ArticlePDF Available


From 1875 to 1878, concurrent multiyear droughts in Asia, Brazil, and Africa, referred to as the Great Drought, caused widespread crop failures, catalyzing the so-called Global Famine, which had fatalities exceeding 50 million people and long-lasting societal consequences. Observations, paleoclimate reconstructions, and climatemodel simulations are used 1) to demonstrate the severity and characterize the evolution of drought across different regions, and 2) to investigate the underlying mechanisms driving its multiyear persistence. Severe or record-setting droughts occurred on continents in both hemispheres and in multiple seasons, with the ''Monsoon Asia'' region being the hardest hit, experiencing the single most intense and the second most expansive drought in the last 800 years. The extreme severity, duration, and extent of this global event is associated with an extraordinary combination of preceding cool tropical Pacific conditions (1870-76), a record-breaking El Niño (1877-78), a record strong Indian Ocean dipole (1877), and record warm North Atlantic Ocean (1878) conditions. Composites of historical analogs and two sets of ensemble simulations-one forced with global sea surface temperatures (SSTs) and another forced with tropical Pacific SSTs-were used to distinguish the role of the extreme conditions in different ocean basins. While the drought in most regions was largely driven by the tropical Pacific SST conditions, an extreme positive phase of the Indian Ocean dipole and warm NorthAtlantic SSTs, both likely aided by the strong El Niño in 1877-78, intensified and prolonged droughts in Australia and Brazil, respectively, and extended the impact to northern and southeastern Africa. Climatic conditions that caused the Great Drought and Global Famine arose from natural variability, and their recurrence, with hydrological impacts intensified by global warming, could again potentially undermine global food security.
Climate and the Global Famine of 1876–78
School of the Environment, Washington State University, Vancouver, Washington, and Lamont-Doherty
Earth Observatory, Columbia University, Palisades, New York
Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York
Lamont-Doherty Earth Observatory, Columbia University, Palisades, and NASA Goddard Institute for Space
Studies, New York, New York
Lamont-Doherty Earth Observatory, Columbia University, Palisades, New York
University of California, Riverside, Riverside, California
(Manuscript received 19 March 2018, in final form 20 September 2018)
From 1875 to 1878, concurrent multiyear droughts in Asia, Brazil, and Africa, referred to as the Great
Drought, caused widespread crop failures, catalyzing the so-called Global Famine, which had fatalities ex-
ceeding 50 million people and long-lasting societal consequences. Observations, paleoclimate reconstructions,
and climatemodel simulations are used 1) to demonstrate the severity and characterize the evolution of drought
across different regions, and 2) to investigate the underlying mechanisms driving its multiyear persistence.
Severe or record-setting droughts occurred on continents in both hemispheres and in multiple seasons, with the
‘‘Monsoon Asia’’ region being the hardest hit, experiencing the single most intense and the second most ex-
pansive drought in the last 800 years. The extreme severity, duration, and extent of this global event is associated
with an extraordinary combination of preceding cool tropical Pacificconditions (1870–76), a record-breaking El
Niño (1877–78), a record strong Indian Ocean dipole (1877), and record warm North Atlantic Ocean (1878)
conditions. Composites of historical analogs and two sets of ensemble simulations—one forced with global sea
surface temperatures (SSTs) and another forced with tropical Pacific SSTs—were used to distinguish the role of
the extreme conditions in different ocean basins. While the drought in most regions was largely driven by the
tropical Pacific SST conditions, an extreme positive phase of the Indian Ocean dipole and warm North Atlantic
SSTs, both likely aided by the strong El Niño in 1877–78, intensified and prolonged droughts in Australia and
Brazil, respectively, and extended the impact to northern and southeastern Africa. Climatic conditions that
caused the Great Drought and Global Famine arose from natural variability, and their recurrence, with hy-
drological impacts intensified by global warming, could again potentially undermine global food security.
Denotes content that is immediately available upon publication as open access.
Supplemental information related to this paper is available at the Journals Online website:
Corresponding author: Deepti Singh,
1DECEMBER 2018 S I N G H E T A L . 9445
DOI: 10.1175/JCLI-D-18-0159.1
Ó2018 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright
Policy (
1. Introduction
During the late nineteenth century, a series of famines
affected vast parts of Asia, causing mortality on a scale
that would be unthinkable today (Davis 2001). Of these,
the so-called Global Famine lasting from 1876 to 1878
was the most severe and widespread in at least the past
150 years (Hasell and Roser 2018;Gráda 2009;Davis
2001). The Global Famine inflicted acute distress upon
populations in diverse parts of South and East Asia,
Brazil, and Africa, with total human fatalities likely
exceeding 50 million. These famines were associated
with prolonged droughts in India, China, Egypt, Mo-
rocco, Ethiopia, southern Africa, Brazil, Colombia, and
Venezuela (Davis 2001;Clarke 1878;Hunter 1877).
Historical documentation indicates famine-related
mortality between 12.2 and 29.3 million in India, be-
tween 19.5 and 30 million in China, and ;2 million in
Brazil (Davis 2001), amounting to ;3% of the global
population at the time. It was arguably the worst envi-
ronmental disaster to ever befall humanity and one of
the worst calamities of any sort in at least the last 150
years, with a loss of life comparable to the World Wars
and the influenza epidemic of 1918/19. The triggers for
the famine were acute droughts, but political and eco-
nomic factors, especially the neglect or destruction of
traditional systems of water and grain storage, were re-
sponsible for translating crop failure into unprecedented
mass mortality (Davis 2001).
Studies published in Nature in 1877 and 1878 proposed
weakened sunspot activity as the primary cause of the
drought over India (Buchan 1877;Derby 1878;Hunter
1877), although this was soon questioned (Blanford
1887). Only a few modern studies have analyzed the
character, dynamics, and causes of the drought condi-
tions and only in some regions during the Global Famine
(Hao et al. 2010;Aceituno et al. 2009;Kang et al. 2013).
