma(sin) [/g+100]
δ
18
O
CHAPTER 3 GEOLOGICAL EVIDENCE
34
alkenone-derived sea surface temperatures from sediment off the coast of Morocco,
deducing a 300-year period of reduced temperature. In the Scotia Sea north-east of the
Antarctic peninsula Bak et al (2007) found that a step change in the diatom assemblage
took place at some time in the 8.4-8.0ky BP period as well as the organic carbon content
of the sediment, and they deduced that what had previously been near-continuous sea ice
had been fairly suddenly transformed into open water. Even across the Pacific in the
Okinawa trough Dou et al (2010) deduced that in 8.2ky BP a fundamental and permanent
shift in the Kuroshio current took place, bringing it west of the Ryukyu Islands (between
Japan and Taiwan), while Xiang et al (2003) and Yu et al (2008) have both reported
short-lived spikes (and dips) in several foraminiferal species and similar dramatic short-
lived changes in mineral proportions from Okinawa trough sediments – very like those
found by Çağatay et al (2002) off the Carolina coast. Yu et al (2008) ascribe these
phenomena to the 8.2ky BP event. Unfortunately in none of these cases is the resolution
good enough to determine over how short a period the changes took place.
The key issue is whether it is reasonable to propose that the flow of glacial water into
the North Atlantic between about 8.5ky BP and 8.1ky BP, which all experts acknowledge
took place, was sufficient to cause changes to the oceans on a worldwide scale. That it
could have affected the climate over much of the world is supported by computer
simulations (e.g. Wiersma and Rennsen, 2006) – and this will not be challenged here.
However it is less clear that it could have initiated dramatic short-lived changes to the
sediments of the East China Sea and the eastern seaboard of the United States, whereas
the proposed polar shift event would have fundamentally altered ocean currents across
the world over a period of several years and it is quite reasonable that these currents
would have taken a long time to re-establish their former routes – and indeed that some
may never have done so.
To conclude this section, it is clear that a dramatic short-lived event took place in
approximately 8.2ky BP towards the end of a period of 200-300 years with cooler
climate. However it is unfortunately much less clear what the exact nature of the event
was or to what extent it was responsible for the many changes in ocean behaviour
reported from this approximate time. Overall therefore the evidence from the oceans
remains inconclusive.
3.3 Ice Sheets
The melting of continental ice sheets has been responsible for over 100m of sea level rise
since the last glacial maximum. However the proposal examined here is that a massive
short-term rise and fall in ocean level may have been responsible for a sudden
acceleration in the rate of loss of continental ice. The effect of massive sea level rise on
continental ice sheets and coastal glaciers is not difficult to predict; they would obey the
laws of physics and therefore float. Loosely attached ice sheets would be likely to break
off; coastal glacier ice and ground-bearing continental ice shelves might remain in
position horizontally due to the constraints of the ground topography but they would
nevertheless tend to rise as the sea level rose. If they were unable to move significantly
CHAPTER 3 GEOLOGICAL EVIDENCE
35
horizontally they would then sink back to their approximate original positions as the
water level fell; if the water rose high enough they would be likely to drift and would
either be returned to a different position or be carried out to sea.
In this context it is therefore of interest that a sudden massive peak in sand-size debris
was reported by Moros et al (2004) in sediments beneath the Norwegian Sea dated
approximately to 8ky BP, within normal 14C uncertainty range of 8.2ky BP (see Figure
3.2). Since ice-rafting represents virtually the only means of transporting sand-size
particles into the open ocean this implies a sudden release of icebergs from continental
ice sheets. The episode was very short-lived. For such large numbers of bergs from
continental ice sheets to have been melting at approximately the same time is, given the
gradual nature of glacier flow and episodic calving of bergs, an anomalous occurrence.
0
0.5
1
1.5
2
2.5
3
6 7 8 9 10
Date (ky BP)
Percentage >150 microns
Figure 3.2. Sudden spike in sand-size debris, Norwegian Sea (Moros et al, 2004)
3.3.1 George VI Ice Shelf
The George VI Ice Shelf is located to the west of the Antarctic Peninsula, spanning 20-
30km between the mainland of Antarctica and the 400km-long Alexander Island. As an
ice ‘shelf’ much of it is bedded directly on the bottom of George VI sound. Bentley et al
(2005) report on a sediment core taken from a so-called epishelf lake, a small area of
open water trapped against Alexander Island by the ice shelf. They discovered that the
epishelf lake had experienced a period of fully marine conditions during the early
Holocene, implying that the ice shelf was no longer present. The actual sequence revealed
by the sediments was of an initial and very short-lived spike in marine foraminifera and
diatom species, followed immediately by deposition of 2m of unsorted ice-raft debris
including igneous clasts with close affinity to the rocks of Palmer Land on the opposite
side of George VI sound (Roberts et al, 2008), followed again by about 0.7m of sediment
containing marine species. The rapidity of the ice-raft debris deposition event was
emphasised by the fact that almost identical 14C dates were obtained from foraminifera
samples taken above and below the debris layer, compared to a spread of several hundred
years determined for the 0.7m deposit above the debris.
CHAPTER 3 GEOLOGICAL EVIDENCE
36
In terms of physical evidence this sequence matches what might be expected
following polar shift extremely well. The prediction is for a local rise in sea level of
around +280m, floating the ice shelf and the lower portions of adjacent glaciers in Palmer
Land and throughout the Antarctic Peninsula, and bringing a temporary influx of much
warmer water. Rapid melting, break up of the ice and deposition of debris would have
been the inevitable consequence – indeed the rapidity of the ice shelf melt appears most
surprising if no catastrophic agent was involved. It is then perfectly reasonable that it
took several hundred years for the ice shelf to re-form across the George VI sound once
more before it could again hem in the epishelf lake against Alexander Island.
The problem however is the date. Both Bentley et al (2005) and Roberts et al (2008)
use a 1300-year correction to 14C dates to account for marine carbonate based on
recommendations by Berkman and Forman (1996) and arrive at a date of 9.6ky BP for
the onset of ice shelf break-up. In order to match the 8.2ky BP event the correction would
have to be about 2700 years, a subject highlighted in the next subsection.
3.3.2 Amery Ice Shelf
The degree of uncertainty present in dating Antarctic samples is illustrated by the 14C
ages of material recovered from sediment beneath the Amery Ice Shelf in East Antarctica
(Hemer and Harris, 2003). Here, seabed surface samples from two different locations,
both beyond the region of ground-bearing ice and therefore currently subject to
sedimentation (i.e. the samples were very recently deposited), provided uncorrected ages
of 6548BP and 11722BP. On this basis the authors decided that it would be appropriate to
replace the expected 1300-year correction for Antarctic waters with 6548 years – which
puts the 2700-year correction required to bring the date of the George VI Ice Shelf break-
up to 8.2ky BP into context.
In fact the single sediment core extracted from beneath the Amery Ice Shelf shows a
sudden spike in the abundance of the diatom species Fragilariopsis curta at the transition
between the uppermost stratum, representing current conditions of sediment deposition,
and the immediately underlying stratum. This suggests that for a short time the water
overlying the core location was within or close to the sea-ice zone rather than being
beneath the ice shelf itself, whereas it is currently 100km behind the ice shelf front; the
implication is that the ice shelf had temporarily (and suddenly) retreated. With a 6548-
year correction this retreat of the ice shelf and accompanying change in sediment regime
is dated to 13.15ky BP; had the authors decided to adopt an 11722-year correction (based
on the other surface sample collected) the date would have been about 8ky BP.
Nevertheless, while the physical data from both the George VI and Amery Ice
Shelves is strongly supportive of an event that led to sudden ice shelf break-up, it has to
be conceded that confident dating is not possible.
3.3.3 Ross Ice Shelf
The Ross Ice Shelf is the largest ice shelf on Earth. It is about 800km by 600km and
much of it is bedded on the bottom of the Ross Sea, a giant inlet in the continent of
Antarctica that extends to within 500km of the South Pole itself. It is effectively part of
the ice cap that covers most of the Antarctic continent and it is the opinion of most
scientists that the Antarctic ice cap dates from the middle Miocene Era, 10-20 million
years ago. It would therefore be expected that the sediments immediately underlying the
CHAPTER 3 GEOLOGICAL EVIDENCE
37
Ross Ice Shelf were of middle Miocene date or earlier. However Raiswell and Tan (1985)
showed from sediment core analysis that this was not the case. They analysed data from a
core taken from the middle of the ice shelf, about 350km from the ocean, and reported
evidence for repeated reworking of the sediments, the most recent such reworking having
taken place no earlier than the late Pleistocene (see also Kellogg and Kellogg, 1981).
Analysis of the sediment content also exposed the fact that, while the majority of the
microscopic flora present was of middle Miocene type, representative Pliocene (5-2
million years ago) and Pleistocene (<2 million years ago) diatoms were also present in
small quantities, mixed in with the Miocene sediments.
The actual layering of the sediments consisted of an upper, recently reworked, layer
overlying a very thin (less than 1cm) iron-stained lens with older, less recently reworked
materials beneath, although the less recently reworked material still included late
Pleistocene diatoms. The iron-stained lens is particularly interesting. According to the
authors it could only have formed due to contact with oxygenated seawater, implying that
the entire ice shelf had been lifted from its bed at some time no earlier than the terminal
Pleistocene – and the dating evidence from Amery Ice Shelf sediments warns that the true
date may have been significantly more recent. Furthermore theoretical considerations
suggested that the iron-stained layer would have formed at about 1cm in 100 years,
implying less than 100 years of seawater contact.
Raiswell and Tan (1985) offered no scientific explanation for their findings, although
Majewski et al (2004) have proposed that a retreat in the Ross Ice Shelf in this period
should be associated with the 8.2ky BP event. Nevertheless there is a difference between
a retreat, which certainly occurred (e.g. McKay et al, 2008a), and a lifting of the entire ice
shelf over a distance of at least 350km.
Raiswell and Tan’s observations are however fully compatible with the event
proposed in this report. The Ross Ice Shelf is constrained by the topography of the land in
that it is effectively pinned by islands and so could not have been moved significantly
even if raised temporarily from its bed. The polar shift motion described in Chapter 2
would have seen a local rise in sea level of about +75m. Thus it would be expected that
sea water would have been sucked beneath the ice shelf and then expelled and that this
would have produced considerable water flow velocities. Reworking of sediments and
introduction of Southern Ocean diatoms of recent origin would be the inevitable
consequence. After about 7 years of raised ocean level the entire ice shelf would have
settled slowly back onto its bed before experiencing a second lifting episode a few years
later, during the second phase of polar shift motion (see Appendix A, Figure A4). The
detailed shape of the bed would inevitably have changed due to sediment reworking and
it is not unlikely that there may have been some horizontal shift of the ice sheet; thus the
match between the profile on the underside of the ice and the sediment surface would no
longer have been perfect, which means that pockets of seawater would have been trapped.
This contact with seawater would have created the iron-stained layer; it would then have
been overlaid in some areas by additional sediment as the ice shelf imposed itself onto the
contours of the sea bed following the second lifting episode, redistributing sediment as
necessary.
This evidence is clearly supportive of the hypothesis that a dramatic sea level rise
episode occurred in the early Holocene, although the lack of precise dating is unfortunate.
It also throws light on the possibility that similar events may have occurred in earlier
CHAPTER 3 GEOLOGICAL EVIDENCE
38
epochs. The sediments found contained diatoms from throughout the Pleistocene and
even earlier, suggesting repeated reworking and episodic ingress of oceanic water. Thus,
if the explanation offered here is correct regarding the most recent reworking phase then
it is possible that a similar explanation applies to several earlier occurrences.
3.3.4 Conclusion
Whilst the physical evidence presented in this section implies a highly unusual event in
the late Pleistocene or early Holocene of very short duration, compatible with the
predicted rise in sea level and ‘floating-off’ of large sections of continental ice sheets and
Antarctic ice shelves, it has to be acknowledged that dating is generally uncertain. The
exception is the Norwegian Sea ice-rafting evidence which comes from high-resolution
samples that were confidently dated to within a few hundred years of 8.0ky BP.
3.4 Lakes and Inland Seas
Lakebed sediment contains an important record of the state of a lake and its catchment.
Sedimentation rate, grain size distribution and chemical proportions are functions of the
sediment content of inflow water, which in turn depends on the nature of the catchment,
vegetation cover, steepness and amount of precipitation. Thus a change in the catchment,
for example decreased vegetation cover, should be reflected in the sediment.
Furthermore, sediment contains micro-organisms and the relative quantities of different
types are affected by temperature, salinity and water depth. It also includes washed-in
vegetative matter from the land and it is possible to detect changes in the relative
proportions of terrestrial and aquatic flora by measuring the carbon/nitrogen (C/N) ratio
present, higher values implying a greater contribution from terrestrial flora, while the
total organic carbon (TOC) content reflects the quantity of dead organic matter being
deposited – usually directly related to the biological productivity of the lake. Isotope
ratios, notably δ18O and δ13C, present in organic matter, can also give information on
climatic factors such as temperature, although there will often be more than one possible
explanation for any change observed. Similarly, changes in the number and distribution
of pollen grains found are likely to reflect climatic changes in the region.
In the context of the proposed 8.2ky BP inundation event it is believed that the key
indicators are likely to be TOC, the C/N ratio, sediment composition and, possibly, pollen
concentration. A sudden change in salinity is likely to kill most lake life while inundation
of the catchment would be expected to do the same for terrestrial plants. The logical
result would be a sudden peak in TOC as large quantities of dead organic matter were
deposited on the lakebed. There would usually be a similar spike in the C/N ratio because
the relative contribution of aquatic and terrestrial flora would change – although whether
that spike was positive or negative would depend on lake and catchment details and
whether the sediment core was retrieved from a location close to an inflow stream or not.
Changes in inorganic sediment composition would be expected if there had been a loss of
vegetation cover in the catchment and therefore an altered susceptibility to erosion, while
CHAPTER 3 GEOLOGICAL EVIDENCE
39
changes in pollen type and concentration would naturally be seen during re-establishment
of catchment vegetation.
It is also possible that other changes would have occurred, particularly those related
to climate, since it is already accepted that the 8.2ky BP event included significant
climate change in many parts of the world. It is unlikely that any signal related directly to
the change in water chemistry would be visible in most open lakes because of the
relatively short duration involved. Freshwater inflow would very rapidly have led to a
near-fresh surface layer – although the time taken to remove high-salinity water from the
lower part of the water column would then depend on the efficiency of mixing and the
depth and topography of the lake. It is possible that in some cases evidence of water
stratification (fresh overlying saline) might be found.
In closed lakes it is more likely that a permanent change in water chemistry would
have occurred, quite possibly combined with the initiation of stratification of the water if
mixing was poor. In these cases δ18O or salinity indicators might be expected to undergo
a sudden jump.
However this is not an easy subject to treat impartially. All lakes and their catchments
undergo continuous development and this sometimes leads to abrupt changes in the
sediment. Thus it is not generally necessary to postulate a saltwater flood in order to
explain peaks or rapid fluctuations in any of the parameters. Furthermore there is always
uncertainty with regard to 14C dating, particularly where significant corrections are made
(or omitted) due to reservoir effects, i.e. the retention of old carbon in the system. In
general it is clear that an uncertainty margin of about ±300 years is to be expected.
