Content uploaded by Franz Rubel
Author content
All content in this area was uploaded by Franz Rubel on Jul 04, 2017
Content may be subject to copyright.
Content uploaded by Franz Rubel
Author content
All content in this area was uploaded by Franz Rubel on Dec 27, 2013
Content may be subject to copyright.
CHARACTERIZING GLOBAL CLIMATE CHANGE BY MEANS OF KÖPPEN
CLIMATE CLASSIFICATION
Christoph Beck
1
, Jürgen Grieser
1
, Markus Kottek
2
, Franz Rubel
2
and Bruno Rudolf
1
1
Global Precipitation Climatology Centre, Deutscher Wetterdienst, Kaiserleistr. 44, 63067 Offenbach, Germany
2
Biometeorology Group, University of Veterinary Medicine, Veterinärplatz 1, 1210 Vienna, Austria
INTRODUCTION
Global climate classifications were originally constructed in order to designate the
manifold existing local climates to an adequate number of climate types and to
determine the spatial distribution of these types on the basis of climatic data for a
reference period. Thus, climate classifications are introduced in order to reflect the
mean spatial climate characteristics.
However, the underlying climate variables are subject to temporal variations and so
are the results of climate classifications. Therefore climate classifications may not
only be used to determine the mean state of the climate. They can also be utilized to
analyse global and regional scale climate variations by applying them to varying time
periods. Spatio-temporal variations of climate types resulting from effective climate
classifications do not only reflect modifications of climatic parameters. By definition
they are closely linked to different environmental conditions (e.g. vegetation) and
therefore they may moreover be used for investigating the potential impact of past,
present and projected future climate change on environmental systems.
Varying aspects of global and regional climate change have been investigated on the
basis of the well-known climate classification according to Köppen (e.g. Köppen
1936, Geiger 1961) by several authors. Fraedrich et al. (2001) analysed the shifts of
modified Köppen climate types on a global and continental scale during the 20
th
century, Grieser et al. (2006) compared the results of a Köppen classification applied
to gridded data from the periods 1951 – 1975 and 1976 - 2000. Focusing on a more
regional perspective Suckling and Mitchell (2000) investigated variations of the
boundary between the Köppen C and D climates in the Central United States and
Wang and Overland (2004) analysed arctic climate change during the 20
th
century on
the basis of a modified Köppen classification. Triantafyllou and Tsonis (1994)
assessed the sensitivity of the Köppen classes to long-term climate change on the
basis of long station time series. By applying the Köppen classification to the output
of general circulation models (hereafter GCM) Guetter and Kutzbach (1990)
estimated the main characteristics of glacial and interglacial climates, Kalvova et al.
(2003) compared the results of Köppen classifications applied to several 20
th
century
observational data sets and varying GCM outputs and Lohmann et al. (1993) used
the Köppen classification to validate GCM control runs and as well to analyse
scenario runs concerning spatio-temporal variations of climate types.
In this contribution the spatio-temporal climate variations during the second half of
the 20
th
century are investigated on global and continental scale by applying the
Köppen climate classification to monthly precipitation and temperature data available
from two most recently constructed globally gridded data sets (Mitchell and Jones
2005, Beck et al. 2005).
DATA AND METHODS
Monthly temperature and precipitation data for the period 1951 - 2000
Monthly mean temperatures and monthly precipitation sums have been taken from
the CRU TS 2.1 data set (Mitchell and Jones, 2005) provided by the Climatic
Research Unit (CRU, University of East Anglia, Norwich, GB) and the VASClimO
v1.1 data set (Beck et al. 2005) available from the Global Precipitation Climatology
Centre (GPCC, Deutscher Wetterdienst, Offenbach, Germany) respectively. Both
data sets cover the analysed period from 1951 – 2000 and provide data for the global
land areas on a 0.5° by 0.5° lat./lon. grid. Due to insufficient station densities
Greenland and Antarctica are not included within the VASClimO precipitation data set
(Beck et al. 2005) and therefore these regions are also excluded from the analyses
presented here.
The same data base has also been used for applications of the Köppen climate
classification on the basis of state of the art data sets most recently performed by
Kottek et al. (2006) and Grieser et al. (2006).
The Köppen climate classification
Most global climate classifications are so called effective classifications which use
threshold values (mostly of monthly temperature and precipitation data) based on
environmental characteristics (e.g. vegetation) for the definition of boundaries
separating different climate types. Probably the most prominent effective
classification is the classification scheme according to Köppen (1936) that is based
on the characteristics of the mean annual cycle of temperature and precipitation.