Hao et al. (2010) showed that the 1876–78 drought in
northern China, which resulted in consecutive crop fail-
ures, was the most severe in the last 300 years based on
seasonal precipitation reconstructions over the Yellow
River basin. Aceituno et al. (2009) showed that north-
eastern Brazil experienced severe dry conditions, and
parts of the northwestern coastal regions and south-
eastern South America experienced intense rainfall
and frequent flooding during the 1877/78 period. These
studies ascribed these extremes to El Niño–like condi-
tions in the Pacific (Aceituno et al. 2009;Hao et al. 2010;
Kiladis and Diaz 1986), as did Davis (2001),which
Kiladis and Diaz (1986) found to be comparable
in magnitude to the strong 1982/83 El Niño but with
stronger global impacts. To the best of our knowledge,
there appears to be no prior global-scale analysis and
attribution of the causes of the drought in the years
before, during, and after the 1877–78 El Niño.
In this study, we detail the characteristics and causes
of the multiyear global drought associated with the
Great Famine, herein referred to as the Great Drought,
with new datasets of hydroclimate and sea surface
temperatures (SST). We combine drought estimates
from four widely used tree-ring-based regional drought
atlases (Cook et al. 2010a,2015,2007;Palmer et al. 2015)
and rain gauge data from the Global Historical Clima-
tology Network (GHCN) (Lawrimore et al. 2011)to
characterize the spatial and temporal features of the
Great Drought and contextualize these features within
the ;140-yr instrumental record and ;800-yr paleo-
climate record. The drought atlases provide an annually
resolved estimation of hydroclimatic conditions, while
the rain gauge data include regions not covered in the
drought atlases and facilitate an examination of the
seasonal evolution of rainfall anomalies and potential
climatic drivers. With the further aid of SST datasets
and climate model simulations, we identify the climatic
conditions that shaped this dramatic multiyear event
across different regions, including conditions preced-
ing, during, and following the outsized El Niño event,
that extended the duration and sever ity of the Gre at
Drought in regions bordering the Atlantic and Indian
Oceans. An understanding of the characteristics and
causes of this event is the first step toward predicting the
occurrence and impacts of similarly widespread and
prolonged droughts, and their consequent impacts on
food security.
2. Materials and methods
a. Hydroclimate data
To characterize the hydroclimatic conditions of the
Great Drought, we employ instrumental records of
precipitation and tree-ring-based drought atlases. Monthly
rainfall data for land-based stations is from the exten-
sive GHCN database (Lawrimore et al. 2011), which
archives data for over 20 000 stations from multiple
sources around the world. The area-weighted average
monthly rainfall for Indian subregions and for the all-
India domain are from the Indian Institute of Tropical
Meteorology (IITM) (Parthasarathy et al. 1995,1993,
1994,1987) and are constructed from a uniformly dis-
tributed network of 306 stations across India with
data availability from 1871 to 2014. IITM defines these
subregions based on the similarity in their rainfall
characteristics [see Fig. 2 in Parthasarathy et al. (1995)
for a map of the subregions]. Rainfall for Fortaleza,
Brazil (3.48S, 38.38W), was accessed from the Joint
Institute for the Study of Atmosphere and Oceans
(JISAO), University of Washington (http://research. Rain-
fall for Shanghai, China (31.48N, 121.478E), and six sta-
tions (Durban, Brakfontein, Graaf Reinet, Somerset East,
Grahamstown, and Port Elizabeth) in the Eastern Cape and
KwaZulu-Natal provinces in South Africa were extracted
from the GHCN database. The stations in the Eastern
Cape–Natal region in South Africa were chosen based on
the availability of data for all years between 1875 and 1997.
For most of these stations, the GHCN records ended in 1997.
In addition to these direct rainfall records, we analyze
tree-ring-based reconstructions of the Palmer drought
severity index (PDSI) from four gridded drought atlases;
the Monsoon-Asia Drought Atlas, version 2 (MADA;
18318)(Cook et al. 2010a), the Old World Drought
Atlas (OWDA; 0.5830.58)(Cook et al. 2015), the North
American Drought Atlas (NADA; 0.5830.58)(Cook
et al. 2010b), and the eastern Australia and New Zea-
land Drought Atlas (ANZDA; 0.5830.58)(Palmer et al.
2015). These atlases extend the current instrumental
record to provide seasonal-scale hydroclimate informa-
tion back to 1500 C.E. in the ANZDA and 1200 C.E. or
longer in the Northern Hemisphere drought atlases. The
Northern Hemisphere drought atlases provide gridded
reconstructions of the boreal summer (June–August)
PDSI, whereas the ANZDA reflects the austral summer
(December–February) PDSI. PDSI is a widely used
measure of the severity of surface wetness or dryness. The
severity of dry conditions based on PDSI values are typi-
cally classified as abnormally dry (from 21.0 to 21.9),
moderate drought (from 22.0 to 22.9), severe drought
(from 23.0 to 23.9), extreme drought (from 24.0
to 24.9), and exceptional drought (,25.0). A similar
classification holds for wet conditions.
PDSI integrates moisture supply (i.e., precipitation)
and demand (i.e., evapotranspiration) over a year or
more and therefore the seasonal PDSI contains hydro-
climate information from the preceding seasons (Cook
et al. 2010a,2015,2007). For example, the reconstructed
June–August PDSI values in the western United States
are strongly influenced by precipitation and tempera-
ture in the preceding winter, during which the region
typically receives the largest fraction of precipitation
(Baek et al. 2017). These properties of PDSI also help
explain why there is not always a one-to-one relation-
ship between rainfall and the drought atlases used
here (e.g., over India, Europe, and eastern Australia). It
should be noted that few tree-ring chronologies from
India go into the MADA. The MADA tends to un-
derestimate the overall severity of the droughts in India
as indicated by the rainfall data. The complexity of the
rainfall patterns over India coupled with the somewhat
sparse tree-ring network used to produce the MADA
over India (Cook 2015;Cook et al. 2010a) probably
contribute to this apparent disparity. However, the re-
constructed drought in 1877 over northeast and penin-
sular India matches the instrumental data reasonably
well. This further supports the use of the MADA here
to complement the extensive rain gauge network that
covers the drought period. In the text, we refer to the
domain covered by the MADA as ‘‘Monsoon Asia.’’