Finally it is inevitable that only a sample of the world’s lakes and inland seas have been
cored and relevant sediment parameters measured at an appropriately fine resolution. It is
therefore difficult to claim that a balanced world view has been achieved. It is also the
case that each research team measures and publishes different parameters for the
excellent reason that they are not all investigating the same issues – and none have been
directly investigating the proposal being made here. Nevertheless, the next subsection
will attempt to take an overview of the data available and to evaluate whether there is any
discernible pattern or a likelihood that it could relate to the proposed inundation event.
This, it is hoped, will lead to an unbiased assessment of the evidence in its generality.
Only once this step has been taken will individual cases be examined in greater detail in
subsequent subsections.
The final three subsections prior to the summary will concentrate on the American
Great Lakes (Superior, Michigan, Huron, Erie and Ontario), the African Great Lakes
(Victoria, Tanganyika and Malawi), and Lake Chad, where specific issues apply.
3.4.1 Overview
This study has examined data relating to 90 lakes and inland seas. This excludes the
Black Sea and the Caspian and it also excludes salt lakes and playa deposits; these will be
investigated separately in later sections.
Of the 90 total, 43 are predicted to have been flooded by the proposed polar shift
event, 44 are definitely above flood level and three are marginal cases and will be
excluded from analysis. The size distribution of each set is shown in Figure 3.3 and it is
clear that the two distributions are not dissimilar. Furthermore each set includes
approximately 25% closed lakes and 75% open.
CHAPTER 3 GEOLOGICAL EVIDENCE
40
The published data for each lake was examined for changes that could plausibly
(given uncertainty in 14C dating) be dated to 8.2ky BP, categorized as follows: a) sharp
and short-lived, b) longer-lived but reversed within a few hundred years, and c) long-
lasting. It was also noted where data clearly showed no significant or unusual change. A
full summary of the findings is given in Appendix B.
Figure 3.3. Size distribution of lakes included in this study
Figure 3.4 presents the percentages in each change category, divided into lakes where
flooding is or is not predicted and also into large and small lakes; and a difference
between the two data sets is immediately evident. While less than 10% of lakes where no
flooding is predicted show short-lived change, this becomes nearly 90% for lakes over
1km2below the predicted flood level and nearly 50% for lakes under 1km2. Since short-
lived changes are exactly the types of change expected from the seawater invasion event
proposed in this report, this evidence is clearly very strongly supportive.
Figure 3.4. Proportions of lakes subject to different change types
0.01
0.1
1
10
100
1000
10000
100000
1000000
0 20 40 60 80 100
Percentage smaller
Lake surface area (km 2)
Flooding predicted
No flooding predicted
0
10
20
30
40
50
60
70
80
90
100
Short-lived changes Longer-lived
changes Long-lasting
changes
Percentage of lakes
Flooding predicted - <1km2
Flooding predicted - >1km2
No flooding predicted - <1km2
No flooding predicted - >1km2
CHAPTER 3 GEOLOGICAL EVIDENCE
41
The other types of change show much less difference between data sets, although
there are still more recorded changes in lakes below the flood level, particularly in the
case of long-lasting changes. This is reasonable. Climate change was occurring rapidly
throughout the early Holocene and it is to be expected that the 8.2ky BP event would
have initiated changes that either failed to reverse or at least lasted for several hundred
years, and such changes would be seen in sediments from all categories of lake. However,
the greater number of long-lasting changes shown for lakes where flooding is predicted
could plausibly be due to permanent alterations to either lake water chemistry or the local
ecosystem, brought about by seawater inundation.
Nevertheless, it is evident that this section should primarily be concerened with the
short-lived changes that would be an almost inevitable consequence of seawater ingress.
Furthermore, it is noted that a higher percentage of large lakes show short-lived changes
than do smaller lakes, which is unsurprising since small lakes, particularly open lakes, are
replenished very rapidly. Thus, if saltwater ingress occurred, freshwater status would be
restored shortly after the flood had retreated. Dead vegetative matter from the land may
have been washed in during the immediate aftermath, but with a small catchment area it
is likely that this phase would also have been relatively short-lived. In effect the whole
process of normalisation would have occurred much more rapidly than might be expected
in large lakes with extensive catchments and so it is much less likely to show up in
sediment samples which are commonly at least 50 years apart in terms of deposition date.
For this reason the next three subsections will concentrate principally on the evidence for
short-lived changes in lakes over 1km2in area.
3.4.2 Total Organic Carbon (TOC)
Records of TOC were obtained from 21 lakes over 1km2, eleven below the predicted
flood line and ten above. In only one case above the flood level, Lake Bosten in Xinjiang
Province in western China (Wünnemann et al, 2003), was any short-lived change noted.
There, records revealed a 400-year period of fluctuating TOC, ascribed to variations in
the input of fluvial sand, with a peak that lasted no more than about 100 years which the
authors suggest was a phase during which the lake level dropped by around 200m due to
reduced rainfall, resulting in a hydraulically closed state. It is clear therefore that a
reasonable explanation exists.
In contrast, of the eleven TOC records from lakes over 1km2where flooding is
predicted, ten reveal short-lived peaks that could, acknowledging uncertainty in dating,
relate to 8.2ky BP. Just one shows no change, namely Lake Winnebago in Wisconsin
(Lovan and Krishnamurthy, 2000); however the resolution of the data from Lake
Winnebago is only 300-400 years, which means that a short-lived TOC increase would
probably have been missed. The shallowness of the lake (5m depth on average) also
means that water retention times are short and the impact of inundation would
consequently also have been short. Of the other ten cases, eight are reproduced in Figure
3.5. In graph a) the dates shown are those given by the authors concerned; in graph b)
they have been adjusted as necessary (by up to 400 years in the case of Lake Huguang
Maar) to show a coincidence of effects at 8.2ky BP. The two lakes that have been omitted
are Nakaumi Lagoon in Japan (Sampei and Matsumoto, 2001), which has a plausible
peak but very poorly constrained dates, and Lake Inchiquin in Ireland (Diefendorf et al,
2007), which has a very minor TOC peak but a much more impressive spike in C/N ratio.
CHAPTER 3 GEOLOGICAL EVIDENCE
42
The data in Figure 3.5 are expressed relative to an average since the absolute TOC
levels range from less than 1% for Lama Lake in northern Siberia (Nowaczyk et al, 2001)
to almost 40% (at 8.2ky BP) in the southern Baltic Sea (Yu et al, 2007b) – the Baltic
having been an open freshwater lake during that period. Naturally the TOC percentage
depends on core location as well as details of the local flora, and the Baltic Sea data
shown in Figure 3.5 is paralleled by data from another core remote from any river inflow
that shows no change in TOC at all. As Yu et al (2007b) deduce, the spike in TOC in
approximately 8.2ky BP was certainly caused by a sudden influx of river-borne terrestrial
organic matter. A similar explanation is proposed by Augustinus et al (2008) in relation
to Lake Pupuke in New Zealand, a lake that in every respect is quite different from the
Baltic Sea. It is a closed crater lake just 1.1km2in area with a very small catchment and is
on the opposite side of the planet – yet the same effect is noted and the authors even
suggest that the increase in terrestrial organic matter may have been linked to climate
change brought about by the 8.2ky BP event. Of the other lakes shown, West Hawk
(Teller et al, 2008) and Huguang Maar (Yancheva et al, 2007) are both deep crater lakes
like Lake Pupuke, although West Hawk is much larger (12km2) and is open. The two
Siberian lakes (Nowaczyk et al, 2001; Andreev et al, 2005) and Haukadalsvatn in Iceland
(Ólafsdóttir, 2010) are medium-sized, open and within the tundra vegetation zone, while
Lake Biwa in Japan (Meyers, 1998) is large (600km2), open and has an extensive
catchment area. All have depths measured in tens of metres.
Figure 3.5. TOC data from lakes >1km2where flooding is predicted
0
50
100
150
200
250
300
350
6 7 8 9 10
Date (ky BP)
TOC (% of mean 6-10ky BP)
Baltic Sea
Lake Biwa [Japan]
Lama Lake [Siberia]
West Haw k Lake [Canada]
Lake Lyadhej-To [Siberia]
Haukadalsvatn [Iceland]
Lake Pupuke [New Zealand]
Lake Huguang Maar [China]
0
50
100
150
200
250
300
350
TOC (% of mean 6-10ky BP)
Dates as reported
(Lake Biwa & West
Hawk Lake estimated
from data given)
Dates adjusted
a)
b)
CHAPTER 3 GEOLOGICAL EVIDENCE
43
If this relatively small sample of lakebed data is representative of lakes across the
planet then this clearly constitutes very strong evidence in support of there having been a
a sudden vegetation wash-in event over large swathes of land at relatively low elevation,
and the fact that a very clear distinction has been found between lakes above and below
the proposed flood level is powerful circumstantial evidence that the cause was indeed
the suggested saltwater flood.
3.4.3 Other Organic Sediment Indicators
The carbon/nitrogen ratio (C/N) is commonly used to determine the source of organic
matter in the sediment, a high value implying increased contribution from terrestrial
sources. In many cases a peak in TOC is mirrored by a peak in C/N. This is true of Lakes
Biwa, West Hawk, Pupuke, Inchiquin and Nakaumi Lagoon, while C/N is not quoted for
the other sites mentioned in the previous subsection other than Lake Winnebago – which
again shows no change. However Lake Potrok Aike in southern Argentina may be added
to the list of relatively large lakes (10km2) with spikes in organic sediment parameters on
the basis of C/N since Mayr et al (2007) and Haberzettl (2006) report a sharp C/N peak.
The authors date this occurrence to about 8.6ky BP, but this depends critically on the age
of a layer of volcanic tephra from an eruption of the southern Chilean volcano Mount
Hudson, which carries considerable uncertainty.
One further lake that should be mentioned, despite its small size, is Lake Flakkerhuk
in Greenland (Bennike and Funder, 1997) which includes a 2cm-thick highly organic
bryophyte-rich sediment layer dated to approximately 8ky BP. The bryophyte in question,
Drepanocladus exannulatus, is a freshwater moss, common in the lake today, and it has
formed a proportion of the organic content of sediment throughout the Holocene. Yet the
sudden and short-lived abundance, having been rare in the layer immediately preceding,
suggests the sudden and widespread death of plants throughout the lake – something that
could potentially have been achieved by saltwater ingress. Similar thin layers of
concentrated bryophyte deposit have also been reported from other lakes in the region
(Funder, 1978).
3.4.4 Inorganic Sediment
A short-lived change in inorganic sediment, either in mineralogy, size distribution or rate
of deposition, is likely to indicate a change in catchment vegetative cover. Inorganic
sediment records were obtained from 19 lakes over 1km2, 12 below the predicted flood
line and 7 above. Again, the only case above the suggested flood level that showed this
type of change was Bosten Lake in China (Wünnemann et al, 2003), discussed previously
in relation to fluctuations in fluvial sand input. In contrast, eight out of twelve records
from potentially flood-affected lakes over 1km2showed short-lived peaks. Admittedly the
effects are different in each case. Lakes Huron and Michigan (Odegaard et al, 2003) show
relatively minor reductions in mean sediment size while Chaohu Lake in eastern China
(Wang et al, 2008) has a minor spike in the percentage retained on 64μm, i.e. the sand
fraction, and Lake Pupuke in New Zealand (Augustinus et al, 2008) saw an increase in
mean grain size coincident with the increased TOC referred to above. The other four
short-lived changes are in mineralogy. The case of West Hawk Lake in Manitoba will be
discussed separately below. The change in Huguang Maar Lake in southern China
(Yancheva et al, 2007) relates to magnetic susceptibility and therefore principally to the
CHAPTER 3 GEOLOGICAL EVIDENCE
44
iron content, although the peaks are relatively minor. A core taken near the shore of Lake
Potrok Aike in Argentina (Haberzettl et al, 2009) showed a very short-lived spike in the
calcium/titanium ratio (Ca/Ti) in the sediment, from a background value of 3 to a peak of
33. Although the authors suggested that this represented a sudden fall in lake level, it is
noted that lesser spikes in Ca/Ti at around 7.5ky BP are replicated in a lake centre core
whereas the 8.2ky BP spike is not, suggesting the possibility of a different cause. Finally,
analysis of a core from the Aral Sea (Le Callonec et al, 2005) includes assessment of five
elements and shows possible peaks in strontium and sodium (although not in manganese,
iron or magnesium). Thus there is clearly no single signature from the inorganic sediment
composition; the only common factor is that some sort of short-lived change occurred –
while there was generally no such change in lakes above the predicted flood level.
The four lakes over 1km2and below the proposed flood level with inorganic sediment
records that show no short-lived change all show long-lasting step changes in one or
more parameter datable to approximately 8.2ky BP. In Lake Inchiquin in Ireland
(Diefendorf et al, 2007) the sedimentation rate increased fourfold; in one of the Baltic Sea
cores (Yu et al, 2007b) it decreased dramatically; in Lama Lake in Siberia (Nowaczyk et
al, 2001) there was an abrupt change to a finer gradation; while in Lake Balaton in
Hungary (Sumegi et al, 2008) the nature of the sediment changed from uniform peat to
laminated deposits including molluscs. Although it is impossible to ascribe any of these
changes to a saltwater inundation event with any confidence, it is a notable coincidence
that no long-lasting changes were recorded from any of the lakes greater than 1km2above
the predicted flood level. In fact other than the short-lived changes to Bosten Lake noted
above, the only other sediment change noted was a broad peak in calcite content in Bow
Lake in Alberta (Leonard and Reasoner, 1999), plausibly attributable to long-term
climate change.
Figure 3.6. Mineralogy spike – West Hawk Lake, Manitoba (Teller et al, 2008)
The case of West Hawk Lake in Manitoba, Canada (Teller et al, 2008) is illustrated in
Figure 3.6 since the data was of particularly high quality and fine resolution and it
50
60
70
80
90
100
110
120
130
140
150
50 55 60 65 70 75 80
Core depth (cm)
ppm (% of mean 50-80cm)
Sodium
Magnesium
Titanium
Strontium
Barium
Zinc
Abrupt change in mineralogy
- in approximately 8.2ky BP
CHAPTER 3 GEOLOGICAL EVIDENCE
45
included X-ray diffraction analysis of the key elements present. This showed that at a
particular depth in the core there were simultaneous spikes in the concentration of 17
elements – six of which are reproduced in Figure 3.6. Dating is imprecise but, based on
parallel pollen evidence showing the sudden arrival of white pine a few centuries later, it
is considered that a date of 8.2ky BP is approximately correct for this spike in mineral
composition, a spike that also ties in with the TOC and C/N data presented previously
and with a massive peak in magnetic susceptibility. While a detailed explanation is
beyond this work, it is logical to propose that an abrupt and short-lived change in the
elements present represents a change to either the catchment or the lake water or both.
The only other lake found where similarly detailed element analysis had been reported
was Erhai Lake in China (Shen et al, 2005), situated well above the predicted flood level,
and for which no such spike is evident throughout the Holocene until recent times.
In summary, the picture is consistent with that for organic sediment in that all lakes
below the predicted flood level show either short-lived changes or non-reversed step
changes in inorganic sediment, while those above generally do not. However the nature
of those changes varies considerably, as might be expected depending on the details of
both catchment area and core location.
3.4.5 Pollen
The composition of pollen grains within sediment is classically used to identify changes
in climate and it would therefore be expected that signs of the widely accepted 200-300
year climatic downturn might be present in many cases in around 8.2ky BP, whether
above or below the predicted flood level. It is also likely that longer-term climate-related
changes might have initiated during this period. These changes in pollen composition
would logically have affected both large and small lakes alike, with no reason for any
contraction in the timescale for smaller lakes, and so no justification for their exclusion.