Threshold values utilizing these climatic parameters have been defined in order to
designate climate types that reflect major environmental characteristics (i.e. the 10°
C isotherm of the warmest month in the year as indicator for tree growth).
The resulting five major climate types according to Köppen (1936) designated by
capital letters are:
-
Tropical rain climates (A) – where the mean temperature of the coldest month
exceeds +18.0°C.
-
Arid climates (B) – are defined as follows on the basis of the average annual
precipitation sum R (cm) and the annual mean temperature T (°C):
R < 2T + 28 (where summer rain is dominating)
R < 2T +14 (where no pronounced annual cycle is observed)
R < 2T (where winter rain is dominating)
-
Temperate rain climates (C) – where the mean temperature of the coldest month
is between –3.0°C and +18.0°C.
-
Boreal forest and snow climates (D) – are characterized by a mean temperature
of the warmest month exceeding 10.0°C and a mean temperature of the coldest
month below –3.0°C.
-
Cold snow climates (E) – are defined by a mean temperature of the warmest
month below 10.0°C.
A further subdivision of these five main types concerning further parameters
representing temperature and precipitation conditions leads to subtypes designated
by two resp. three letter codes (e.g. Cf – where f indicates humid conditions
throughout the year, and Cfa – where a indicates hot summer conditions). In this
paper the presentation of results is restricted to the five main Köppen types described
above.
A more comprehensive overview on the Köppen classification can be found for
example in Kraus (2001).
Analysing spatio-temporal variations of Köppen climates
In order to investigate the spatiotemporal variations of Köppen climate types during
the 1951 – 2000 period the classification scheme has been applied to monthly mean
values determined for sliding 15-year intervals, as has been proposed by Fraedrich et
al. (2001). For each of these 36 intervals (indicated by the respective central year)
the areas that are designated to each Köppen climate type have been calculated and
continuous time series of the varying relative area (percentage of the global resp.
continental land area excluding Greenland and Antarctica) occupied by the certain
types have been obtained.
For each 15-year interval its similarity to the classification results obtained on the
basis of long term mean values from the whole 50-year period from 1951 – 2000 is
estimated in terms of the percentage of the global land area assigned to the same
climate type.
The spatial distribution of the differences between classification results obtained for
the 50 year period and the one 15-year interval showing greatest deviations were
determined and the respective transfer matrix summarizing the respective
redistributions between the five main climate types was calculated.
As a change from one climate type to another between two periods may be caused
by rather small differences concerning the relevant climatic parameter (e.g. annual
precipitation sum) for those gridcells assigned to different climate types in the two
periods it is furthermore investigated in how far these differences reflect a statistical
significant climate change concerning those parameters that are relevant for the
respective redistribution by applying the non-parametric U-test according to Mann-
Whitney (e. g. Bahrenberg et al. 1990). Concerning for example a change from
climate type A to climate type C it is tested if the temperature of the coldest month
shows statistical significant differences (at the 90% or 95% level) between the two
periods.
RESULTS
The Köppen classification restricted to the five main types based on monthly mean
values for the period 1951 – 2000 is shown in Fig. 1. The temporal variations of the
relative area occupied by each of the five main Köppen climates estimated for sliding
15-year intervals during the 1951 – 2000 period are depicted in Fig. 2 for the whole
globe and in Fig. 3 for the continental scale regions (marked by grey rectangles in
Fig. 1).
Concerning the long-term variations of the global land area occupied by the five main
Köppen types (see Fig. 2) the most striking features are a distinct reduction of the
areas assigned to the polar E and boreal D climates as well as a concomitant
expansion of dry B climates. Tropical A and temperate C climates on the other hand,
although showing marked decadal scale variations exhibit no such distinct long term
trends.
Figure 1: Spatial distribution of the five main Köppen climate types determined for the period
1951 – 2000. Rectangles indicate the continental sub-regions for which selected results are
presented.
As depicted in Fig. 3 the trends in land fractions occupied by Köppen classes for
each of the continental-scale regions may differ considerable from the global trends.
North America
Most striking for North America appear distinct reductions of polar E and as well dry
B climates. Simultaneously the area occupied by the temperate C and boreal D
climate types increases.
Asia
In Asia polar E climates exhibit a reduction in coverage as well. But in contrast to
North America dry B climates expand whereas temperate C and boreal D climates
feature downward trends in coverage.