b. Climate indices and SST data
We use monthly time series of sea level pressure
(SLP) at the Madras Observatory in India (1796–2000)
(Allan et al. 2002), the Niño-112, -3, -3.4, and -4 indices
(1870–present) (Trenberth and Stepaniak 2001), and the
Atlantic multidecadal oscillation (AMO) index (Enfield
et al. 2001) (1871–present) from the National Oceanic
and Atmospheric Administration Earth System Re-
search Laboratory’s Physical Climate Division (NOAA
ESRL PSD) database. The Niño indices are area-
weighted averages of the SST anomalies relative to the
1901–50 mean over the Niño-112(08–108S, 908–808W),
Niño-3 (58S–58N, 1508–908W), Niño-3.4 (58S–58N,
1708–1208W), and Niño-4 (58S–58N, 1608E–1508W) re-
gions. The AMO index is the unsmoothed, detrended,
area-weighted average SST over the North Atlantic
(08–708N). In addition, we use the monthly SLP re-
cord from Darwin, Australia (1866–present), which
is closely related to the Southern Oscillation index
(SOI), a measure of the pressure difference between the
Darwin and Tahiti stations and an indicator of the large-
scale El Niño–Southern Oscillation (ENSO) variability
(Trenberth et al. 2016). Trenberth et al. (2016) recom-
mend using the Darwin SLP instead of the SOI index
due to the lack of reliability of the Tahiti record prior
to 1935. To represent the Indian Ocean dipole (IOD),
we use the monthly dipole mode index (DMI; 1856–
present), calculated as the SST difference between the
western (508–708E, 108S–108N) and eastern (908–1108E,
108S–08N) equatorial Indian Ocean (Saji and Yamagata
2003), which is available from the Japan Agency for
Marine-Earth Science and Technology.
Global monthly SSTs are from the Extended Recon-
structed Sea Surface Temperature (ERSST) dataset, ver-
sion 5, which is available at a 28328spatial resolution
(Huang et al. 2015). In addition, Hadley Centre Global Sea
Ice and Sea Surface Temperature (HadISST), version 1.1
(Rayner et al. 2003), and Kaplan Extended SST, version 2
(Kaplan et al. 1998;Reynolds and Smith 1994), data are
used to quantify the uncertainty in the SST-derived
Niño indices. Gridded precipitation data (0.5830.58)
for 1900–present are from the Climatic Research Unit
(CRU) dataset, version 4.01 (Harris et al. 2014).
1DECEMBER 2018 S I N G H E T A L . 9447
Globally gridded (28328) surface temperature, hu-
midity, sea level pressure, and precipitation are from
the 56-member NOAA CIRES Twentieth Century Re-
analysis Project, version 2c (20CR). In addition to the
monthly means, the variability in the 20CR ensemble for
each variable is quantified using the standard deviations
between the 56 members. A comprehensive analysis of
the performance of 20CR precipitation against obser-
vations and other reanalysis products is provided in Lee
and Biasutti (2014), where it is shown that 20CR better
represents rainfall over tropical land and is comparable
to other reanalyses over the midlatitudes. Relevant to
this study, we compare the climatology of 20CR pre-
cipitation and teleconnection patterns with CRU (see
Fig. S1 in the online supplemental material). The main
climatological features of precipitation and its correla-
tions with the Niño-3.4, AMO, and DMI indices in
20CR are similar to CRU. Precipitation–Niño-3.4 cor-
relations are weaker over South Asia, precipitation–
AMO correlations are weaker over the Mediterranean
and stronger over southern Africa, and precipitation–
DMI correlations are weaker over Australia in 20CR
relative to CRU. However, the sign of these correlations
is consistent across all regions relevant to this study.
c. Drought characteristics
To evaluate the long-term context of the hydroclimate
conditions over Asia during the three years of the Great
Drought, we characterize the spatial extent and severity
of drought for each year in the MADA. The spatial
extent of the drought is defined as the fractional area
of this domain with abnormally dry conditions (i.e.,
PDSI #21.0). The drought severity is the area-
weighted average PDSI over the entire domain. We
limit our analysis to the period after 1200 C.E. during
which there is spatial coverage of PDSI across the
entire MADA domain (;108S–608N, 658–1508E) for
d. ENSO characteristics
In this study, we characterize El Niños based on the
area-weighted average SST anomalies over the Niño-3.4
region (Barnston et al. 1997). We examine the intensity
and duration of historical El Niño events to understand
their differing impacts. Their duration is defined as the
number of consecutive months with SST anomalies over
the Niño-3.4 region exceeding 0.58C, following the
NOAA Climate Prediction Center threshold. Their cu-
mulative intensity is defined as the sum of the monthly
Niño-3.4 SST anomalies over the entire duration of the
event. This metric is a combined measure of the strength
and duration of an event, both of which are important
for interpreting its regional impacts.
e. Climate model experiments
We use an ensemble of SST-forced simulations (1856–
2016) with the atmospheric component of the NCAR
CCM3 to examine the role of the tropical Pacific SST
conditions relative to global SST conditions in shaping
this multiyear drought. The first ensemble of simulations
involves lower boundary forcing from the observed
global SSTs, referred to as the global ocean–global at-
mosphere (GOGA) simulations. SSTs for these simu-
lations are blended from the Kaplan dataset (Kaplan
et al. 1998) used in the tropical Pacific (208N–208S) and
HadISST dataset (Rayner et al. 2003) used outside of
the tropical Pacific. The second ensemble of simulations,
referred to as the Pacific Ocean Global Atmosphere–
Mixed Layer (POGA-ML), only specifies SSTs in the
tropical Pacific from the Kaplan dataset; the SST
anomalies in other regions are computed using an ocean
mixed layer ocean model. Heat exchange between the
atmosphere and the ocean occurs at the surface based on
the computed energy fluxes from the atmosphere model,
allowing SST variations outside the tropical Pacific to be
forced by the tropical Pacific SSTs. Therefore, the cli-
mate response in the POGA-ML can be driven either
directly by the tropical Pacific or indirectly by remote
SST variations forced by the tropical Pacific. To capture
the role of internal atmospheric variability, each en-
semble has 16 members with identical boundary forcings
that only differ in the initial atmospheric conditions.