No distinction will therefore be made here with regard to lake area. The actual
distribution of the different types of change found is shown in Figure 3.7.
Figure 3.7. Numbers of lakes subject to different pollen change types
0
5
10
15
20
25
Flooding predicted No flooding predicted
Total number of records
Short-lived changes
Longer-lived changes
Long-lasting changes
No change
CHAPTER 3 GEOLOGICAL EVIDENCE
46
Despite an expectation that climate change would affect both high and low ground
alike and therefore affect all types of lake equally, it is actually clear that many more
changes potentially attributable to 8.2ky BP are seen in lakes where flooding is predicted
than those where it is not. Just two changes out of 14 records apply to lakes above the
predicted flood level, both from north-west China. Lake Zhuyieze (Chen et al, 2001)
revealed a broad peak in conifer pollen between 8.3ky BP and 8.1ky BP, although it
should be noted that this peak is just one of five similar peaks between 11ky BP and 7ky
BP; while in Lake Sanjiaocheng (Zhu et al, 2002) there was a sharp spike in Ulmus and a
corresponding dip in Picea, suggesting a temporary increase in temperature.
In contrast, twelve out of 21 records from potentially flooded locations indicate a
significant change. Lake Raigastvere in Estonia (Seppä and Poska, 2004) shows relatively
short-lived peaks in Betula and Picea with corresponding reductions in deciduous taxa
such as Alnus,Corylus and Ulmus, suggesting a short period of cooler climate. A similar
but less marked effect is seen in two other Estonian lakes, Lake Viitna and, allowing for
minor dating adjustment, Lake Ruila (Seppä and Poska, 2004). Another Estonian lake,
Lake Rouge (Veski et al, 2004), with very similar changes in pollen proportions, also
evidenced an extraordinarily short-lived spike in deposition rate for all taxa – reproduced
in Figure 3.8. This evidence, which is provided by very high resolution sampling, is most
revealing; not only does there appear to have been a change to a cooler climate lasting
about 300 years but superimposed on that is an abrupt (less than 15 years) pollen wash-in
event, most pronounced in Betula, at that time the dominant taxon comprising over 50%
of the grains found.
0
50
100
150
200
250
300
350
400
7.5 8 8.5 9
Date (ky BP)
Influx (% of mean 7.5-9ky BP)
Betula
Alnus
Tilia
Ulmus
Figure 3.8. Pollen concentration –Lake Rouge, Estonia (Veski et al, 2004)
Højby Sø in Denmark (Rasmussen et al, 2008) also has high-resolution pollen-count
data and also shows a similar reduction in deciduous taxa over a 100-200 year period and
a slight increase in Betula and Pinus, suggesting a short climatic downturn. Although it
shows no clear spike in pollen deposition rate, Højby Sø has no natural inflow stream and
so could not have experienced any wash-in event.
CHAPTER 3 GEOLOGICAL EVIDENCE
47
Seppä et al (2007) report on several other Scandinavian lakes and find very similar
changes in pollen composition in Lakes Flarken in Sweden and Arapisto in Finland to
those noted above from Estonia and Denmark – although it has to be admitted that no
such change was evident from Lakes Nautajärvi and Korttajärvi in Finland (Tiljander,
2005), also below the predicted flood level.
Moving away from Scandinavia, Lake Potrok Aike at the southern extremity of
Argentina has been extensively studied. Mayr et al (2007) measured a very high pollen
count commencing very suddenly in about 8.7ky BP with a 20- to 30-fold increase.
Despite the authors’ date estimate, it is considered likely here that the pollen increase,
which coincides approximately with the C/N peak referred to above and a sudden 20-fold
increase in sedimentation rate, is actually related to the event described in this report,
since dating is uncertain and relies heavily on the timing of volcanic tephra layer
deposition. If so, then these effects could plausibly have been caused by a sudden local
deglaciation event, potentially achieved by a temporary sea level rise (over 400m
predicted locally), floating and breaking up low-elevation ice sheets. A very similar effect
can be noted from Grandfather Lake in Alaska (Vlag and Banerjee, 1999), once on the
edge of the Akhlun glacier system, where a sudden and dramatic shift in vegetation can
be seen in the pollen record in approximately 8.2ky BP – although the resolution of the
data is admittedly low. The proportion of Alnus suddenly increased from near zero to
about 70%, where it has remained ever since; Betula and other cold-loving taxa decreased
accordingly. Clearly this shift was due to glacial retreat; the question is whether this
retreat was assisted by the sea level rise (+190m predicted) suggested here.
Also in the far north is Lake CF8 on Baffin Island (Axford et al, 2009), where two
sharp changes in the pollen composition imply short periods of much cooler climate, the
second of which is suggested by the authors to correspond to the 8.2ky BP event. A
similar picture is obtained from Lake Lyadhej-To in northern Siberia (Andreev et al,
2005) with a temporary pollen-inferred temperature reduction of 1-1.5in approximately
8.2ky BP.
The final potentially flooded lake where pollen changes have been recorded is Lake
Shuangchi Maar on Hainan Island (Zheng et al, 2003), which displays a very abrupt
increase in pollen levels, ascribed by the authors to the onset of a much more humid
climate – which is a logical inference whatever the cause.
Of those lakes where flooding is predicted but which have been classified as showing
little or no pollen-related change, Lake Chaohu in eastern China (Wang et al, 2008)
shows a sharp but minor dip in Quercus and a corresponding rise in certain shrub taxa;
Lama Lake in Siberia (Nowaczyk et al, 2001) shows a similarly small dip in arboreal
pollen; the changes at Lake Pupuke in New Zealand (Augustinus et al, 2008) and Blood
Pond in Massachusettes (Hou et al, 2006) appear negligible; while the data from Lake
Flakkerhuk in Greenland (Bennike and Funder, 1997) and Lake Juusa in Estonia
(Punning et al, 2005) are of very low resolution and so are impossible to interpret with
confidence.
Overall the contrast between the large number of pollen-related changes evident in
lakes that are predicted to have been flooded and the lack of change above the predicted
flood level is stark, and this suggests that what have conventionally been interpreted as
regional climatic phenomena could in some cases be phenomena restricted to low-
elevation land, potentially caused by seawater-induced damage to the local ecosystem.
CHAPTER 3 GEOLOGICAL EVIDENCE
48
3.4.6 Other Parameters
Oxygen isotope ratio δ18O is the prime quantity that has been used to identify the 8.2ky
BP event since δ18O in water tends to relate principally to temperature, and several δ18O
measurements from lake sediment carbonate or cellulose have been used to support the
original data from the Greenland ice sheet. It is also true that many of these lakes are
situated above the predicted flood level – for example the Ammersee in Germany (von
Grafenstein et al, 1998). In this study a lowered value of δ18O was noted in one further
lake above the predicted flood level, namely Hawes Water in England (Marshall et al,
2007) and in four where flooding is predicted, namely Lake Rouge in Estonia (Veski et
al, 2004), Lakes Arbovatten (St. Amour, 2009) and Igelsjön (Hammarlund et al, 2003) in
Sweden and Nordan’s Pond in Newfoundland (Daley et al, 2009); the period of lowered
δ18O lasted 100-300 years. These signals should logically be related to local or regional
climate and so may not be directly indicative of a seawater inundation event.
Nevertheless it is of note that all four of the lakes studied with δ18O data that showed no
significant reduction, namely Moon Lake in North Dakota (Valero-Garcés et al, 1997),
Lakes Oikojärvi in Finland and Keitjoru in Sweden (St. Amour, 2009) and Lake
Tibetanus in Sweden (Hammarlund et al, 2002), were from above the predicted flood
level.
Another widely used proxy, affected by both temperature and salinity, is the mix of
diatoms present, established by some but not many of the studies used. It will not be
detailed here since it does not appear that significant additional information can be
gained, although in some cases the data supplements pollen analysis and can be related to
climate change. Similarly, carbon isotope ratio δ13C, which is believed to be related to
algal activity, and which was quoted by several authors, appears to offer no further
insight into the 8.2ky BP event, and so will not be presented here.
At this stage therefore the evidence points to two effects. The first is a 200-300 year
long period of climate change; the second is a very short-lived event, revealed by spikes
and sudden changes in TOC, C/N, mineralogy and pollen influx. Significantly, this
second effect is almost entirely restricted to lakes that are below the suggested sea level
rise contour.
3.4.7 The American Great Lakes
The foregoing subsections have done no more than sample the information to be gleaned
from the planet’s lakes, reliant on the vagaries of research and publication. However,
while it is reasonable that hard evidence is sometimes difficult to find in small lakes
which would recover quickly from saltwater inundation and which would not have been
significantly affected by the straining and tilting of the planet’s mantle described in
Chapter 2, this is not the case for the planet’s largest lakes. Here one would certainly
expect direct evidence. This subsection will therefore investigate the case of the
American Great Lakes, five of the largest on Earth and all below the proposed flood
level.
The basins of the American Great Lakes (Figure 3.9) were all ice-covered during and
after the last glacial maximum. Their modern history therefore only commenced in about
14ky BP when the first parts of Lake Erie became exposed, and the ice only left northern
Lake Superior in around 10ky BP. Studies have revealed that it was a complicated history
due to progressive isostatic rebound of the land as the Laurentide Ice Sheet receded. From
CHAPTER 3 GEOLOGICAL EVIDENCE
49
around 11ky BP (Odegaard et al, 2003) the main drainage for the upper Great Lakes
(Superior, Michigan and Huron) was through Lake Nipissing to the Ottawa River, and the
depressed level of the Lake Nipissing region (due to the proximity of the ice sheet) meant
that levels were generally lower (relative to the land) than those today across most of
Lakes Michigan and Huron. They fluctuated dramatically however. According to Lewis
et al (2005, 2008) there were three short-lived highstands during which levels were some
40m above those in the intervening centuries, the precise explanation for which is unclear
but which must certainly have involved the disintegration and re-forming of glacial
barriers. The last so-called ‘Mattawa’ highstand was in around 9.0ky BP, after which lake
levels in the upper Great Lakes fell, according to Lewis et al (2005, 2008) by around
60m, to the third so-called ‘Stanley’ lowstand phase. During this same period Lake Erie
became restricted to a small area at the eastern end of its basin (Holcombe et al, 2003).
Lake Ontario would also have been depressed at its western end relative to today’s
shoreline, in line with evidence from Hamilton harbour (Duthie et al, 2004). A major
reason for reduced water inflow during this phase was the fact that glacial lakes Agassiz
and Ojibway had now coalesced, allowing the water from Lake Agassiz, which had
flowed south into Lake Superior during the Mattawa highstands, to flow through Lake
Ojibway and directly into the Ottawa River. Many studies (Croley and Lewis, 2006;
Lewis et al, 2008; McCarthy and McAndrews, 2010) also suggest that some or all of the
Great Lakes then became closed basins, with insufficient inflow to overcome
evaporation, implying a much drier climate than today. This was the situation
immediately prior to the final draining of Lake Agassiz-Ojibway in around 8.2ky BP.
Figure 3.9: Great Lakes region prior to 8.2ky BP
The elastic rebound following the emptying of Lake Agassiz-Ojibway would have led
to a sudden rise in the sill between Lake Huron and the Ottawa River, by approximately
5m according to the calculations carried out here, an effect that has not been taken into
account in the modelling presented by Lewis et al (2005, 2008). It is therefore
unsurprising that forests have been discovered on the beds of Lakes Huron and Michigan
that show signs of sudden immersion. Lewis et al (2009) deduce that a sudden 7m rise
took place in Lake Erie, also expected due to the sudden isostatic rise of the Niagara sill
L. Superior
L. Michigan
L. Huron
L. Erie
L. Ontario
R. St Lawrence
R. Ottawa
R. Mississippi
Laurentide Ice Sheet
Glacial Lake Ojibway
L. Nipissing
L. Nipigon
CHAPTER 3 GEOLOGICAL EVIDENCE
50
relative to western parts of the lake. Miller et al (2000) report about 3m of sediment
around the remains of tree stumps on the bed of southern Lake Michigan that all date to
the same approximate time, 8.45-8.4ky BP according to their paper, although there is
considerable uncertainty in the correction required to 14C dates in periglacial regions.
Whatever the true date, it seems inescapable that this event occurred at the end of the
lowstand phase and that this corresponded with the isostatic changes caused by the loss of
water from Lakes Agassiz and Ojibway.
However, according to the polar shift hypothesis presented in this report, the final
draining of Lake Agassiz-Ojibway brought about much more than a sudden isostatic
change. The entire Great Lakes region would have experienced three and a half years of
seawater immersion, and this would have been preceded by a very sudden filling of all
five basins as the ocean overtopped the lowest point of the rim of each. The 3m of sudden
sediment deposition over the buried forest at the southern end of Lake Michigan,
including unusually intact – and therefore rapidly buried – clam shells and the remains of
land beetles (Miller et al, 2000), may plausibly be due to this sudden infill event. Colman
et al (1994) reported that low 18O values (-6‰ to -10‰) in Lake Michigan, typically
associated with glacial meltwater, came to a sudden end at a date constrained to lie
between 8.9ky BP and 8.03ky BP, rising suddenly to around -2‰ to -3‰, and this date
can be further constrained by the fact that Lewis et al (2005) found highly negative 18O
values in Lake Huron down to about 8.4ky BP. This change is compatible with seawater
inflow since seawater 18O lies between 0‰ and -1‰.
In Lake Superior Mae et al (2007) investigated 18O levels from ostrocode species
found in cores through sediments in deep parts of the lake and also noted highly negative
18O values (-28‰ to -15‰) during early phases. From about 8.4ky BP however no
ostrocodes were found, a fact that the authors attribute to a lack of preservation due to a
lack of sedimentation. It could of course also have been due to a change in water
chemistry. In fact Forrester et al (1994) report a similar paucity of ostrocode shells from
Lake Michigan during this same period.
However, much the most revealing geological evidence from the American Great
Lakes is the sudden appearance of black iron sulphide bands in sediments from this time,
bands that continued to form for the next few thousand years. They have been reported
from all five lakes (Odegaard et al, 2003), in each case commencing immediately above
the unconformity that brought the final Stanley lowstand to an end. The appearance of
these bands is unequivocally associated with anoxia in the bottom water of a stratified
water column (Odegaard et al, 2003; Davison and Heaney, 1978). If this observation
were restricted to a single basin then one could legitimately seek an explanation unrelated
to the polar shift event described in this report. However, for the same phenomenon to
have occurred in all five basins at a time when there was no connection between the
upper lakes and Lake Erie almost demands an event that was large enough to change the
chemistry of all lake water in the region. An influx of seawater would constitute an event
of appropriate magnitude; it would have induced stratification of the water column since
subsequent freshwater river inflow, being of lower density, would have tended to remain
on the surface. The result would have been a situation not unlike that applying in the
Black Sea today, with the difference that there would have been no continuing
replenishment of the saline bottom water. With time therefore the limited mixing that
occurred, combined with river outflow, would have reduced the salinity of the bottom
CHAPTER 3 GEOLOGICAL EVIDENCE
51
water, eventually seeing an end to anoxic conditions and the disappearance of the iron
sulphide bands. The duration of anoxic conditions would have depended on the water
retention time (191 years in Lake Superior today for example) and the degree to which
currents induced intermixing between upper and lower water masses. A few thousand
years is reasonable.