Europe
An expansion of dry B climates can also be stated for Europe but most important
appears a sharp increase / decrease of the area occupied by temperate C / boreal D
climates affecting up to 10% of the European land area.
South America
In South America positive trends occur for temperate C and tropical A climates
whereas dry B climates undergo a long term reduction in coverage. Additionally a
slightly decreasing trend in the coverage of cold E climate can be detected.
Africa
The expansion of dry B climates that can also be seen on the global scale and in
Europe and Asia appears most striking in Africa where its increase in percentage
since 1951 reaches around 5%. This expansion of dry B climates is compensated by
concomitant reductions of the area covered by tropical A and especially temperate C
climates.
Australia
Australia shows marked inversely coupled decadal scale variations of tropical climate
types A (and partly temperate C climates) on the one hand and dry B climates on the
other hand. However no clear-cut long-term trends can be deduced.
Figure 2: Time series of the relative area (percentage of the global land area excluding
Greenland and Antarctica) occupied by the main Köppen types.
North America
Asia
Europe
South
America
Figure 3:
Time series of the relative area (percentage of the resp. continental land area
excluding Greenland and Antarctica) occupied by the main Köppen types. For the continental
scale subregions indicated in Fig. 1.
Africa
Australia
Figure 3: cont.
With respect to Fig. 2 it may be concluded that the last decades of the 20
th
century
represent exceptional climate conditions in terms of the distribution of Köppen climate
types. For example the area assigned to dry B climates reaches its maximum in the
1986 – 2000 period while boreal D and cold E climates exhibit global minima
concerning their coverage on the global scale. For each 15-year interval Fig. 4 shows
the relative area assigned to the same climate type as in the 1951 – 2000 period. It
becomes clear that the 1986 – 2000 period indeed shows the most distinct
differences to classification results obtained for the whole period. This is consistent
with the general knowledge that the 1986 – 2000 interval represents a period of
accentuated global warming as it comprises 13 of the 15 warmest years within the
period 1951 – 2000. Thus, it may be considered as a recent representation of a
projected globally warmer future climate documented by e.g. IPCC (2001).
Figure 4: Time series of the relative fraction of global land area denoted to the same Köppen
climate type as in the period 1951 – 2000. Sliding 15-year intervals have been applied.
Against this background it is desirable to take a closer look at the differences
between the results of Köppen classifications applied to the 1986 – 2000 interval and
the whole analysis period from 1951 – 2000. The respective redistributions between
the five main Köppen climate types are summarized in the transfer diagram given in
Fig. 5. The resulting differences between the two periods concerning the spatial
distribution of climate types are depicted in Fig. 6.
96.7% of the global land area are assigned to the same main Köppen climate in the
two periods under consideration. For the remaining 3.3% land fraction redistributions
between the five classes occur between 1951 – 2000 and 1986 – 2000. According to
Fig. 5 major transfers between the five main climates took place from climate type C
to A, from climate types A, C and D to B, from D to C and from E to D. As result the
area occupied by the dry climate type B increases towards the end of the 20th
century while the boreal and cold climates D and especially E exhibit reductions in
coverage in the most recent 15-year interval. The tropical A climate and the
temperate C climate feature only minor differences – slight expansions during 1986 –
2000 - between the two periods.
Fig. 5: Transfers between the five main Köppen climate types from the period 1951 - 2000 to the
most recent 15-year interval from 1986 - 2000 in terms of percentages of the global land area
(excluding Greenland and Antarctica). The four rectangles and the oval represent the five main
Köppen climate types characterised by temperature and precipitation conditions respectively.
Numbers above/below the one letter Köppen codes indicate the relative area occupied by each
type during the 1951 – 2000/1986 – 2000 period respectively. Blue/red Numbers to the right
represent the associated area gains/losses. Arrows indicate redistributions between climate
types exceeding 0.01%.
Only around 92% of the global land area appear as stable climates when comparing
the two selected periods on the basis of the two letter coded 14 Köppen subtypes.
Except for the dry B climates these 14 classes are differentiated concerning
precipitation characteristics. Thus it can be stated that considerable redistributions
occur within the five main climate types, mainly due to precipitation variations (not
shown here). In summary the differences between the period 1951 – 2000 and the
interval 1986 - 2000 regarding the extension of the five main Köppen climates as
displayed in Fig. 5 support a global warming tendency towards the end of the 20
th
century. This warming trend is reflected particularly by a reduction of the area
occupied by boreal D and cold E climate types as well as by a decrease in
precipitation implying an expansion of dry B climates.