Additional details of these experiments are described in
Seager et al. (2005).
On comparison with CRU observations for the 1901–
50 baseline period, the CCM3 GOGA simulations cap-
ture the main spatial features of the observed annual
precipitation climatology over most regions except parts
of Asia (Figs. S1a,c). Over South Asia, the model does
not simulate the heavy precipitation center over central
India and along the Himalayas (Figs. S1a,c). In addi-
tion, the model has a wet bias over central China and a
dry bias over eastern China. Similarly, teleconnection
patterns with the three modes of variability—ENSO,
AMO, and IOD—are reasonably well represented in
the CCM3 GOGA simulations, supporting the use of the
model for this study (Fig. S1). Notable biases in regions
of relevance to this study include spatially varying biases
over South and East Asia in the Niño-3.4–precipitation
correlations, stronger than observed teleconnections in
the AMO–precipitation correlations over Europe, and
weaker than observed teleconnections in the DMI–
precipitation correlations over Australia.
To examine the precipitation responses associated
with tropical Pacific versus global SST conditions, we
compute standardized precipitation anomalies for each
ensemble member relative to the ensemble average
mean precipitation. For each region that experienced
drought conditions during the 1876–78 period, we cal-
culate the area-weighted average precipitation anoma-
lies over land during their major rainy seasons. The
significance of the differences between the distributions of
area-averaged precipitation anomalies from the two en-
sembles are calculated using the Kolmogorov–Smirnov
statistical test.
All anomalies for observed and modeled quantities are
calculated relative to a climatology evaluated over 1901–50.
3. Results
The Global Famine was initiated by severe droughts
in several regions that persisted for multiple seasons
between 1875 and 1878. In Fig. 1, we identify the tem-
poral evolution of these regional droughts. The drought
FIG. 1. Drought extent and SST evolution: (a) Monthly evolution of the Niño-3.4 index, AMO index, and the Indian Ocean DMI during the
Global Famine of 1876–78. The approximate beginnings and durations of dry conditions in major regions based on PDSI values (,21.0) or
seasonal rainfall anomalies (,21.0s), or a combination of both, are indicated by arrows and lines. Since PDSI from the regional drought atlases
is annually resolved, the durations of drought in the regions identified based on PDSI (i.e., East Asia, North Africa, and Southeast Asia) are
indicated for the 12-month period ending in the reconstruction season (i.e., September–August for MADA, OWDA, and NADA). Gray lines
indicate seasons outside the main rainy seasons. (b) Extent of dry conditions (colored regions) during each of the three years based on negative
PDSI (,21.0; brown) or low rainfall (,21.0s; pink) conditions. Gray areas highlight the extent of the drought atlases, and white areas indicate
absence of data. To identify the characteristics of dry conditions, PDSI values are from the regional drought atlases and rainfall anomalies
(relative to 1901–50) are from the GHCN database. Note that the dry regions depicted in (b) are approximate and for illustrative purposes. Some
subregions within the broader area depicted here might have differing conditions.
ECEMBER 2018 S I N G H E T A L . 9449
started in India with a failure of the 1875 winter mon-
soon season, and dry conditions persisted through the
summer of 1877. In East Asia, the drought started in
spring 1876, and the lack of rainfall persisted through
summer 1878. Subsequently, droughts developed in
parts of South Africa, northern Africa, and northeastern
Brazil in following seasons that lasted till at least 1878.
Relatively shorter but severe droughts also occurred in
western Africa, Southeast Asia, and Australia between
mid-1877 and 1878. Droughts in most of these regions
are often associated with the occurrence of El Niño
events (e.g., Kumar et al. 2006;Slingo and Annamalai
2000;Ropelewski and Halpert 1987;Wang et al. 2017;
Xu et al. 2004). While previous studies (Kiladis and Diaz
1986;Aceituno et al. 2009) have identified the pres-
ence of a strong El Niño during the Great Drought, the
El Niño conditions only developed in 1877 and waned
in 1878. However, the drought in key areas afflicted
by famines—including India, northeastern Brazil, and
China—started prior to the development of the El Niño
or lasted longer than its duration.
a. Character and historical context of the drought
Parts of India, East Asia, Central Asia, and Southeast
Asia simultaneously experienced abnormally dry con-
ditions (PDSI ,21.0) between 1876 and 1878, with the
peak spatial extent in 1877 (Fig. 2). East Asia, the region
with the highest number of reported famine deaths
(Davis 2001), witnessed the most widespread and per-
sistent droughts across all three years. The drought
was most extensive in South, central, and East Asia in
1877. In Southeast Asia, the drought was also the most
severe and widespread in 1877 and persisted in many
regions through 1878. In addition to Asia, moderate to
severe (PDSI ,22.0) drought conditions covered much
of northern Europe in 1876. In 1877, abnormally dry
conditions (PDSI ;21.0) occurred over parts of eastern
Australia and severe drought conditions (PDSI ,23.0)
occurred over California and the Mediterranean basin
including northern Africa and central Europe (Fig. 2b).
In 1878, these dry conditions intensified in the Mediter-
ranean basin. In addition, moderate to severe droughts
(PDSI ,22.0) spread across much of eastern Australia
in 1878 while much of the conterminous United States
experienced severe wet conditions (PDSI .3.0), both
typical of El Niño years (Fig. 2c). These regional droughts
were associated with substantial SST anomalies in mul-
tiple ocean basins. In the tropical Pacific Ocean, cool
SSTs in 1875/76 reversed to warm SSTs in 1876/77 that
strengthened in 1877/78 (Figs. 2a–c). In the Atlantic
Ocean, warm SSTs developed in the subtropics in 1867/77
and expanded into the tropics in 1877/78 (Figs. 2b,c). In
the Indian Ocean, warm SST anomalies developed across
the western and northern parts of the basin in 1877/78
(Fig. 2c).