In summary therefore it is fair to state that the combination of phenomena that took
place at the end of the last Stanley lowstand phase is fully in accord with expectations
had seawater inundation occurred. Thus, allowing for the uncertainty in dating, American
Great Lakes geology is definitely supportive of the polar shift hypothesis.
3.4.8 The African Great Lakes
Lake Malawi in East Africa, at +474m and within 15of the equator, is at much too high
an elevation to have been flooded during the proposed polar shift event, and this is
confirmed by the lack of evidence from sediment cores for any significant period of
raised salinity at this time (Finney and Johnson, 1991). However the geography of Lake
Malawi means that polar shift should nevertheless have had an effect. The lake is about
560km in length, typically 50km in width, and is orientated approximately north-south. It
is bounded on all sides by steep-sided mountains, with the exception of the valley at the
southern end down which the Shire River carries excess lake water to the Zambezi.
Although not in danger of seeing any seawater ingress, it is predicted that the proposed
polar shift would have imposed a gradient on the region, effectively creating a 50m
difference in level between the northern and southern ends of the lake. During the first
phase of motion the body of water would simply have migrated north, but during the
second phase around 600km3of water would have flowed out down the Shire River,
producing an average 25m reduction in lake level. In this context, the observation by
Gasse et al (2004), who present a diatom analysis of sediments from a core taken from
the northern basin of the lake, that there was probably a short-lived regression in the 8.5-
8.2ky BP period, is therefore relevant. The date range quoted, as with all dates relying on
chemical dating techniques, should be seen as approximately correct rather than precise.
A direct comparison could legitimately be made with Lake Tanganyika, also in East
Africa, which is geologically and morphologically very similar to Lake Malawi. It too is
long (about 600km), narrow (typically 50km), orientated north-south and surrounded by
steep-sided mountains. However the key difference is that the Lukuga River that drains
the lake flows out about half way along its length rather than at one end and this means
that very little water would have escaped due to tilting of the land. No regression would
therefore be expected and no evidence of one has been found (Steinkamp, 2006).
Further north however, Lake Victoria would also have been subject to tilting and,
since the Nile flows out at the northern end, the first phase of polar shift would have
resulted in significant water outflow. The predicted reduction in lake level is about 11m.
In this context the finding by Stager et al (2003) that an abrupt change occurred at this
time leading to a stratified water column would appear to be relevant. The authors
attribute it to a sudden reduction in rainfall; but since their core site was close to the Nile
ouflow, stratification is more likely to be a sign that outflow had ceased temporarily.
CHAPTER 3 GEOLOGICAL EVIDENCE
52
3.4.9 Lake Chad
Today Lakes Victoria, Tanganyika and Malawi comprise the African ‘Great Lakes’. But
in 8.2ky BP much the largest lake on the continent, indeed the largest freshwater lake in
the world after Lake Agassiz-Ojibway had disappeared, was Lake Chad in west-central
Africa, measuring some 900km across. Although now a rather small, shallow terminus
for the Chari River with a water retention time of the order of 2-3 years, in 8.2ky BP
Lake Chad was around five times the current area of Lake Superior, today the largest
freshwater lake in the world – see Ghienne et al (2002) based on satellite images. It
stretched from Maiduguri in northeast Nigeria to the southern slopes of the Tibesti
Mountains in the central Sahara, with a total volume in the order of 10,000km3. Ghienne
et al (2002) suggest that the lake was approximately at its maximum size, outflowing into
the Benue River system via a sill at +308m, from 8.5ky BP to 5.5ky BP with the
exception of relatively short periods of regression around 8.3, 7.1 and 6.5ky BP. While
others have arrived at slightly differing dates and levels, e.g. Gumnior and Thiemeyer
(2003) who suggest a maximum at around 6.3ky BP, there is broad agreement amongst
researchers that Lake Chad was a genuinely vast body of water in the early Holocene.
With the proposed polar shift, the level of the ocean would have been insufficient to
reach Lake Chad. However, as in the case of Lakes Malawi and Victoria, the surface of
the land would have been tilted significantly with respect to the Earth’s gravitational
field. The first phase of motion would merely have moved the lake several kilometres to
the north-east, while the second phase would have tilted the land to the south-west,
temporarily flooding previously dry regions in north-eastern Nigeria and emptying a
significant proportion of the lake’s contents down the Benue River. When the tilting
ceased, Lake Chad would have been left at a level of around +280m. Although the now-
dry Bodele depression would have remained full of water, large areas of land would
suddenly have been exposed – and whichever climatic model is used it would have taken
several centuries for the water to regain the +308m sill level. This proposed water loss
matches the 8.3ky BP regression noted by Ghienne et al (2002) based on other studies.
The findings of Thiemeyer (1997) are also relevant. He investigated the dune field
complexes of north-eastern Nigeria, presenting evidence as to when they were submerged
and when they were exposed to the air, and discussing the factors affecting erosion. From
12ky BP to 8ky BP (approximately) he found that all the dune fields north-east of a
particular low ridge near the city of Maiduguri were submerged, and similarly from 7ky
BP to 5.5ky BP; however between 8ky BP and 7ky BP the picture was confused. It was
clear that dunes had been reworked due to exposure to the atmosphere, but it was not
clear exactly why since a gradually retreating lake level should have allowed sufficient
vegetation to form for erosion to be resisted. Furthermore there had been unexpected
changes to older dune fields that should never have been affected by Lake Chad. He used
the term ‘unknown dynamic’ to indicate that changes had occurred but that the
mechanism was unclear. On the other hand the tilting of the land proposed here would
mean that regions well beyond the boundaries of Lake Chad would have experienced up
to two years of inundation, leading to loss of vegetation across marginally stable dune
fields; and, as explained above, this would have immediately been followed by a
reduction in lake level which would suddenly have exposed large areas of previously
submerged land, probably for centuries. The observation of an abrupt change to the dune
fields is supported by Holmes et al (1999) who noted the sudden onset of a short period
CHAPTER 3 GEOLOGICAL EVIDENCE
53
of aeolian sand deposition in 8.2ky BP in the sediments of Lake Bal, a closed playa lake
200km west of the present Lake Chad. Lake Bal itself would almost certainly have been
flooded by Lake Chad if the proposed polar shift event actually occurred.
In fact it is difficult to explain such a significant regression in Lake Chad at this time
other than through the mechanism proposed here. Baumhauer (2004) reports no change in
the lakeland conditions pertaining just to the north-west of the 8.2ky BP Lake Chad; and
in the eastern Sahara the late 9th Millennium BP sees the onset of very humid conditions
(Haynes et al, 1989). Yet Lake Chad suffered significant loss of water for some reason.
Thus, while dating is imprecise it is certain that something occurred in the Lake Chad
region towards the end of the 9th Millennium BP, something that caused a significant and
relatively rapid regression of the lake. The proposed polar shift episode provides a logical
explanation.
3.4.10 Conclusion
It is abundantly clear based on the evidence of lakebed sediments not only that a
significant climatic shift occurred between about 8.4ky BP and 8.1ky BP but that a very
short-lived event took place in approximately 8.2ky BP that led to sudden increases in the
organic carbon content of sediments, changes in sediment mineralogy and in pollen
composition and influx rate, all plausible consequences of saltwater inundation.
Furthermore it has been shown that these short-lived changes are almost entirely found in
lakes below the sea level rise contour proposed here. The fact that these signals are
concentrated mainly in larger lakes (>1km2) is to be expected because of the longer-
lasting effects of inundation compared to small lakes with small catchments.
In the particular case of the American Great Lakes, the simultaneous onset of black
iron sulphide bands in sediments from all five lakes dating from this exact period clearly
signals anoxic bottom water, one of the known causes of which is a difference in salinity
(and therefore density) between the upper and lower parts of the water column. No
explanation for this phenomenon has yet been agreed.
Finally, it has been shown that effects took place in the major lakes of Africa that are
compatible with the calculated tilting of the land surface during polar shift and for which
no other cause is readily apparent.
In total therefore it is fair to state that the evidence from lakes and inland seas
provides considerable and in many cases well-dated support to the polar shift hypothesis.
3.5 Salt Lakes and Playas
The distinction between freshwater and saltwater lakes is slightly artificial since many
salt lakes have a complex history and have been freshwater lakes in the past, with an
outfall, but reduced rainfall has led to disconnection and the accumulation of salt. Thus
the set of lakes considered in the previous section actually included a few that are saline
today, such as the Aral Sea. However this section is directed principally at ephemeral
lakes, lakes that regularly dry out completely, or permanently saline bodies of water such
as the Dead Sea. Australia contains the greatest concentration of ephemeral salt lakes in
CHAPTER 3 GEOLOGICAL EVIDENCE
54
the world and therefore presents an excellent opportunity to look for evidence,
particularly since much of Australia lies below the projected flood level. The Sahara also
includes numerous low-lying salt playas. Other specific locations requiring investigation
are the Danakil Depression in East Africa, the Salton Sea in California and, as mentioned
above, the Dead Sea, all at elevations below current ocean level and all predicted to have
been flooded in 8.2ky BP. However, before considering each of these locations in turn,
the first subsection will address the issue of the likely effect of a short-duration (typically
5-6 years) saltwater flood on groundwater chemistry.
3.5.1 Saline Groundwater
It is acknowledged by hydrologists that there are several natural origins of saline
groundwater in addition to human-initiated salinity, namely:
a) Connate seawater – seawater that has been present within a sediment since its
deposition, which may have been millions of years ago.
b) Seawater that has accessed a layer in a coastal location by lateral flow from the
ocean.
c) Seawater that has arrived due to a past marine transgression.
d) Salts that have accumulated due to millennia of evaporation from an enclosed
basin, followed by re-dissolution by rainwater and leaching into the groundwater.
e) Salts that have been dissolved into initially fresh rainwater as it percolated
through a soluble rock layer.
In most cases it is clear whether the origin is marine or not, since the mix of chemicals in
the ocean is distinctive (primarily sodium and chloride) and generally different from
those that leach from the rocks of the Earth. However there is often no way to be sure of
the means by which oceanic water arrived, nor the timeframe involved. Clearly in the
context of the proposal being put forward in this report this is an important point. For
example, there is saline groundwater beneath large swathes of the Argentinian lowlands
(van Weert et al, 2009) and the conventional explanation is that this must be connate
since it is not generally believed that there has been any recent marine incursion in this
region – yet this cannot be proved. In fact it is very surprising that saline water should be
present so far from the ocean. This may be contrasted with the situation in the Nubian
sandstone aquifer that covers much of northeast Africa where, at Bahariya oasis in central
Egypt, the entire 1800m depth of the layer contains fresh water (Himida, 1970), despite
the view of most researchers being that much of the Nubian sandstone was deposited as a
marine sediment; its original connate water would therefore have been saline. Thus, in
Africa, the passage of about 50 million years has seen the replacement of the original
saline pore water with fresh rainwater inflow even though the climate over much of the
region has been arid for a large percentage of that time; yet this has not been the case in
South America for some reason.
It has apparently also not been the case in water-bearing strata beneath the eastern
seaboard of the United States, Florida, the Mississippi basin and the Texas lowlands (van
Weert et al, 2009). Neither has rainwater infiltration managed to flush out connate
seawater from relatively shallow strata beneath the whole Northern European Plain, nor
parts at least of the plain of Central Asia.
CHAPTER 3 GEOLOGICAL EVIDENCE
55
The point being made through these observations is that ‘connate water’ looks to be
an inadequate explanation for saline groundwater in regions distant from the sea that have
seen millions of years of rainwater inflow, when compared with other parts of the world
where initial marine pore water has been replaced. If the event described in this report did
indeed take place then it would have brought seawater to cover most of the low-lying
regions of the world for a period of up to about six years, and it is inevitable that there
would have been some intrusion into the groundwater, although the extent to which this
took place would have been critically dependent on the previous depth of the
groundwater table and the permeability of the underlying strata. In regions of heavy clay
the effect would have been negligible; in formerly-dry sandy strata the impact would
have been significant and long-lasting.
However, because of the large uncertainty in modelling the likely penetration of salts,
this point will not be amplified further here.
3.5.2 Australian Salt Playas
It is conventionally expected that the salts found on the bed and in the ground below an
ephemeral salt lake have been washed in by the rivers that feed that lake. They will
therefore be salts that have leached out of the ground and they will generally bear no
close relationship with the combination of salts found in seawater. This, for example, is
the case in the playas of western China (e.g. Wu et al, 1986) and North America (Last,
1990). It is therefore of surprise to many researchers that much of the salt in the majority
of Australian salt lakes is unequivocally of marine origin (e.g. De Deckker, 1983).
Known exceptions include small crater lakes in upland western Victoria and Lake
Buchanan in Queensland (Chivas et al, 1986).
Chivas et al (1991), who studied the sulphur isotope parameter 34S, produced a set of
contours covering Western Australia that indicated 100% sea salt origin for gypsum
sulphates near the coasts, estimating that this reduces to about 55% at Lake Way, about
500km inland. They even suggested that the Lake Amadeus basin in the centre of the
continent may have received up to two thirds of its sulphur from oceanic sources. A
marine origin for these salts presents a conundrum since it is not generally considered
that there has been any marine transgression for about 60 million years. Chivas et al
(1991), in common with numerous other researchers, therefore proposed that these salts
arrived as airborne sea salt. Thus the contours of 34S across Western Australia would be
expected to correlate with distance from the ocean, and they do – approximately.
However, when the measured 34S values are plotted against the predicted maximum
extent of flooding – see Figure 3.10 – a clear pattern can be seen. In all locations where
flooding is predicted, values are in the 19-23‰ range, compared to 21‰ for seawater;
those where no flooding is predicted lie in the 13-18‰ range. Thus Lake Seabrooke is at
20‰ while Lake Austin, a similar distance from the ocean, is at 17‰; Lake Percival is at
20‰ while Lake Waukarlykarly, much nearer to the ocean but on higher ground, is at
14‰. While it does indeed appear likely that inland lakes have received large quantities
of airborne sea salt and that this forms a proportion of the salts currently present, it is the
contention here that inundation in 8.2ky BP was the principal agent causing raised 34S
values in low-lying regions.
Chivas et al (1991) admit that their findings in relation to South Australia are less
certain because of the complexity of the surface chemistry and the raised likelihood of
CHAPTER 3 GEOLOGICAL EVIDENCE
56
additional sulphur ingress from bedrock sources. However Vengosh et al (1991) carried
out a similar study based on the boron isotope parameter 11B and came to the conclusion
that the ultimate origin of the boron present in Australian playa deposits in South
Australia as well as the southern half of Western Australia is marine. They also proposed
that the source was airborne sea salt.
Figure 3.10: Locations of Australian salt playas, 34S values (in ‰) for Western
Australia (Chivas et al, 1991), predicted flooding superposed
This explanation however is logically flawed in the case of the playas of the Great
Artesian Basin. Several studies of this region (Johns and Ludbrook, 1963; Johns, 1968;
Callen, 1977) have revealed that, though current salts are sodium chloride-rich and of
marine origin, underlying strata contain carbonate-rich sediments and therefore cannot
have been. The sodium chloride-rich salts are assigned to the Pleistocene and Holocene
(i.e. within the last 2My); the carbonate-rich deposits are older. No explanation has been
offered as to why the chemistry of these lakes should have changed two million years ago
in an environment that had been essentially unchanged for many tens of millions of years.