As can be seen from Fig. 6 the reduction of the area designated to the boreal and
cold climate types D and E appears most pronounced in high latitudes of North
America and in the north-eastern part of Asia where cold E climates are replaced by
boreal D climates, which in turn are substituted over large areas by temperate C
climates in Eastern Europe and the USA. Expansions of dry B climates on the other
hand occur most strikingly in southern and sub-Saharian Africa and the eastern and
central parts of Asia. Concerning tropical A climates - as temperate C climates
exhibiting slight increases in coverage on the global scale (Fig. 5) – the spatially most
coherent changes can be found in southern Africa where tropical A climates gain
area mainly on the expense of temperate C climates.
Figure 6: Köppen climate types for the period 1986 - 2000 for gridcells with different Köppen
climates within the periods 1951 – 2000 and 1986 – 2000, respectively. Grey and black dots
indicate gridcells for which a difference concerning the respective relevant climate parameter
between the two periods can be determined on the 90% or 95% level of significance,
respectively.
With respect to the statistical significance of differences of the parameters relevant
for classification results between the two periods (as indicated by grey and black dots
for exceedance of the 90% and 95% level of significance respectively in Fig. 6) it
turns out that for around 1.63% of the global land areas a change from one climate
type to another is accompanied by a statistical significant shift in the mean of the
respective relevant climatic parameter at least at the 90% level of significance. Thus
for about 50% of the area for which changes are found these changes can
significantly be attributed to the main variable. Statistical significant differences over
large areas occur especially
-
in Africa where temperate C climates are substituted by tropical A climates,
-
in Africa and eastern Asia where dry B climates replace tropical A and temperate
C climates or boreal D climates respectively, and
-
in the sub-polar regions of North America and north-eastern Asia where boreal D
climates increased their coverage at the expense of cold E climates.
Also climatic differences linked to the distinct expansion of temperate C climates in
eastern Europe and the USA frequently reach statistical significance although most
often merely at the 90% level.
Regarding the expansion and partly also reduction of dry B climates it has to be
mentioned that there are regional differences concerning the climatic parameter to
which the detected redistributions can be attributed. Especially in the western and
central Sahel regions and throughout Australia but also in the USA redistributions
affecting dry B climates seem to be rather connected with temperature variations
than with precipitation changes.
SUMMARY AND CONCLUSIONS
The application of the global climate classification according to Köppen (1936) to
globally gridded climate data for sliding 15-year intervals in the period 1951 – 2000
yields an overview of the spatio-temporal variations of main climatic types in a global
and when applied to regional data subsets in a continental perspective.
On the global scale the main temporal variations concerning the area occupied by
the five main Köppen climate types are characterized on the one hand by long-term
declining trends of the area designated to the climate types representing boreal and
cold climates (D and E) and on the other hand by increasing coverages of the dry B
climates. Although tropical A and temperate C climates show marked decadal to
multi-decadal scale variations - including a decline of A climates until the mid 1980s
and a distinct increase of C climates since the 2
nd
half of the 1960s – no such clear
cut tendencies lasting over the whole 1951 – 2000 period as detected for the above
mentioned types become obvious for these two types. However respective variations
on the continental scale show partly distinctly differing behaviour.
The comparison of classification results achieved for individual 15-year intervals with
the spatial distribution of climate classes obtained on the basis of the whole 1951 –
2000 period reveals that the most recent sub-period from 1986 – 2000 shows
greatest deviations to the long-term mean climate in terms of area fractions of
different climate types.
The redistributions between the five main Köppen climate types that occurred
between the 1951 – 2000 period and the 1986 – 2000 interval affect 3.3% of the
global land area. But only for about the half of this area differing classification results
can be attributed to statistical significant differences of the respective relevant
climatic parameter between the two periods. Transfers between climate types result
in a most recent increase of the dry B climate and concurrent reductions of boreal D
and cold E climates whereas for tropical A and temperate C climates only minor
expansions become visible.
The above findings concerning spatio-temporal variations of the five main Köppen
climates during the 2
nd
half of the 20
th
century are on a global scale generally in line
with the results obtained by several authors who applied the Köppen classification or
modifications of it to observed 20
th
century climate data (e.g. Fraedrich et al. 2001,
Kalvova 2003, Grieser et al. 2006) and with respect to more regional scale variations
are furthermore supported for example by Wang and Overland (2004) and Serreze et
al. (2000) concerning the reduction of cold E climates and by Gonzalez (2001)
concerning the expansion of dry B climates in Africa.