To complement the drought atlas estimates and ex-
amine regions not covered by them, we analyze cumu-
lative 12-month (September–August) rainfall anomalies
at GHCN stations for 1875/76, 1876/77, and 1877/78
(Fig. 3). Rain gauge data from the nineteenth century
need to be treated with caution, and coverage is sparse
outside of India and parts of Europe and North Amer-
ica. Nonetheless, the GHCN station data largely confirm
the regional droughts identified in the drought atlases,
identify other droughts in regions not covered in the
drought atlases, and bring the drought in India into
sharp focus. Several stations in peninsular India re-
corded anomalously low rainfall exceeding 21.5 stan-
dard deviations (,21.5s) in 1875/76 (Fig. 3a). These
rainfall deficits intensified and spread across India in
1876/77. Strong rainfall deficits (,21.5s) also occurred
in parts of eastern Australia, southern Africa, the Bra-
zilian Nordeste, and the southwestern and northeastern
United States, and moderate deficits (,21.0s) at sta-
tions in Southeast and East Asia in 1876/77 (Fig. 3b).
Strong rainfall deficits (,21.5s) persisted in the Bra-
zilian Nordeste, Mediterranean, southern Africa, and
Southeast Asia in the following year (1877/78; Fig. 3c).
In contrast, rainfall anomalies over the United States
and western and peninsular India reversed from very
drytoverywetin1877/78(.1.5s). Differences be-
tween station-based rainfall estimates and PDSI-based
hydroclimatic conditions might exist because of the
misalignment between the seasons of PDSI recon-
struction and the September–August period used for
the cumulative rainfall anomalies. Despite that, these
station-based measurements, along with the tree-ring-
based hydroclimatic indicators, demonstrate severe
concurrent droughts across the tropics and subtropics
that persisted for multiple seasons within this 3-yr pe-
riod, implicating climate anomalies as a trigger for the
Global Famine.
To quantify the extreme, record-setting nature of this
drought, we examine key instrumental records dating
back to the 1870s for the rainy seasons of four regions
that experienced major economic or political transi-
tions following severe famines during 1876–78 (Fig. 4).
Across most of India, the summer (June–September)
monsoon season is the dominant source of rainfall
but the winter (October–December) monsoon season
contributes substantially to total annual rainfall over
peninsular India (Rajeevan et al. 2012). Following four
consecutive years of weak winter-monsoon rains since
1871, rainfall across India was extremely low (,21s)
for the four consecutive rainy seasons from the 1875
winter season to the 1877 summer season (Figs. 4a,b).
Rainfall deficits were extreme in 1876/77 when the
October–December rainfall (;21.5s), which was the
third lowest on record, was followed by the all-time re-
cord low summer monsoon rainfall (;23.1s) in 1877
(Figs. 4a,b). This record weakest summer monsoon is
consistent with the highest SLP ever recorded at the
Madras Observatory, which is an indicator of the strength
of the monsoon (Allan et al. 2002)(Fig. 4c). There is
limited station availability in China but Shanghai falls
within the region of persistent drought (Fig. 2). Although
summer is the main rainy season, rainfall in northeastern
China starts in spring, and El Niño impacts are found to
extend across the spring and fall seasons (Wang et al.
2017). Shanghai had below normal rainfall from spring
through fall (March–Novemb er) b etwe en 1876 and
1878. With rainfa ll anomalies below 22sin 1876 and
below 21.0sin 1877, 1876–77 in Shanghai had the
lowest 2-yr average rainfall on record. At Fortaleza in
the Brazilian Nordeste, where the main rainfall season is
February–May (Polzin and Hastenrath 2014), rainfall
was at least 1.5sbelow normal for three consecutive
rainy seasons during 1877–79 (Fig. 4e), the only 3-yr
FIG. 2. The Great Drought 1876–78: September–August (12 month) average SST anom-
alies during 1875/76, 1876/77, and 1877/78, and PDSI from the drought atlases for each of the
three years. MADA, NADA, and OWDA provide June–August PDSI, and ANZDA pro-
vides the December–February PDSI. Since PDSI integrates the moisture supply and demand
over preceding seasons, we provide average detrended SST anomalies for the 12-month
period from the preceding September to the concurrent August. SST anomalies are calcu-
lated relative to the 1901–50 baseline period.
ECEMBER 2018 S I N G H E T A L . 9451
period on record with persistently low rainfall, and 1877
had the strongest rainfall deficits (;22.0s) within the
1870–2010 record. In the Eastern Cape and Natal re-
gions in South Africa, the October–March rainy season
(Goddard and Graham 1999) rainfall was very high
(.1s) in 1874/75 and 1875/76 but very weak (,21.2s)
in 1876/77 and 1877/78, with this 2-yr average rainfall
(1876–78) being the fifth lowest between 1874 and 1995
(Fig. 4f).
Instrumental observations, tree-ring-based measure-
ments (Figs. 14), and historical documents all indi-
cate that the most severe, persistent, and widespread
impacts were in Asia (Davis 2001;Hao et al. 2010). We
therefore quantify the spatial extent (fraction of area
with PDSI ,21.0) and severity (area-average PDSI) of
drought across Monsoon Asia (see inset in Fig. 5a for
domain) within the ;800-yr-long MADA record (1205–
2012) from the MADA (Fig. 5). At its peak in 1877,
the spatial extent of the drought was 48%, ranking a
close second to the drought in 1495, which covered
49% of the domain (Fig. 5a). The 1877 drought was the
most severe over the same period and by quite a large
margin (Fig. 5b): area-weighted average PDSI for the
Monsoon Asia region in 1877 was approximately 21.0,
whereas no previous historical events exceeded 20.77.
Although the drought diminished in 1878 (area-average
FIG. 3. Global rainfall anomalies during the Great Drought: Standardized anomalies of
12-month (September–August) cumulative rainfall at available GHCN stations during the
three periods: (a) 1875/76, (b) 1876/77, and (c) 1877/78. Anomalies are with reference to the
1901–50 climatology.
PDSI ;20.25), 32% of Monsoon Asia remained in
drought (PDSI ,21.0). At least 25% of Monsoon Asia
experienced droughts in all three years, with the 3-yr
average ranking 31st highest in extent and 12th in se-
verity during this 800-yr period.