Another slightly surprising finding is that there is so little salt in many of the
Australian playas, although it has to be admitted that it is extremely difficult to predict
the balance between surface salt deposits and salt dissolved in groundwater. Nevertheless
if a comparison is made with the well known Lake Bonneville in Utah the contrast is
stark. In Lake Bonneville there has generally been about 2m of salt in historic times
(Mason and Kipp, 1998), reducing in recent years due to local brine extraction, and the
overall volume is compatible with the 11ky of water inflow that the lake has experienced
Eyre
Frome
Torrens
Gairdner
Mackay
Carnegie
Percival
Amadeus
Way
Buchanan
Great Artesian
Basin
13
15
14
17
17
18
19
19
19
20
20
20
21
23
14
15
16
15
14
16
16
20
Waukarlykarly
16
22
23
20
Seabrooke
Austin
CHAPTER 3 GEOLOGICAL EVIDENCE
57
since it last flowed out into the Columbia River system. In the great Australian lakes by
contrast the thickness is commonly of the order of a few centimetres. Kotwicki (2008)
has calculated that the current salt content in Lake Eyre equates to about 5.5ky of run-off
at current rates – contrasting with an age of many millions of years.
These observations are however fully explicable if it is accepted that there has been a
relatively recent inundation by the sea. This would have had the effect of dissolving most
of the accumulated surface salts present and leaving behind only the marine salts
contained in trapped seawater, much of it in the Great Artesian Basin. This would explain
the relatively low quantities seen today.
The observation that the salts in many of the playas have only been of marine origin
for two million years is also compatible with the expectation that catastrophic polar shift
has accompanied many or all of the terminal stages of past ice ages. Prior to two million
years ago there were no ice ages and so no opportunity for release of meltwater on the
scale of Lake Agassiz and therefore no massive polar shift episodes; thus it would be
expected that the salts accumulating in the playas could only be derived by leaching out
of the rocks. Carbonate-rich salts occur all over the world where inflowing rivers traverse
limestone deposits. It therefore seems likely that it was only after an episode of polar
shift, following an early Pleistocene glaciation phase, that sodium chloride of marine
origin came to dominate the South Australian playas.
3.5.3 Saharan Oases
The freshwater reservoir in the Nubian sandstone aquifer has already been mentioned, a
reservoir that extends many hundreds of metres below sea level and yet which is, at
Bahariya oasis and for hundreds of kilometres to the south, entirely fresh. The importance
of this resource to the economy of the Sahara is immense, and there has been much
investigation and modelling over the years to determine the source of the water and
whether it is being continuously replenished by rainwater falling far to the south. The
near consensus (e.g. Heinl and Brinkmann, 1989) nowadays is that replenishment is
minimal and that the source was almost entirely vertical infiltration of rain and lake water
during the Saharan pluvial period between about 10ky BP and 6ky BP, refilling a
reservoir that had been created during previous pluvial periods.
Yet between Bahariya oasis and Moghra oasis, about 150km further north, the
situation changes significantly; the upper 1000m of water in the Nubian sandstone layer
beneath Moghra is relatively fresh, but below that the waters are very saline. The
conventional explanation (e.g. Aref et al, 2002) is that a saltwater wedge has intruded into
the aquifer from the Mediterranean Sea, just 80km north of Moghra. The problem with
this explanation is that 150km south-west of Moghra, at the lowest point of the Qattara
Depression (-133m), and 150km from the sea – see Figure 3.11, the level of saline water
is actually some 900m higher, such that it emerges at the surface. Yet the Mediterranean
is still held to be responsible because no marine transgression filling the Qattara
Depression is known to have occurred.
Logically however, if the water within the Nubian aquifer is principally the product of
vertical infiltration from freshwater lakes during the pluvial period, with little large-scale
horizontal water movement, then the argument that extensive horizontal flow of
Mediterranean water has taken place appears questionable. An explanation which is more
in line with the evidence is that vertical infiltration from a saltwater lake has occurred.
CHAPTER 3 GEOLOGICAL EVIDENCE
58
Salts in the Qattara Depression are concentrated at the south-western end, the lowest part
and therefore the area that would have seen the final evaporation of a hypersaline lake in
the event that the whole depression had ever been filled with seawater. The fact that the
groundwater remains saline at the surface in that area is then explicable since every
rainfall event dissolves large quantities of surface evaporite deposit and reinforces the
salinity of the groundwater. Moghra on the other hand is around 100m higher in
elevation, and so could never have experienced hypersaline conditions even if the
depression had once been filled by seawater. Saline saturation and precipitation of salts
could not have occurred until the water level had fallen significantly lower. Thus, the
marine water that would have penetrated the groundwater at Moghra following seawater
inundation would since have been depressed by thousands of years of subsequent
freshwater inflow from the Nubian sandstone aquifer.
In summary, the state of the groundwater at either end of the depression exactly
matches the logical consequences of a seawater infill episode. In contrast, if the
Mediterranean really is the source of the saline groundwater then it is difficult to see a
reason for the level being so much higher at the end of the depression furthest from the
sea.
Figure 3.11: Saharan Oases region, predicted flooding superposed
The logic of this position is enhanced when the geological history of the Qattara
Depression is examined in more detail. As is the case in many low-lying playas, wind-
induced erosion exceeds the rate of airborne or waterborne deposition at the south-
western end of the depression. This means that the depression is becoming ever deeper.
The detailed mechanism of erosion is that halite crystallisation within the rock strata
forming the floor of the depression is fragmenting the rock. While the halite remains in a
solid state it acts as a binder and the result is a hard agglomerate, resistant to erosion; but
when the halite is dissolved by rainfall the fragmented grains are released as fine sand
Mediterranean Sea
Qattara
Depression
Moghra
Bahariya
Farafra
Dakhla Kharga
Siwa
Red
Sea
Al Khufra
R. Nile
Libyan
Plateau
Dead
Sea
CHAPTER 3 GEOLOGICAL EVIDENCE
59
and silt particles and, when the playa dries, these are then readily transported by the wind
(Aref et al, 2002). Interestingly, Aref et al (2002) find that this process has been in
operation for about two million years, since the start of the Pleistocene Era, the period
during which the Earth has been subjected to repeated ice ages and interglacials. They
also found residual terraces of intact halite-bound agglomerate at three higher levels at
the south-western end of the depression, at approximately -100m, -75m and -25m,
suggesting there may have been at least four major inundation-evaporation cycles during
the Pleistocene.
In the context of the proposal being made in this report, the evidence from the Qattara
Depression is important. Polar shift and consequent flooding explains the high salinity of
groundwater beneath the depression and the marked difference between the two ends.
The suggestion that similar occurrences have probably taken place during the terminal
phases of several previous glacial cycles is supported by the finding of three higher
agglomerate terraces, and the fact that the process appears to have begun at the start of
the Pleistocene is also explained.
Importantly also, the fact that Bahariya oasis and oases to the south (Farafra, Dakhla,
Kharga) all overlie an entirely freshwater reservoir means that no seawater incursion can
have occurred in those regions. Given the land elevation between these oases and the
Qattara Depression, this places an upper limit on sea level rise in Egypt of about +200m;
the proposed polar shift would have induced a rise of about +190m at the sill between the
Qattara Depression and Bahariya Oasis. It is also of note that, with this magnitude of
shift, none of the major Libyan oases of the Fezzan would have been flooded; the fact
that these contain freshwater is therefore compatible with this proposal.
3.5.4 The Tunisian Chotts
The Chott-el-Djerid complex is a network of playas in central Tunisia and extending
westward into Algeria. It is last thought to have been connected to the Mediterranean Sea
in about 90ky BP, based on dating of mollusc remains (Causse et al, 1989). It then
became isolated due to a combination of falling sea levels and land uplift, and this
resulted in playa formation. Swezey et al (1999) have charted evidence for seven
subsequent depositional phases, the first of which was dated by thermo-luminescence to
86ky BP, but the sediment from which had then steadily eroded over most of the playa
region to reveal underlying Miocene or early Pleistocene strata. Net deposition
recommenced in around 13ky BP with the onset of moister conditions and continued until
5.5ky BP. Between 10ky BP and 7.5ky BP the playa became a permanent lake, re-
dissolving the marine salts that had been deposited during earlier connection to the
Mediterranean and forming a highly saline body of water. Thus, if there had been any
further marine incursion in 8.2ky BP this would have done nothing more than effect a
change in the total salt load carried by the water, although whether it would have resulted
in a net gain or loss is impossible to determine. However, a possible indication that there
may have been an influx of seawater postdating the disconnection from the
Mediterranean in 90ky BP is that the mix of salts currently present, as revealed during
flooding and dissolution in 1990 (Bryant et al, 1994), is still extremely close to that of the
ocean, whereas one might have expected that inflowing waters from the North-West
Sahara Aquifer System and from the rivers that fed the enclosed Chott-el-Djerid lake
from around 10ky BP would have brought their own distinctive chemical signature. This
CHAPTER 3 GEOLOGICAL EVIDENCE
60
therefore invites the suggestion that a recent marine incursion may indeed have taken
place.
3.5.5 The Danakil Depression
The Danakil Depression, straddling the Eritrea-Ethiopia border, is known to have been a
bay of the Red Sea until about 30ky BP, betrayed by the existence of ancient corals and
molluscs (Bonatti et al, 1971). It then became progressively severed from the Red Sea by
a build-up of basaltic magma across the mouth of the bay, resulting in greatly increased
water salinity; in effect it became an evaporation lagoon. Finally, following complete
severance from the sea a few thousand years later, it dried out completely leaving the
fresh salt deposits seen on the floor of the Danakil Depression today. If there had been a
subsequent infilling of the basin in 8.2ky BP this would simply have added a further
halite layer to an already massive deposit, and there would therefore be no direct sign that
this infill event had occurred. It is certain that the most recent salt deposition (other than
minor reworking) predates what is known as the Afrera Formation, comprising soft
limestone, clay and gypsum and dated to 5.8ky BP (Bannert et al, 1970) or 7-10ky BP
(Waltham, 2005), but it is impossible to state from direct evidence alone by how much it
predates it. However the lack of any dated overlying strata earlier than the Afrera
Formation suggests that the uppermost salt layer may not have predated it by as much as
the 20ky that had by then elapsed since disconnection from the Red Sea. Inundation in
8.2ky BP is therefore plausible.
3.5.6 The Salton Sea
The Salton Sea in southern California is an enigmatic body of water. In its current form it
dates from 1905 when an attempt was made to irrigate the then-dry Imperial Valley by
diverting fresh water from the Colorado River. The attempt failed catastrophically and a
large part of the basin was flooded. Within about 10 years of the 1905 infill event the
salinity of the Salton Sea had risen to above that of the ocean (Cohen and Hyun, 2006)
which, since the inflowing Colorado River water was essentially fresh, implies that
evaporite salts had been dissolved into the new lake. The composition of salts in the
Salton Sea (Cohen and Hyun, 2006) compared to those of the ocean and the Colorado
River (Gloss et al, 1981) suggests that the source is approximately 70% marine and 30%
from the Colorado (see Appendix C). The issue therefore is the origin of the marine
component.
Prior to about AD1300 it is known (e.g. Waters, 1983) that the Salton basin was filled
or part-filled by a large fresh or brackish water body known to researchers as Lake
Cahuilla. It was fed by part of the flow from the Colorado River and it underwent phases
of ‘overflow’ during which the water level reached +12m, the level of the sill at the
south-eastern end of the basin, and phases of reduced inflow during which the level
dropped below the sill. In fact Waters (1983) identified four overflow phases within the
last two thousand years. Li et al (2008a) studied the strontium isotope ratio 87Sr/86Sr from
the modern Salton Sea, from the Colorado River and from ancient Lake Cahuilla tufa
carbonates, and their finding was that all three matched – and that the ratios were
considerably higher than that found in the ocean. The implication is that Lake Cahuilla
formed part of the Colorado freshwater system and this is supported by the discovery of
freshwater molluscs in Lake Cahuilla sediments (Waters, 1983). Furthermore Li et al
CHAPTER 3 GEOLOGICAL EVIDENCE
61
(2008a) found that the carbonate tufas around the Lake Cahuilla shoreline had been
growing continuously between 20.5ky BP and AD1300. Since they had sampled from an
elevation of -24m, about 45m above the modern Salton Sea surface, and since there was
no discernible hiatus in carbonate accumulation, they deduced that Lake Cahuilla had
fallen no lower than -24m during its entire 19.8ky life.
Superficially it is difficult to see where the marine salts that now form a large
percentage of those in the Salton Sea could have come from if Lake Cahuilla was a
freshwater lake. They would have to have been precipitated from the lake as it dried
following the loss of Colorado inflow water in about AD1300. Since Lake Cahuilla, when
full, was some 25 times the volume of the modern Salton Sea, the salinity prior to draw-
down could in theory have been as little as 2‰ – if it is assumed that all the evaporite
materials from Lake Cahuilla have now been re-dissolved into the Salton Sea; however it
is more likely that only a proportion was actually available for re-dissolution after 600
years of exposure to a sub-aerial environment, in which case the salinity of Lake Cahuilla
prior to draw-down was probably rather higher than 2‰, making the water slightly
brackish, an unexpected condition for a lake that is essentially part of a freshwater river
system.
However, this unexpectedly high salinity is made possible by the topography of the
basin. The Colorado (or a proportion of it) would have flowed in at the southern end of
the lake and the outflow (during overflow phases) would have been located fairly close to
the inflow. The result would have been that the source water for the outflow was
substantially made up of near-fresh Colorado inflow water. Furthermore Li et al (2008b),
studying 18O values from tufa carbonates, deduced that the majority of inflowing water
must have been lost to evaporation, leaving a much reduced outflow volume even during
overflow phases. A salinity gradient would have been induced: near fresh at the southern
end of the lake; more saline at the northern end.
This situation can be modelled fairly simply; however a key unknown is the start
condition. The growth of carbonate tufas commenced in 20.5ky BP, implying that this
was the earliest phase of Lake Cahuilla, caused by a change in the course of the lower
Colorado River. At that date however the ocean had retreated to around -100m and the
Salton basin, once the northern end of the Gulf of California, would have dried out
entirely. It is therefore reasonable to propose that substantial marine evaporite deposits
had accumulated during the final oceanic phase in the basin. When Lake Cahuilla first
formed it would have re-dissolved a proportion of these evaporites, creating a saline lake.
Modelling (see Appendix C) then requires a series of assumptions. The salinity of the
Colorado has been taken to be 0.065‰ (Gloss et al, 1981). The outflow volume is
assumed to average 10% of the inflow (1km3/year against 10km3/year) based on the 18O
values reported by Li et al (2008b). The total volume of the lake is taken to be 200km3.
The end point is also known approximately, namely that the lake should have reached a
salinity of a little more than 2‰ by AD1300 and that the salts should still be about 70%
marine by that time. Adjusting the variables to match these conditions, if for example the
initial salinity in 20.5ky BP was 70‰, i.e. twice that of the ocean, then it is necessary to
assume that 96% of the lake outflow was derived directly from the Colorado inflow
rather than other lake water. This produces a salinity of 2.1‰ by AD1300 at which time
the oceanic salt component would have reduced to 70%, as required.