As 13 out of the 15 globally warmest years within the 1951 – 2000 period occured
between 1986 and 2000 this most recent period does not only comprise the effects of
the global warming observed in the 20th century but it may also be considered as an
analogue for a globally warmer climate as it is projected for the future (e.g. IPCC
2001). Thus it may be concluded that the shifts of the main climate types detected for
the 1986 – 2000 interval will persist or may even be amplified under most likely future
conditions of further global warming. This conclusion is supported by the results of
applications of the Köppen climate classification performed on the basis of model
output of coupled and uncoupled GCMs (e.g. Lohmann et al. 1993, Kalvova et al.
2003) indicating an expansion of tropical A and dry B climates and a reduction of
boreal D and cold E climates under the assumption of further greenhouse gas
warming.
REFERENCES
Bahrenberg, G., E. Giese and J. Nipper (1990): Statistische Methoden in der
Geographie – Band 1: Univariate und bivariate Statistik.- Teubner, Stuttgart.
Beck, C., J. Grieser and B. Rudolf (2005): A New Monthly Precipitation Climatology
for the Global Land Areas for the Period 1951 to 2000. Climate Status Report
2004, 181 - 190, German Weather Service, Offenbach, Germany. Reprint
available at http://gpcc.dwd.de.
Fraedrich, K., F.-W. Gerstengarbe and P. C. Werner (2001): Climate shifts during the
last century. Climatic Change 50: 405-417.
Geiger, R. (1961): Überarbeitete Neuausgabe von Geiger, R.: Köppen-Geiger / Klima
der Erde. (Wandkarte 1:16 Mill.), Darmstadt.
Gonzalez, P. (2001): Desertification and a shift of forest species in the West African
Sahel. Climate Research 17: 217-228.
Grieser, J., M. Kottek, C. Beck, F. Rubel and B. Rudolf (2006): World Maps of
Köppen-Geiger Climates calculated from CRU TS 2.1 and VASClimO v1.1, in
preparation.
Guetter, P. J. and J. E. Kutzbach (1990): A modified Koeppen classification applied
to model simulations of glacial and interglacial climates. Climatic change 16:
193 – 215.
IPCC (2001): Climate change 2001: The scientific basis. Contribution of working
group I to the Third Assessment Report of the IPCC. Houghton, J. T. et al.
(eds) Cambridge University Press.
Kalvova, J., T. Halenka, K. Bezpalcova and I. Nemesova (2003): Köppen climate
types in observed and simulated climates. Stud. Geophys. Geod. 47: 185-202.
Köppen, W. (1936): Das geographische System der Klimate (Handbuch der
Klimatologie, Bd. 1, Teil C).
Kottek, M., J. Grieser, C. Beck, B. Rudolf and F. Rubel (2006): World Map of
Köppen-Geiger Climate Classification Updated. submitted to Meteorologische
Zeitschrift.
Kraus, H. (2001): Die Atmosphäre der Erde. 2. Ed. Springer, Berlin.
Lohmann, U., R. Sausen, L. Bengtsson, U. Cubasch, J. Perlwitz and E. Roeckner
(1993): The Köppen climate classification as a diagnostic tool for general
circulation models. Clim. Res. 3: 177-193.
Mitchell, T. and P. Jones, 2005: An improved method of constructing a database of
monthly climate observations and associated high-resolution grids. Int. J.
Climatol. 25, 693-712.
Serreze, M. C., J. E. Walsh, F. S. Chapin III, T. Osterkamp, M. Dyurgerov, V.
Romanovsky, W. C. Oechel, J. Morison, T. Zhang and R. G. Barry (2000):
Observational evidence of recent change in the northern high-latitude
environment. Climatic Change 46: 159-207.
Suckling, P. W. and M. D. Mitchell (2000): Variation of the Köppen C/D climate
boundary in the Central United States during the 20
th
century. Physical
Geography 21: 38-45.
Triantafyllou, G. N. and A. A. Tsonis (1994): Assessing the ability of the Köppen
system to delineate the general world patterns of climates. Geophysical
Research Letters 21: 2809-2812.
Wang, M. and. J. E. Overland (2004): Detecting arctic climate change using Köppen
climate classification. Climatic Change 67: 43-62.
View publication statsView publication stats