Together, these multiple sources of data provide
quantitative evidence of a severe, global-scale, multi-
year drought between 1875 and 1878 associated with
record-setting droughts in several regions, particularly
in Asia, where it was an extreme 3-yr drought and the
central year of 1877 was the worst single-year drought in
the last 800 years.
b. Spatiotemporal characteristics of the drought in
Within the regions impacted by the Great Drought,
India is unique for its dense network of rain gauge ob-
servations that extend back into the nineteenth cen-
tury, and droughts in India have a close relationship with
Pacific SST conditions. Using monthly, area-averaged
rain gauge data across homogenous rainfall regions in
India from the IITM, we examine the characteristics of
the drought in further detail. Figure 6 shows that the
Great Drought started with dry conditions in peninsular
India in winter 1875. Note that the winter monsoon
typically brings ;30%–60% of the annual rainfall to
peninsular India (Rajeevan et al. 2012)(Fig. 6f). Fol-
lowing weak winter rainfall in late 1875, the all-India
rainfall (AIR) was anomalously low for most months
(except July and September) in 1876 (Fig. 6a), partic-
ularly during the late monsoon and following early
winter season. The winter season anomalies were par-
ticularly extreme in peninsular India, which experi-
enced four consecutive months of near-record lows from
September to December 1876, coinciding with the start
of the famine in India (Fig. 6f). Intensifying the drought,
multiple Indian subregions including the northwest,
west central, and central northeast experienced con-
secutive near-record low rainfall during the 1877 mon-
soon months (Figs. 6b–e), consistent with the developing
FIG. 4. Severity of rainfall anomalies: Time series of (a) standardized anomalies of summer (June–September)
monsoon rainfall for the all-India region, (b) standardized anomalies of winter (October–December) monsoon
rainfall for the all-India region, (c) sea level pressure at Madras, India, (d) standardized anomalies of March–
November rainfall at Shanghai, China, (e) standardized anomalies of February–May rainfall at Fortaleza in the
Brazilian Nordeste, and (f) standardized anomalies of October–March rainfall in eastern South Africa (average of 6
stations in the Eastern Cape and KwaZulu-Natal provinces). The four red dots highlight years from 1875 to 1878,
and the horizontal red line indicates the magnitude of the peak anomalies within this 4-yr period for reference. The
length of the Madras SLP record is limited by the length of the available time series, and the South African rainfall
time series is limited by the unavailability of data at multiple stations in the GHCN database post-1997. Rainfall for
the 1997/98 rainy seasons at Fortaleza is missing in the record. Gaps in any of the records represent missing values.
Anomalies are calculated from the 1901–50 baseline period.
ECEMBER 2018 S I N G H E T A L . 9453
El Niño(Fig. 1)(Kumar et al. 2006;Pokhrel et al. 2012;
Ihara et al. 2008). Although individual months have re-
corded lower rainfall in some subregions, the consistently
low all-India rainfall during the 1877 peak monsoon
season (Singh et al. 2014;Pai et al. 2016) is unsurpassed.
Notably, 1876–77 is one of only two consecutive 2-yr
periods with annual rainfall anomalies persistently lower
than 21.0s(21.5sin 1876 and 21.8sin 1877), the other
being 1904–05 although it had weaker anomalies. Rains
over peninsular India recovered in the 1877 winter and
were followed by very wet conditions in 1878 over all
subregions except the central northeast (Fig. 6). These
1876–77 rainfall failures across many subregions of India,
with its monsoon-dependent agriculture, contributed to
the severe food shortage and ensuing famine in India that
started in peninsular India in 1876. Subsequently, the
extremely wet conditions in late 1877 and 1878 across
India led to substantial loss of lives by facilitating the
spread of infectious diseases in a famine-weakened pop-
ulation (Whitcombe 1993).
c. Natural climate variability: El Niño and beyond
Pan-tropical rainfall failures, such as occurred in 1877,
are often caused by the warm phase of ENSO (Lyon and
Barnston 2005). However, the precise impacts of an El
Niño depend on the timing, duration, and location of
FIG. 5. Monsoon Asia drought severity and extent: Time series of (a) the fraction of the
Monsoon Asia region (inset) experiencing drought (PDSI ,21.0), expressed as a percent,
and (b) the area-weighted average PDSI across the region from 1205 to 2012. The drought
severity and extent for each year of the drought are indicated in the corresponding panels.
Dashed lines indicate the magnitude of these characteristics during the peak of the drought in
1877. The 1877 drought extent is the second highest and drought severity is the strongest since
the early 1200s.
peak SST anomalies (Kumar et al. 1999,2006;Ihara
et al. 2008). While previous studies of the 1876–78 pan-
tropical drought attributed the blame to an El Niño
(Aceituno et al. 2009;Hao et al. 2010;Kiladis and Diaz
1986), the reasons for the associated record severity
of impacts in multiple regions have not yet been de-
termined. Further, the other climate factors that con-
tributed to the prolonged multiyear drought conditions
before and after the El Niño are largely unexplored.
Here, we examine the spatiotemporal features of the El
Niño and identify the extraordinary sequence of SST
configurations in three major ocean basins that led to
this multiyear, global-scale extreme event (Fig. 1).
The 1877–78 El Niño was associated with the warmest
annual-mean SSTs in the central Pacific (Niño-3.4 re-
gion) between 1870 and 2015 (Fig. 7a), consistent with
the extreme impacts experienced across multiple
regions and seasons (Figs. 13). The record high mag-
nitude of the Niño-3.4 anomalies in the instrumen-
tal record is consistent across multiple SST datasets
(Fig. S2a). The annual-mean SLP at Darwin, Australia,
another indicator of the strength of El Niños, suggests
that the 1877 event was the fourth strongest between
1866 and 2015 (Fig. 7b). Although they differ in their
estimates of the extremeness of the event due to the
varying spatial signatures of different flavors of
El Niños, both indicators suggest an extreme El Niño
event. In addition to the extremely strong El Niño, we
identify three other extreme or record-setting condi-
tions that are responsible for the multiyear duration of
this drought (Fig. 7). First, we find that anomalously cool
tropical Pacific conditions preceded the El Niño and
initiated droughts in some regions. Second, the North
Atlantic was anomalously warm in 1877–79, with a peak
following the peak of the El Niño(Fig. 7c). Third, a
positive IOD event (Saji et al. 1999) developed in the
latter half of 1877 along with the developing El Niño
(Fig. 7d). These SST conditions that subsequently de-
veloped in the tropical Indian and Atlantic Oceans were
extreme versions of their typical responses to El Niño
(Alexander et al. 2002;Enfield and Mayer 1997;Elliott
et al. 2001) and were crucial in shaping the overall multi-
year drought in these different regions and seasons (Fig. 1).