CHAPTER 3 GEOLOGICAL EVIDENCE
62
Whilst this set of conditions is not impossible, it may be noted that if a seawater
ingress event had taken place in 8.2ky BP this would allow a more relaxed set of
assumptions to be made while achieving the same condition in AD1300, namely a
reduction in the outflow percentage derived directly from the Colorado River, from 96%
to 91.5%. Clearly neither case can be proved; equally clearly the seawater ingress
hypothesis is compatible with the evidence.
3.5.7 The Dead Sea
The Dead Sea is in effect a large-scale experiment in the natural formation of a
concentrated brine. It is a chloride-rich solution with a significant bromide content but
very little residual sulphate or carbonate. The greatest concentration of positive ions is
that of magnesium, followed by sodium and calcium. The principal evaporite minerals
found in deposits from the last few hundred thousand years are aragonite (CaCO3) and
gypsum (CaSO4.2H2O), their formation being dependent on the quantity of carbonate and
sulphate being brought into the Dead Sea via the river Jordan and other sources, together
with a certain amount of halite (NaCl).
Many researchers have investigated the history of the Dead Sea brine (e.g. Neev and
Emery, 1967; Klein-BenDavid et al, 2004; Katz and Starinsky, 2009), basing conclusions
on the types of evaporite deposit present as well as evidence of past water levels, general
geological history and present brine chemistry. There is now near unanimity that the
Dead Sea basin underwent a long period during which it was connected to the
Mediterranean Sea via what is now the Yizreel valley in northern Israel, and that for
about a million years (Klein-BenDavid et al, 2004) it was acting as an evaporation
lagoon. Sea water flowed into the basin; evaporation took place; a much reduced flow of
more concentrated saline solution returned as a bottom current back into the
Mediterranean. For much of this period the connecting channel was shallow enough for
the concentration of salts within the lagoon (known as the Sedom lagoon) to have risen
sufficiently for halite precipitation to have occurred. This situation can readily be
modelled, and calculations carried out for this report show that the known 2000m
evaporite deposit from the Sedom lagoon phase (Klein-BenDavid et al, 2004; Katz and
Starinsky, 2009) could certainly have formed within a million years. The principal
mineral present within the Sedom deposit is halite, with a certain amount of gypsum,
anhydrite and dolomite (Klein-BenDavid et al, 2004), and this is exactly what would be
expected from seawater under these conditions.
This interpretation of the evidence for the Sedom period is widely accepted and it is
estimated that the connection with the ocean was severed in about 700ky BP (Katz and
Starinsky, 2009) as the land was uplifted. However it is much less clear what happened
next. Klein-BenDavid et al (2004) have presented a compelling study of a large number
of saline groundwater samples from the Dead Sea basin as well as water from the Dead
Sea itself and Lake Kinneret. They characterised the brines in terms of two molar ratios,
namely sodium/chloride (Na/Cl) and bromide/chloride (Br/Cl), considering these
elements to be relatively reliable indicators since they would be unlikely to react
chemically with the surrounding rocks. Their key finding was that all the brines from the
Dead Sea region fell on a single line when these two isotope ratios were plotted – see
Figure 3.12 – and that this strongly implied a mixing of two end-member brines
(designated Brine A and Brine B) in varying proportions. Brine B could only have been
CHAPTER 3 GEOLOGICAL EVIDENCE
63
produced by evaporation of seawater beyond bischofite saturation to the point where the
Br/Cl molar ratio reached 0.0155. Logically this would have occurred following final
disconnection of the Dead Sea basin from the ocean. For example, if connection to the
ocean had gradually reduced over a 3000 year period then the Br/Cl ratio would have
risen from about 0.0023, the typical Sedom lagoon value according to the model used in
this study, to 0.004. If this is assumed to be the point of disconnection, the water volume
would then need to have contracted to just over a quarter of its previous value in order to
raise the Br/Cl ratio to 0.0155 as a consequence of halite precipitation, producing Brine
B. A contraction of this order is reasonable considering the topography of the Dead Sea
basin.
Figure 3.12: Dead Sea basin brines (after Klein-BenDavid et al, 2004)
The other end-member brine proposed by Klein-BenDavid et al (2004), Brine A, is
less certain. They proposed that it is actually a brine that had seeped into the ground for
many kilometres during the Sedom phase. Over the course of time, they argued, Brine A
had seeped out and mixed with the Brine B liquid that had formed following
disconnection from the ocean. However the mechanism by which the brines mixed is far
from obvious; nor is it immediately clear how there could have been sufficient volume of
Brine A within the surrounding rocks to effect the necessary change to the Dead Sea
water itself, particularly since the entire basin floor and walls should in theory have
become sealed by the formation of halite. The presumption is that Brine A water seeped
out at elevations above the surface level of the lake, finding its way into the streams
feeding the lake while some of Brine B seeped into the rock below the lake surface level,
mixing with the Brine A liquid already present. The obvious alternative, shown in Figure
3.12, is that Brine A was actually seawater.
If Brine B formed as proposed following disconnection of the lake from the ocean in
700ky BP it would have been composed almost entirely of magnesium, chloride and
sulphate. Since that time, changes in brine composition will have taken place due to a
combination of river and spring inflow, seepage outflow, ion exchange with surrounding
rocks, evaporation and precipitation. This has been modelled (see Appendix D), taking
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0 0.2 0.4 0.6 0.8 1
Na/Cl
Br/Cl
Predicted brine evolution path
Dead Sea
Lake Kinneret
Groundwater samples
Ocean water
Brine B
Brine A
CHAPTER 3 GEOLOGICAL EVIDENCE
64
the chemical composition of inflow water to be that of the River Jordan in the early 20th
Century (Irwin, 1923), and the modern Dead Sea brine composition (largely magnesium,
sodium and chloride with some potassium, calcium and bromide) can be successfully
obtained. Unfortunately however the predicted evaporite deposits do not then match those
observed. Instead of being 60% aragonite, 28% gypsum and 12% halite (Katz and
Starinsky, 2009) – ignoring the clastic fraction – the prediction is for 34% aragonite, 16%
gypsum and 47% halite, with traces of bischofite. The predicted halite content is much
too high. The only way to match the actual evaporite proportions found would be to
suggest a much increased groundwater seepage rate, but the problem is that this would
produce a significantly incorrect brine composition for the modern Dead Sea, with over
twice the actual sodium content, half the actual magnesium content and much too low a
Br/Cl ratio.
The principal reason why the model fails to predict the correct sodium chloride
(halite) content in the system is the assumption that the recent chemistry of the river
Jordan – which is a high-salinity river – applied throughout 700ky of history. Yet there is
no reason to expect average past salt inflow quantities to have been lower than in the
early 20th Century. If the salts that gained access to the rivers were emerging from ‘Brine
A’ reservoirs that filled surrounding strata during the Sedom lagoon phase, one could
only expect the concentration to decrease with time as these reservoirs became depleted,
replaced by fresh water from rainfall. It is true that during highstands such as that of Lake
Lisan in about 26-20ky BP (Bartov et al, 2002) there may have been some brine re-entry,
but this would have been at elevations not exceeding that of Lake Kinneret today. The
upper Jordan, which has (or had until recently) a chemistry and concentration only
slightly different from that of the lower Jordan (Irwin, 1923), would not have been
affected.
In order to match the composition of the Dead Sea brine as it is today without
predicting greatly excessive deposits of halite, it is necessary to assume a much reduced
salt inflow from the Jordan in the past, which means that it is then necessary to postulate
a mechanism by which the salt content of the springs that feed the river could have been
replenished in relatively recent (but pre-industrial) times. Seawater inundation is clearly a
theoretically possible mechanism. If the sea were ever to have re-flooded the basin it
would have left behind a saline lake which would have taken several centuries to return to
its former size by evaporation. The groundwater pressures this would have generated
would inevitably have driven saline water deep into any porous rock strata in the region.
Then, when the water level fell and normal river flow resumed, the saline groundwater
would have re-emerged and the result would be a high salt content in the Jordan and other
rivers, gradually reducing over time due to the action of rainwater.
In this context it is of note that the period since 700ky BP can be divided into four
phases, separated in the sediment record by three thick (typically 6m) halite bands. These
bands are dated approximately (Katz and Starinsky, 2009) to 420, 200 and 10ky BP; the
last is particularly well known and has been variously dated by other researchers to
between 11 and 8ky BP (Yechieli et al, 1993; Migowski et al, 1998 and 2004) or 8-7ky
BP (Ben-Avraham et al, 1999). Conventionally, these bands are assumed to be due to
rapid lake level draw-down events.
The hypothesis put forward here is that the three halite bands each signal inundation
by seawater and subsequent draw-down by evaporation, the last one being in 8.2ky BP.
CHAPTER 3 GEOLOGICAL EVIDENCE
65
With this assumption, the whole 700ky phase has been re-modelled (see Appendix D for
details). Ratios between the volume of water held within the basin immediately following
inundation and the minimum volume after subsequent evaporation were assumed to be
6:1, 7:1 and 8:1 at 420, 200 and 8.2ky BP respectively, reflecting estimated changes in
basin topography. Seepage of brine into the surrounding rocks was taken to be negligible
due to the very low permeability of the evaporite beds. To optimise the match between
prediction and observation in terms of the thicknesses of different minerals in precipitated
strata and also current Dead Sea brine composition, a logarithmic reduction in river salt
content (above a background value) following inundation is proposed, with a half-life of
20ky. Figure 3.13 shows that the current Dead Sea brine is still matched reasonably well
with these assumptions, at least in terms of Br/Cl and Na/Cl ratios. In fact the magnesium
and potassium contents are predicted to be rather higher (47g/l instead of 44g/l; 13g/l
instead of 7g/l) than is currently the case, but magnesium and potassium both have the
potential for reaction with surrounding rocks and therefore loss to the system. The
sodium and chloride contents are fairly accurately modelled. Also, importantly, the
predicted evaporite composition (excluding the three concentrated halite bands) becomes
64% aragonite, 8% gypsum, 22% halite and 5% bischofite – closer to the 60% aragonite,
28% gypsum and 12% halite reported by Katz and Starinsky (2009) than was possible
without any seawater ingress. The low predicted gypsum content implies that the
background river salt composition should have contained more sulphate than that
deduced in Appendix D, which is quite possible given the uncertainty in the assumptions
made. This would also have led to a decrease in the proportion of halite.
Figure 3.13: Predicted Dead Sea brine evolution – including seawater inundations
In summary therefore there is excellent physical reason to suggest that the Dead Sea
was indeed refilled with seawater on three separate occasions, marked by three thick
halite bands within the Dead Sea basin evaporite sequence. The dates, 420, 200 and 8.2ky
BP (approx), all fall a few millennia after the ends of glacial periods (Petit et al, 1999)
and so are logical dates for sudden escapes of water from glacial meltwater lakes; the
three seawater infill events could therefore all have been caused by similar phenomena.
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0.02
0 0.2 0.4 0.6 0.8 1
Na/Cl
Br/Cl
Predicted brine evolution path
Dead Sea
Lake Kinneret
Groundwater samples
Ocean water
Predicted Dead
Sea brine
CHAPTER 3 GEOLOGICAL EVIDENCE
66
Furthermore an early Holocene highstand in the Dead Sea basin is also supported by
other researchers (e.g. Frumkin et al, 2001; Goodfriend et al, 1986), and Stein et al (2004)
note a sudden change in the 14C reservoir age that occurred at that time, signalling a
mixing event that ended thousands of years of stratified water conditions.
3.5.8 Conclusion
It is believed that the evidence from Dead Sea sediments is extremely powerful. It does
more than merely fit in with the proposed seawater inundation in 8.2ky BP; it is hard to
explain in any more conventional way. Furthermore it provides solid evidence of two
earlier inundation episodes, each at the close of an earlier ice age. This is directly
supported by physical evidence in the Qattara Depression in Egypt in which three terraces
of evaporite suggest three earlier filling and evaporation cycles. Additionally, the current
dessication process on the floor of the Qattara Depression was found to have commenced
at the start of the Pleistocene Era, about two million years ago, coincident with the onset
of glaciation-deglaciation cycles. Coincidentally, the beginning of the Pleistocene was
also the approximate time at which the groundwater chemistry in the Great Artesian
Basin in Australia suddenly changed from a carbonate-dominated salt combination to one
of acknowledged marine origin. Taken in conjunction with evidence from the Danakil
Depression, the Chott-el-Djerid and the Salton Sea, these findings provide excellent
reason to believe not only that a massive incursion of seawater took place in many parts
of the world in 8.2ky BP but that at least three similar incursions have occurred over the
last two million years.
In passing it should also be reported that bodies of water that are not predicted to have
been reached by this flood event show no signs of anything unusual. Lake Balkhash in
western Kazakhstan is one such and, though partially saline, the mix of salts is distinctive
(sulphate-rich) and quite different from that of the oceans, clearly resulting from tens of
thousands of years of river input (Kawabata et al, 1999). The Great Salt Lake in Utah
(Hahl and Langford, 1964) is another highly saline lake, at an elevation well above the
predicted flood level, and, though the salts are predominantly sodium and chloride they
are compatible with local river inflow; similarly the Tüz Gul in Turkey (Camur and
Mutlu, 1996), Lake Urmia in Iran (Esmaeili Dahesht et al, 2010) and several other
relatively large salt lakes in highland regions in many parts of the world.
Thus it may fairly be said that a study of the Earth’s saltwater bodies and residual
playas, together with associated groundwater systems gives significant support to the
hypothesis that a widespread saltwater flood episode occurred in about 8.2ky BP.
However, one particular body of salt water has so far been omitted from this discussion.
It is on a quite different scale from those considered so far and it is a body of water that
has already been linked by several researchers with a massive saltwater flood. It is the
Black Sea.
CHAPTER 3 GEOLOGICAL EVIDENCE
67
3.6 The Black Sea Basin
Since Ryan et al (1997)’s proposal of a ‘Black Sea Flood’ the history of this body of
water has been a controversial topic and there is still considerable disagreement between
experts as to the exact sequence of events. The works of Ryan et al (1997, 2003) attracted
considerable interest because of the suggested links between a proposed catastrophic
Black Sea flood and the story of Noah. They drew attention to the evidence in the Black
Sea for shoreline features over 100m below current sea levels that appeared to have been
submerged rapidly since these features had not been destroyed by subsequent wave
action. These observations have since been supported by other researchers, for example
Lericolais et al (2009) and Gökaşan et al (2005, 2010). Ryan et al (1997) also noted the
remarkable fact that the lowermost level of organic-rich deposit known as sapropel
containing marine organic matter of Mediterranean origin was deposited at a consistent
date (c. 7.6ky BP) at all levels sampled, between 200m and 2200m depths. These points
were seen as evidence for a sudden and catastrophic inflow of Mediterranean water
through the Bosphorus Strait, transforming what is acknowledged previously to have
been an isolated fresh or brackish water lake into the marine water mass that exists today.
However the explanation offered by Ryan et al (1997) has failed to satisfy most
marine geologists. There would appear to be a near consensus that the level of the Black
Sea, while it had certainly been much lower in earlier millennia, was already high enough
to flow out through the Bosphorus by 7.6ky BP (Aksu et al, 2002; Görür et al, 2001;
Giosan et al, 2009). The sudden inflow event has therefore tended to be pushed back in
time (Ryan et al, 2003).
Since the entire issue is highly topical as well as being controversial, involving
diametrically conflicting views, the next subsection is included as a discussion of the
likely history of the Black Sea prior to 8.2ky BP, attempting to synthesise the relevant
evidence. The question of what may have happened in 8.2ky BP will be addressed
subsequently.
Figure 3.14: Model used in Black Sea and Caspian Sea simulation (Thom, 2010b)
3.6.1 Modelling Black Sea History
A simple model has been designed, representing the entire Black Sea-Caspian Sea system
over a period of several thousand years, as shown in Figure 3.14. The model takes
account of the effects of river inflow (including flow through the Manych channel from
the Caspian Sea), evaporation and sea level rise. Thom (2010b) gives details.
Black
Sea Caspian
Sea
Ocean ManychBosphorus
Rivers Rivers
Evaporation Evaporation
CHAPTER 3 GEOLOGICAL EVIDENCE
68
Ocean level changes have been taken from the Fairbanks (1989) curve and are
assumed to apply to the Mediterranean and the Sea of Marmara. Both the Black Sea and
the Caspian Sea are described by simple equations relating water volume to surface level.
Exact river inflow and evaporation rates are of course unknown; they were therefore
considered variables that could be adjusted within reasonable constraints of known
climatic history in order to generate outputs in accordance with observed evidence. In
fact evaporation rate (more correctly the difference between evaporation and
precipitation) was linked directly to temperature deduced from the Greenland ice core
record (Alley et al, 1993). A salt content of 0.1‰ was assumed for all rivers, a figure
only found today in rivers unaffected by industrial and agricultural run-off. By way of
comparison the salt content of the modern River Volga is around 0.17‰ (Clauer et al,
2000). Detailed assumptions are presented in Thom (2010b). Figure 3.15 shows the
results of the simulation, omitting the effect of the event described in this report.
The predictions would appear to match the available evidence reasonably well prior
to 8.2ky BP. Transgression of the Caspian Sea is predicted in 16-12ky BP, matching
Svitoch (2009), coinciding with a transgression of the Black Sea, in line with the opinions
of several researchers (Balabanov and Izmailov, 1988; Antonova and Svitoch, 2008).
Caspian Sea salinity is predicted to have reached a peak of nearly 13.5‰ before reducing
steadily to 8.5‰ at the end of the outflow period in 12.3ky BP, in agreement with Svitoch
(2009). At the same time Black Sea salinity is predicted to have decreased steadily to
about 4‰, in line with Antonova and Svitoch (2008).
According to the simulation, the early Holocene then brought regression to both seas.
In the Caspian the calculated minimum level is -46m, close to the -50m level suggested
by Svitoch (2008); in the Black Sea the minimum level is -163m, in line with the
evidence for dune systems (Lericolais et al, 2009), erosion surfaces, dessication cracks
and root remains (Ryan et al, 2003) to depths of -150m. From 10.55ky BP a 200-year
period of northward flow through the Bosphorus is predicted, although not the massive
Black Sea flood suggested by Ryan et al (1997). By 10.35ky BP the Black Sea is shown
to have reached the level of the world’s oceans, agreeing with Giosan et al (2009)’s
Danube delta evidence, and the Bosphorus then became a large outflowing river carrying
around 120km3/year, supporting the view of Aksu et al (2002) and others. The sudden
change in 87Sr/86Sr ratio in the Black Sea to normal ocean levels in 9.4ky BP reported by
Major et al (2006) logically coincides with the Bosphorus being deep enough for
commencement of two-way flow, relatively fresh surface water outflow, marine
strontium-rich bottom water inflow, as is the case today. The situation in 8.2ky BP would
therefore have been that of a brackish water (c. 15‰ salinity) lake standing about 30cm
higher than ocean level, sufficient to cause net outflow through the Bosphorus. The
following subsections discuss the evidence as to what happened next.
3.6.2 The Lacustrine-Marine Transition
One indicator of raised salinity in the Black Sea is the reported presence of marine
mollusc species. Ryan et al (2003) date their first appearance in the Bosphorus Strait
itself to 8.0ky BP (after calibration), although Kerey et al (2004) suggest 7.4ky BP at the
southern end and significantly later at the northern end. Marine molluscs appear in the
Black Sea itself just before the sudden onset of sapropel formation in 7.6ky BP.
CHAPTER 3 GEOLOGICAL EVIDENCE
69
-200
-150
-100
-50
0
50
100
-16000 -14000 -12000 -10000 -8000 -6000 -4000 -2000 0
Surface Level (m)
Calendar years BP
Oceans
Black Sea
Caspian Sea
Bosphorus Sill
Manych Sill
-100
-50
0
50
100
150
200
250
-16000 -14000 -12000 -10000 -8000 -6000 -4000 -2000 0
Volume flow (km3/year)
Calendar years BP
Bosphorus
Manych Spillway
0
5
10
15
20
25
30
-16000 -14000 -12000 -10000 -8000 -6000 -4000 -2000 0
Salinity (‰)
Calendar years BP
Black Sea
Caspian Sea
Bølling-
Allerød
Younger
Dryas
8.2ky BP
event
Bølling-
Allerød
Younger
Dryas
8.2ky BP
event
Bølling-
Allerød
Younger
Dryas
8.2ky BP
event
Figure 3.15. Predicted changes in water level, flow rate and salinity (Thom, 2010b)
Sapropel formation however appears to present a clear anomaly. The deposit includes
a fully marine faunal assemblage and has been dated everywhere in the Black Sea at all
depths below 200m to 7.6ky BP (Jones and Gagnon, 1994; Ryan et al, 1997), at which
time the average salinity should have been rising only slowly. This is therefore a major
surprise. Under conventional assumptions of post-glacial sea level rise it is quite unclear
what hydraulic mechanism could have produced the stratification of the water column
required for sapropel formation at such a moderate salinity together with the seemingly
instantaneous introduction of marine organisms. Both Degens and Ross (1972) and
Boudreau and Leblond (1989) calculated that the process should have taken around 2000
CHAPTER 3 GEOLOGICAL EVIDENCE
70
years; it should also have been progressive, commencing in the deepest basins. A sudden
transformation implies a sudden seawater influx, probably predating the onset of sapropel
by a few centuries (Kerr, 1998). Inflow of fresh water from rivers would in that case
immediately have started to overlie the marine water mass, restricting mixing of the water
column, creating anoxic conditions at depth and inducing sapropel formation. Since the
Black Sea was already at the level of the Mediterranean, the only mechanism by which
rapid marine inflow could have been induced is through a substantial rise in
Mediterranean Sea surface level – in line with the hypothesis proposed here.
A sudden seawater inflow also explains the phenomenon of marine life being present
in the Black Sea at a similar date or even before being found in the Bosphorus. The
inrush would inevitably have introduced Mediterranean fauna. However within a few
years the Bosphorus would have seen a resumption of predominantly brackish water
outflow as the stratified water column developed, only suitable to species tolerant of
salinities lower than about 18‰. It may have been several centuries before two-way flow
resumed and truly marine bottom water conditions developed.
3.6.3 The Bosphorus
The event proposed in this report was essentially non-violent in nature. However the
predicted rise and fall in ocean level relative to the land would have had violent
consequences in the case of flow into the Black Sea. Hydraulic modelling (see Appendix
E for an outline) suggests that the flooded Bosphorus channel would have had to
accommodate over 100km3/day at the peak of the inflow period, sustaining flows of up to
7m/sec in average velocity. During subsequent outflow the predicted peak average
velocity is also around 7m/sec, occurring in two major pulses as the planet first righted
itself and then tilted the other way. The final phase would have been a further violent
south-to-north inflow with average velocities up to 6m/sec, very similar to that initially
proposed by Ryan et al (1997), as the Mediterranean regained its proper level. In total it
is predicted that average flow velocities of over 4m/sec would have been taking place for
a cumulative total of approximately nine years; the scour induced would have been
considerable, removing sediment from the floor of the Bosphorus and rapidly exposing
bedrock.
One might expect the physical effects of such massive flows to be obvious – and they
are. kaşan et al (2005) report a synthesis of evidence from seismic and bathymetric
studies of the floor of the Bosphorus and they conclude that a “massive erosion” event
took place some time prior to 7.4ky BP and consider that it was caused by a south-to-
north flow of water. The key evidence is seen in seismic profiles showing that parallel
bedded sediments have been deeply incised. They estimate that between 20m and 34m of
sediment must have been eroded, in many places exposing basement rock from the Upper
Cretaceous period. Kerey et al (2004) present a detailed evaluation of sediment cores and
seismic profiles from the central Bosphorus and they found that only the deepest basin
(on the transect they surveyed) contained pre-Holocene sediment; elsewhere the
basement rock was directly overlain by Holocene deposits post-dating the erosion event.
Those pre-Holocene sediments that were present dated to 26-16ky BP (Çağatay et al,
2000); they were lacustrine and included typical Black Sea fauna (gastropods, ostracodes
etc.), suggesting that they had been deposited during a long period of freshwater outflow
from the Black Sea. However these deposits had clearly been truncated at a level 87m
CHAPTER 3 GEOLOGICAL EVIDENCE
71
beneath the current water surface; neighbouring cores, one in a similarly deep basin,
found only bedrock beneath the more recent Holocene deposits. Gökaşan et al (2005) also
draw attention to submarine canyons at either end of the Bosphorus, suggesting that the
northern one should be associated with this erosion event.
This evidence is exactly what would be expected as a consequence of the flooding
predicted following polar shift. The scouring of basement rock implies extremely rapid
flow. In fact the profile of the bedrock (Algan et al, 2001) is a rather strange feature of
the Bosphorus, with three sills about 80m below the current surface and deep basins (to
around -150m) between them, and it is possible that this profile is itself chiefly due to
repeated episodes of violent and extremely turbulent water flow. Majority opinion
appears to be that the Bosphorus has been in existence throughout most of the last two
million years since there is evidence of Black Sea fauna in the Sea of Marmara during
this period, although the lack of sediments pre-dating 26ky BP has forced Gökaşan et al
(2005) to propose that the connection between the Black Sea and the Sea of Marmara was
via the Gulf of Izmit and the Sakarya River in former times, a solution also proposed by
Yanko-Hombach et al (2007) to explain the appearance of marine fauna in the Black Sea
before they are found in the Bosphorus. Yet careful examination of the geology of the
Izmit plain and the adjacent Sapanka Lake region (Gürbüz and Leroy, 2010) has revealed
no physical evidence for this. Clearly the flood flows expected due to polar shift provide
an alternative explanation for the denuding of the rock surface and consequent removal of
sediments, just as they provide an explanation for the early arrival of marine fauna and
the onset of sapropel formation.
Finally, the presence of an under-sea canyon at the northern mouth of the Bosphorus
is an expected consequence of the final violent re-filling phase, lasting about four years,
immediately prior to which the level of the Black Sea north of the Bosphorus would have
stood only slightly above the level of the eroded Bosphorus sill. The much broader,
shallower canyon at the southern end reflects the fact that most of the southward high-
flow-rate phase would have taken place in a deeper water environment.
Thus it would appear that the physical evidence of the Bosphorus is hard to explain in
any way other than as a consequence of the event – or series of events – proposed in this
report. If this interpretation of the evidence is correct then the same should also be true of
the strait to the south, the Dardanelles.
3.6.4 The Dardanelles
The Dardanelles Strait, connecting the Sea of Marmara and the Aegean, is broader and
deeper than the Bosphorus and the maximum flood flow during the polar shift event is
predicted to have been less rapid than that through the Bosphorus – see Appendix E for
details. However, average velocities of 6m/sec are predicted during the second phase of
tilting motion, during which the level of the Aegean would have fallen by 230m relative
to the land and the Dardanelles would have become a vast outflowing cascade. Average
velocities of 3m/sec are predicted during initial inflow and 2m/sec during final infilling.
Clearly, very substantial erosion should have occurred.
Gökaşan et al (2010) present the key evidence, again derived from seismic profiles
and bathymetric measurements. They identify three distinct layers of sediment overlying
what is considered likely to be basement rock, the latest of which (their Unit 1) represents
Holocene deposits, laid down under relatively low velocity flow conditions. Through
CHAPTER 3 GEOLOGICAL EVIDENCE
72
much of the strait itself these deposits directly overlie bedrock. However at either end
they overlie what Gökaşan et al (2010) suggest are prograding delta deposits (Unit 2),
comprising irregular seismic reflection patterns. In the context of the event proposed here
these materials represent the sediment that used to line the Dardanelles channel but which
was scoured and re-deposited up to 100km from the northern mouth of the strait; at the
southern end the deposit is less in volume and extends about 50km into the Aegean. The
total volume of this deposit is estimated here, from the information given by Gökaşan et
al (2010), to equate to 30-40m removed from the floor of the Dardanelles. The fact that
the largest quantities are found at the northern end is logical in that the first episode of
violent flow would have been the south-to-north infilling phase, following which the bed
of the Dardanelles would have been substantially denuded.
Most interestingly Gökaşan et al (2010) also identify a further deposit (Unit 3), also
missing through most of the strait itself but present at either end (particularly the northern
end) where it underlies Unit 2. In the strait, pockets of material assigned to this stratum
were found in channels that had been incised into the bedrock. The inference would
appear to be that an earlier and possibly even greater water flow event had been
responsible for incising these channels, while subsequent events had only succeeded in
truncating the sediments, leaving remnants in the deepest channels. The thickest Unit 3
deposits were found north of the Dardanelles, between 50km and 100km out into the Sea
of Marmara. It is suggested here that any deposits that had previously existed closer to
the strait were removed by the 8.2ky BP flows. The same explanation is proposed for the
relatively thin deposit found at the southern end of the strait.
Finally, as in the case of the Bosphorus, the Dardanelles channel feeds into under-sea
canyons at either end, most dramatically the Şarköy Canyon at the Sea of Marmara end.
As in the case of the Bosphorus, the presence of these canyons represents further
evidence in support of the event or events proposed here.
3.6.5 Conclusion
The physical evidence for Black Sea and Caspian Sea levels, sediments and salinities
during the late Pleistocene and Holocene eras has been compared to a simulative
hydrological model. While in most respects the model can be constrained to match the
evidence by using reasonable input values, it is quite unable to support the observed
sudden onset of sapropel formation in the Black Sea in 7.6ky BP together with the arrival
of Mediterranean fauna. This fact demands a sudden ingress of marine water at a date a
few hundred years prior to the start of sapropel formation and therefore provides strong
evidence for the sea level rise event outlined in this report.
Furthermore, evidence of sudden, massive and highly erosive water flow has been
found in both the Bosphorus and the Dardanelles. Whilst authors such as Gökaşan et al
(2005 and 2010) ascribe this to the ‘Black Sea Flood’ event described by Ryan et al
(1997), the evidence that this flood never took place in the form initially envisaged, for
example the detection of marine strontium input in about 9.4ky BP (Major et al, 2006), is
now widely accepted. Thus the explanation offered here would appear to be the only
rational one remaining.