During 1870–76, the central tropical Pacific was in a
prolonged cool phase for 7 years (Fig. 7a). This was as-
sociated with persistent and severe droughts in the west-
ern United States and much of Europe apart from the
British Isles (Herweijer and Seager 2008) and persistently
weak all-India winter rainfall (October–December) from
1871 to 1876 (Fig. 4b), consistent with the suppression of
the winter monsoon during La Niñayears(Rajeevan et al.
2012). The largest negative rainfall anomalies coincided
FIG. 6. Temporal evolution of drought over India: Rainfall for the all-India domain and five subregions defined by the IITM basedon the
similarity in their rainfall characteristics [see Fig. 2 in Parthasarathy et al. (1995)]. Colored lines highlight the evolution of drought across
different regions between 1875 and 1878. Gray colors show all other years between 1870 and 2013.
ECEMBER 2018 S I N G H E T A L . 9455
with the strongest negative SST anomalies in 1875/76, and
the start of the Great Drought in India in winter 1875.
These low or negative SST anomalies (,0.28C) persisted
for ;80 consecutive months, the longest cool period on
record between 1870 and the present (Fig. 7a).
Coincident with the developing record El Niño was an
unsurpassed positive IOD event in late 1877, with warm
anomalies in the Somali Current region and cool SSTs
off the western Australia coast (Fig. 7d; see also
Figs. S2b–d). The DMI, a measure of this gradient across
the Indian Ocean that typically peaks following the
monsoon season, was the strongest on record (Fig. 7d).
The very weak monsoon circulation associated with the
extreme 1877 summer monsoon rainfall failure resulted
in weak summertime cooling of the western Indian
Ocean by upwelling and evaporation, which likely led to
warmer SSTs in the region and the development of the
positive IOD configuration. Positive IOD events that
develop and peak during the monsoon season tend to
enhance rainfall over the subcontinent but typical IOD
events, such as the 1877 event, that develop and peak
in the post-monsoon season (September–November)
are normally associated with relatively weaker Indian
monsoon rainfall (Anil et al. 2016). To analyze the
regional precipitation impacts of these individual and
co-occurring conditions, we compare the composite
July–December 20CR rainfall patterns during years
(excluding 1877) with the following three conditions:
1) El Niño events without positive IOD events, 2) positive
IOD events in the absence of El Niño events, and 3) co-
occurring IOD and El Niño events (Fig. 8). The selected
season coincides with the typical cycle of positive IOD
events. We find that positive IOD conditions amplify the
drying effect of El Niños over parts of Southeast Asia,
eastern Australia, and southern Africa (Figs. 8a–c), con-
sistent with previous studies (Ummenhofer et al. 2013;
FIG. 7. ENSO and SST features: Long-term detrended time series of the (a) Niño-3.4 index,
(b) Darwin SLP anomalies (hPa), (c) AMO index, and (d) Indian Ocean DMI. Dashed red
lines indicate the peak magnitude of the index between 1876 and 1878. Vertical blue lines in
(a) highlight the four longest prolonged cool periods in the equatorial Pacific, which are
defined as consecutive months with Niño-3.4 SST anomalies consistently below 0.28C.
Anomalies are calculated from the 1901–50 baseline period.
FIG. 8. Influence of different ocean basins: Composite standardized rainfall anomalies for July–
December from 20CR during (a) years with developing El Niño events without positive IOD events,
(b) years with positive IOD events without developing El Niño events, and (c) years with positive IOD and
developing El Niño events. July–December is selected to coincide with the cycle of positive IOD events.
Also shown are composite average rainfall anomalies for February–May during (d) years with strong El
Niño events during nonpositive (neutral or negative) AMO events, (e) years with positive AMO events
without El Niño events, and (f) years with strong El Niño events during positive AMO phases. February–
May is selected to coincide with the peak rainy season over Nordeste Brazil and the peak AMO following
a strong El Niño event. El Niño events are selected based on the November–March Niño-3.4 index ex-
ceeding 1.0s. (Note that years in the left column indicate years of developing El Niño events.) Positive
AMO events are selected based on February–May AMO index exceeding 1.0s. Positive IOD events are
selected based on the July–December DMI index .1.0s. Standardized rainfall anomalies are calculated
based on the mean and standard deviation sof the baseline period 1901–50.
ECEMBER 2018 S I N G H E T A L . 9457
Goddard and Graham 1999;Cai et al. 2011,2009;Ashok
et al. 2003). All years when IOD events occurred along
with developing El Niño events had severe rainfall
deficits in these regions. While positive IOD conditions
enhance rainfall over South Asia in the absence of El
Niño events, rainfall is relatively suppressed when pos-
itive IOD events occur with an El Niño. Furthermore,
although central Asia does not show a robust rainfall
response during either IOD or El Niño events, severe
rainfall deficits occur across central Asia during all 5 years
with co-occurring IOD and El Niñoevents(Fig. 8c). This
suggests that the observed severity of rainfall deficits in
these regions during 1877 is likely associated with the
simultaneous occurrence of a record strong positive IOD
and El Niño(Fig. 7). Basinwide warming of the Indian
Ocean followed in 1878 (Fig. 2c), which reduces the
drying impacts of the continuing warm tropical Pacific
SSTs on the 1878 India summer monsoon (Ihara et al. 2008).