CHAPTER 3 GEOLOGICAL EVIDENCE
73
3.7 The Aral-Caspian Basin
The prediction made here is that the rise in ocean levels caused by polar shift would have
been sufficient to flood the entire plain of Central Asia, including the Caspian and Aral
Seas. The sequence of events has been estimated by constructing a simplified flow model
to describe the system as a whole (see Appendix E), of necessity an approximate model
but sufficient to generate working estimates of flow velocities. As reported in the
previous section, a very high initial inflow is predicted through the Dardanelles and the
Bosphorus, reaching average velocities of 3m/sec through the Dardanelles and 7m/sec
through the Bosphorus. When the Black Sea level had risen sufficiently, water would
have spilled across the Manych sill – see Figure 3.16 – and into the Caspian, reaching a
peak average velocity of nearly 2m/sec at the narrowest point. However at about the same
time Arctic-originating water would also have started to flow into the Aral-Caspian basin
through the Turgay Valley, rapidly reaching a peak flow rate of over 1500km3per day; a
further ingress would also have occurred across the Dvina-Pechora-Kama watershed west
of the Ural Mountains. However, peak flow velocities would only have endured for a
matter of days before rising water levels led to the drowning of all these valleys and
channels. There would then have been about two years of near-zero flow velocity
throughout the system, followed by level decrease and overspill initially through the
Turgay valley and Dvina-Pechora-Kama systems and then, for a period extending to
around 14 years, through the Manych and into the Black Sea. Thus it would be reasonable
to seek evidence both in the water chemistry of the Caspian and Aral Seas and in the
sedimentation-erosion pattern found in the Manych spillway and the other valleys
mentioned.
Figure 3.16. Manych spillway location
In this context it is of note that Chepalyga (2007) has proposed that a series of so-
called ‘superfloods’ occurred across the various valleys and water bodies of Central Asia
during the early Khvalynian transgression of the Caspian Sea (16-12ky BP approx.),
caused by pulses of water from melting ice sheets. The evidence for this included the
Black Sea Caspian Sea
Volga
L. Manych-Gudilo
Kuban
Don
Donets
Dnieper Ural
Terek
Kuma
Caucasus Mountains
Manych
Spillway
Sea of
Azov
CHAPTER 3 GEOLOGICAL EVIDENCE
74
geology of the Manych spillway region, where he drew attention to the occurrence of
longitudinal ridges between parallel valleys. His suggested flow through the Manych
spillway at the peak of a superflood was 1000-1500km3/year, between four and six times
the flow of the Volga River today. In opposition to this interpretation of the evidence
Svitoch (2009) is among those who have argued that there never was any such source of
water and that therefore these floods could not have occurred.
3.7.1 The Manych Spillway
The surface of the Manych spillway today reaches an elevation of +26m where it cuts
through the Yergeni ridge. The Yergeni ridge otherwise forms a 100m-high barrier
between the Black Sea and the Caspian, the same barrier that also keeps the waters of the
Don and Volga rivers apart (other than via the Volga-Don canal). To the west of the
+26m watershed lies the 80km-wide Manych depression in which lies Lake Manych-
Gudilo, over 100km long but only 1-2km in width over much of its length. To the west of
this is the equally narrow and only slightly shorter Veselovskoye reservoir from which
the Manych River carries excess water to the Don and thence to the Sea of Azov and the
Black Sea. The average gradient of the spillway is about 1 10-4.
Despite the current sill level being at +26m, there is abundant evidence (Chepalyga,
2007; Antonova and Svitoch, 2008; Svitoch, 2008) that the Caspian Sea reached a water
surface elevation of about +50m at some stage during the early Khvalynian transgression.
Hydraulic flow modelling carried out for this report gave a maximum difference of 7m
between the surface of the Caspian and the level of the sill, suggesting a minimum sill
level of at least +40m (as used in the predictions in the previous section). Mamedov
(1997) has argued that tectonic uplift has significantly raised some parts of the Caspian
region since that time, although the estimated rate is only 0.2-0.4m per millennium. This
could mean that the sill had been as low as +35m during the Khvalynian phase. However
Rychagov (1997) reported high-level early Khvalynian deposits in coastal Dagestan in a
region not subject to this tectonic uplift effect. On balance therefore the conclusion
appears unavoidable that the sill was at or above +40m at some stage during the early
Khvalynian transgression, which means that something has since caused significant
erosion.
Popov (1983) developed a series of geological sections through the sediments of the
Manych, widely referenced and reproduced by other authors. Svitoch (2009) highlights a
section across the Manych depression, reproduced in Figure 3.17 in schematic form. In
summary, 40m of deposits of Khazarian age (c. 400-200ky BP) were found between
-50m and -10m in elevation. Into this a wide valley had been formed and then later filled
with Karangatian sediments (c. 120-70ky BP). Overlying these layers were 30-40m of
Burtass-Gudilovian loams (c. 45-20ky BP – Pilipenko et al, 2009), bringing the ground
level to +30m in places. These strata are all readily explicable. During the warm
interglacial phases around 420, 320, 220 and 120ky BP (Petit et al, 1999) the Manych
would have been a strait of at least 10km width connecting the Caspian and the Black
Seas. These conditions, relatively deep water and low flow rates, would have allowed
sedimentation to occur. The incised Karangatian deposit suggests a period of erosion due
to faster flow at some stage during the rapid warming phase between 120ky BP and
110ky BP, before rising ocean levels led to increased depth and renewed sedimentation.
Faunal remains within the Burtass-Gudilovian deposits are of Caspian origin, implying
CHAPTER 3 GEOLOGICAL EVIDENCE
75
that these sediments were laid down during outflow across a near-flat plain during a
prolonged Caspian highstand.
Figure 3.17: Schematic of section across Manych Depression (after Svitoch, 2009)
However the 25-30m deep incisions in the Burtass-Gudilovian layers shown in Figure
3.17 are much less readily explicable. They have created a series of flat-bottomed troughs
with up to 5m of Khvalynian material at the bottom of each. The ridges between these
troughs were the features noted by Chepalyga (2007) as indicating a ‘superflood’. Svitoch
and Yanina (2001) reported on the mollusc fauna assemblage from these so-called
Khvalynian deposits, in particular the finding of a 1.5m thick mixed layer containing
mollusc shells from many different eras, from pre-Khazarian to Khvalynian. Clearly these
shells had been eroded and re-deposited and, based on the latest of the specimens found,
this re-deposition had occurred at the end of the early Khvalynian period. Their most
recent 14C date, 8.07ky BP, was obtained from a 2m thick deposit at another site
containing only early Khvalynian remains. Mamedov (1997) also highlighted the widely
scattered ages determined from supposedly Khvalynian materials: 3-31ky BP from 14C;
10-24ky BP from U-Th; 42-71ky BP from thermo-luminescence. Logically these
discrepancies are at least partially due to the extensive mixing and re-deposition of
materials found by Svitoch and Yanina (2001).
The modelling introduced in the previous section suggested that the maximum flow
volume through the Manych spillway during the early Khvalinian transgression was
87km3/year, in agreement with Svitoch (2009), which should not have been responsible
for more than about 1m of erosion to the then +40m sill (Thom, 2010b). If eroded
material had been deposited within the flat Manych depression, where the water flow rate
would have been low, the thickness could not have been significant; nor could the width
of flow have extended across more than a few hundred metres. Yet many metres of
erosion have evidently occurred, creating channels totalling well over 10km in width and
leaving the sill at only +26m. In order to generate erosion on this scale, assuming a
normal river flow velocity of 0.75m/sec and a gradient of 1 10-4, the water depth would
have to have been around 5m (Manning, 1891). Over a 10km width this would give a
flow rate of nearly 1200km3/year, about five times that of the Volga today – matching the
proposal made by Chepalyga (2007). However, to achieve the 25-30m of erosion shown
in Figure 3.17 would require thousands of years – and no such flow is believed to have
occurred over a prolonged period. A further deduction can be made from the fact that
there are up to four parallel eroded channels, recalling the channels incised into the
bedrock beneath the Dardanelles. In normal riverine flow this is practically
+30m
+20m
+10m
0m
–10m
–20m
–30m
–40m
–50m
+30m
+20m
+10m
0m
–10m
–20m
–30m
–40m
–50m
Khazarian
Karangatian
Burtass-Gudilovian
Khvalynian
Holocene
25km
CHAPTER 3 GEOLOGICAL EVIDENCE
76
inconceivable. Meandering can lead to channel divisions – but in this case the flow
channels are almost straight. The logical inference is that these channels were not
produced by thousands of years of flow from a very large river but by a much shorter
period of much higher velocity flow.
The issue of sediment deposition also has to be addressed. The mixed sediments
encountered by Svitoch and Yanina (2001) imply extensive and violent erosion by fast-
flowing water, followed by a period of lower flow rate; yet no such set of events is
known to have occurred during or after the Khvalinian transgression. Mamedov (1997)
also remarked that it is unclear how such thick sediments came to be deposited during the
Khvalynian phase.
This evidence is however entirely compatible with the polar shift event proposed
here. The principal erosion phase would have occurred over a 14-year period as the water
that had filled the Aral-Caspian basin spilled out through the Manych at up to thirty times
the rate suggested by Chepalyga (2007) and at predicted average velocities of up to
around 2m/sec – although much higher local velocities are likely. As flow volume and
velocity decreased, extensive sedimentation would then have taken place along the near-
flat floors of the scoured-out channels.
3.7.2 The Turgay Valley
The Turgay – see Figure 3.18 – is the valley that connects the Aral-Caspian basin to the
West Siberian plain to the north, cutting through a belt of relatively high (+200m to
+500m) ground. The floor of the valley itself rises to a sill level of +126m. It is a very
wide shallow valley, apparently much too large for the Ubagan River which today runs
north to join the much larger Tobol River. The physical form of the valley strongly
suggests that it has acted as the conduit for a significant flow of water at some stage in its
history, and it is widely acknowledged (e.g. Mangerud et al, 2001) that this was the case
at certain times during the Pleistocene when glaciation effectively dammed the West
Siberian plain, creating a large glacial lake that overspilled south through the Turgay. The
presence of the valley itself – or at least its southward manifestation toward the Aral Sea
– can therefore be explained without recourse to the catastrophic event described in this
report.
There is little obvious evidence of sill erosion in the Turgay. The floor of the valley
comprises thick deposits of wind-blown sand, but this deposition phase came to an end in
the early Holocene, presumably because increased rainfall allowed vegetation to stabilise
adjacent sandy soils and prevent further erosion. A date of 11ky BP has been obtained
from sediment just 4m below the current surface, compared to 19ky BP at 34m depth
(Mangerud et al, 2001), illustrating the cessation of deposition some time after 11ky BP.
These dates also imply that only a few metres of material can have been removed during
the 8.2ky BP event. On the other hand the flow through the Turgay is predicted to have
been of relatively short duration in both directions, about two years either way, and the
sand grain size involved is much larger than that of the silty loams of the Manych and so
much more difficult to erode. Nevertheless the size of the valley leading north definitely
implies that significant erosion took place at some stage, much greater erosion than could
possibly be ascribed to the modern Ubagan River. It appears unlikely that this could have
been achieved by the 8.2ky BP event alone, given the constraints imposed by the thick
CHAPTER 3 GEOLOGICAL EVIDENCE
77
wind-blown sand deposit, but it may quite possibly have been achieved by similarly large
flows at the close of earlier glacial periods prior to wind-blown sand deposition.
Figure 3.18: The Turgay valley and other spillways
3.7.3 The Chemistry of the Caspian Sea
The water of the Caspian Sea differs chemically from that of the oceans – but by
surprisingly little. Tuzhulkin et al (2005) estimated that the difference would appear to
have developed over only about 10ky, brought about by interaction with terrestrial run-
off, a deduction which is incompatible with current understanding of Caspian Sea history,
since the last time that ocean waters are believed to have penetrated the Caspian was
during the Karangatian phase, over 100ky ago.
The calculation is sensitive to one key parameter however and that is the extent to
which Caspian Sea water has been drained off through the Kara Bogaz Gol (see Figure
3.18), a large bay in the eastern part of the sea that acts as an evaporation lagoon. Current
outflow from the main body of the Caspian through the channel to the Kara Bogaz Gol is
about 10km3/year (Aladin et al, 2008), representing about 3% of the total river inflow to
the Caspian (the remainder being lost through evaporation); yet in the early 20th Century
it is known that the outflow rate was much higher. In fact it is possible to deduce from the
evaporite deposits (mainly halite and sodium sulphate) found covering the floor of the
Kara Bogaz Gol that there have been three (Garrett, 2001) or possibly four (Karpychev,
2007) long phases of evaporite deposition since about 8ky BP, the last of which continues
today. It is estimated here that the total known salt deposit beneath the Kara Bogaz Gol
represents a loss of Caspian Sea salinity of around 5-10‰, and it is certain that the total
salt loss has been significantly greater than this since much will have been lost to saline
groundwater. Since an average flow of 10km3/year for the period since 8ky BP would
Black Sea
Barents Sea
Aegean
Sea
Baltic
Sea
Caspian
Sea Aral Sea
Manych
Spillway Turgay
Valley
Bosphorus
Dardanelles
R. Volga
R. Kama
R. Ural
R. Tobol
R. Ubagan
R. Irtysh
R. Ob
R. Don
R. Dnieper
R. Dvina
R. Pechora
R. Yenisei
Ural
Mountains
Dvina-Pechora-
Kama watershed
Kara Bogaz
Gol
CHAPTER 3 GEOLOGICAL EVIDENCE
78
have resulted in the loss of about 8‰ it is therefore reasonable to suggest that the actual
average flow may have been significantly higher than this.
In fact the calculation carried out for this report (Appendix F) has found that an
average outflow of 17km3/year would be required to transform the moderately
concentrated seawater left following the proposed oceanic incursion and subsequent
evaporation, producing the current salt balance in most major respects. The only
significant anomalies are the calcium content – predicted to be much higher than is
actually the case – and the sulphate content – under-predicted. However the lack of
calcium is readily explained by the very low solubility of calcium in water and its
propensity to precipitate out, and the additional sulphate is equally readily explained by
assigning a greater percentage of inflow water to western rivers, notably the sulphate-rich
River Kura, and to inflow from the Aral Sea (see next subsection). In theory it would be
possible to devise an alternative explanation for the current salt balance, invoking
substantial input from as-yet uncharted sub-sea thermal spring sources, but this would
appear to be a denial of a much more straightforward explanation. Clauer et al (2000) are
forced to propose such sub-sea thermal spring sources in order to explain the current
87Sr/86Sr ratio of the Caspian Sea, which is higher than that of the Volga and almost all
other inflow rivers. Yet when a concentrated seawater origin dating to 8.2ky BP is
assumed, together with a 17km3/year average outflow to the Kara Bogaz Gol, then the
predicted strontium content and isotope ratio are both very close to those actually present.
3.7.4 The Aral Sea
In most respects the Aral Sea would be expected to provide no evidence of any seawater
inundation event since for much of its Holocene history it has overflowed into the
adjoining Sarikamish basin and thence via the Uzboy channel to the Caspian, where it
would have introduced the high-sulphate-content water of the Syr Darya River. However
this may not have been the case for a period of around 300 years immediately following
the 8.2ky BP event, predicted by climatologists (e.g. Wiersma and Renssen, 2006) to
have been very dry in Central Asia. Thus it is legitimate to expect that there may be
something in the sediment record that points to a period of significantly different water
chemistry. In that respect the 15-20cm thick gypsum-rich layer reported by Le Callonec
et al (2005) dating to this approximate period and covering about 200-300 years of
sedimentation may be significant, together with the sudden spike in the strontium content
reported in Section 3.4.4. However, it has to be admitted that these signals, though
supportive, are inconclusive.
3.7.5 Conclusion
The event proposed in this report explains both Caspian Sea chemistry and the
morphology of surrounding geological features. If it never took place then there is
currently no explanation for the erosion that has occurred along the Manych channel,
which appears to have been formed by a flood flow of massive proportions, nor for the
deposition of sediment that has occurred at the bottom of those channels; neither is there
any simple explanation for current Caspian Sea chemistry. With additional suppo