Atlantic SSTs north of the equator were anomalously
warm in 1877, 1878, and 1879, shifting the intertropical
convergence zone (ITCZ) northward and causing three
consecutive dry rainy seasons over the Nordeste region
(Hastenrath and Greischar 1993;Uvo et al. 1998;Lucena
et al. 2011)(Fig. 7c). The AMO index, which represents
SSTs in the North Atlantic (Schlesinger and Ramankutty
1994;Enfield et al. 2001), peaked three months after the
peak of the El Niño to a record high magnitude between
1870 and 2015 (Fig. 1a). The severity of the impacts in the
Brazilian Nordeste in 1878 is consistent with the warmest
North Atlantic SSTs in that year (Fig. 7b). To evaluate
the individual influence of these conditions on regional
rainfall anomalies, we compare the composite February–
May 20CR rainfall patterns during years (excluding 1877/
78) with the following sets of conditions: 1) El Niño events
during a neutral or cold phase of the AMO, 2) extreme
positive AMO events in the absence of El Niños, and 3) El
Niño events during a warm phase of the AMO (Figs. 8d–f).
The February–May season coincides with the main rainy
season in the Brazilian Nordeste and the peak warming in
the North Atlantic following El Niños. El Niño events that
occur in the cold or neutral AMO phase have a drying
effect over northern Brazil although the impacts do not
consistently extend into the Nordeste region (Fig. 8d).
However, all six historical events with El Niño coinciding
with warm AMO phases have more severe and widespread
drying across northern and northeastern Brazil (Figs. 8d–
f), underscoring the importance of their combined occur-
rence in shaping the 1876–78 drought in this region.
d. Role of tropical Pacific versus global SST forcing
To examine the relative role of the tropical Pacific in-
cluding the 1877–78 El Niño relative to global SST anom-
alies unrelated to the tropical Pacific in driving regional
precipitation anomalies, we compare precipitation anoma-
lies from the 16-member GOGA and POGA-ML ensem-
bles of climate simulations with the NCAR CCM3 (see
section 2e). A comparison of these ensembles highlights the
importance of SST anomalies outside of, and not forced by,
the tropical Pacific that could aid in the predictability of
future occurrences of a similar event.
For these comparisons, precipitation anomalies are
calculated for the major rainy seasons in each region
that experienced dry conditions: June–September sum-
mer and October–December winter monsoons in India,
March–November in northeast China, October–March
in South Africa and eastern Australia, January–March
in the Mediterranean region, and February–May in
Nordeste Brazil (Fig. 9). Although the ensemble mean
response does not simulate the rainfall deficits during
the 1875 boreal winter monsoon in India (October–
December), it does correctly simulate anomalously dry
conditions in all other regions, albeit with lower mag-
nitudes than observed in some regions. The simulated
mean response of both ensembles shows a similar range
of negative rainfall anomalies during the 1877 boreal
summer (July–September) monsoon season over India,
the 1877 boreal spring–fall (March–November) sea-
son in northeastern China, and the 1877/78 austral
spring–summer seasons (October–March) in Australia
(Fig. 9) when these regions experienced the most severe
dry conditions. In these regions, the distributions of
precipitation anomalies from the two ensembles are
indistinguishable at the 5% significance level, suggesting
that the rainfall deficits during 1875–77 are largely
forced by tropical Pacific SSTs or by SST anomalies in
other regions that were forced by the tropical Pacific.
This SST forcing includes the strong 1875/76 La Niña,
the strong 1877–78 El Niño, and the 1877 positive IOD
conditions. While the negative rainfall anomalies during
October–December in India and March–November in
East Asia are comparable, they are relatively smaller in
magnitude to observations. Perhaps internal variability is
driving the severity of these rainfall deficits, but more likely
differences from observations are due to model deficiencies
in accurately simulating the Asian monsoon rainfall and its
teleconnections with natural modes of variability (Hurrell
1995), as discussed in section 2e (Fig. S1).
In contrast, negative rainfall anomalies during the 1877
and 1878 rainy seasons (February–May) over Nordeste
Brazil (Figs. 9c,d), the 1877/78 austral summer season
(October–March) in South Africa (Fig. 9g), and the 1878
winter rainy season (January–March) in the Mediterra-
nean basin (Fig. 9h) cannot be attributed to the tropi-
cal Pacific forcing alone. For these regions, the GOGA
and POGA-ML ensembles simulate significantly dif-
ferent distributions of rainfall anomalies with opposite
FIG. 9. Role of tropical Pacific vs global oceans: (a)–(h) Standardized rainfall anomalies in years associated with extreme deficits in
different regions (see red boxes on map) from the 16-member ensemble of (black) GOGA and (red) POGA-ML simulations. For each
region, seasons are chosen to coincide with the local main rainy seasons. Observed average rainfall anomalies for each region are indicated
by blue diamonds. Also shown are average February–May North Atlantic (08–708N, 808W–08) surface temperature (TS) anomalies in
(i) 1877 and (j) 1878, and (k) the average July–December TS gradient between the western (108S–108N, 508–708E) and eastern (108S–08,
908–1008E) equatorial Indian Ocean in 1877 in both simulations. No blue dots to represent observations are included in (i) and (j) since TS
from the GOGA simulations closely track the observed SSTs that are used as boundary conditionsfor the model. In the box-and-whisker
plots, the boxes represent the 25th–75th percentiles, and whiskers represent the 5th–95th percentiles of the 16-member ensembles.
Numbers in the top left indicate pvalues from the Kolmogorov–Smirnov test for difference in distributions. Low pvalues indicate that the
distributions are significantly different. All anomalies are calculated relative to the 1901–50 climatology.
ECEMBER 2018 S I N G H E T A L . 9459
or weaker mean rainfall responses in the latter. This
suggests that the regional rainfall anomalies during
1877/78 are associated with SST variations outside the
tropical Pacific that are unrelated to the El Niño, such as
the North Atlantic extremely warm SSTs, or are not
fully captured by the POGA-ML model, such as the late
1877 IOD even though that is likely a response to the El
Niño. POGA-ML simulates near-zero average SST
anomalies in the North Atlantic in spring 1877 although
observed SST anomalies were positive (Fig. 9i). POGA-
ML does simulate the warm North Atlantic response in
1878 to the strong El Niño, which occurs via atmo-
spheric teleconnections and induced surface heat
flux anomalies (Alexander et al. 2002;Enfield and
Mayer 1997;Elliott et al. 2001) although the observed
anomalies were significantly stronger (Fig. 9j). These
differences in precipitation and SST anomalies suggest
that the Brazil Nordeste rainfall deficits are intensified
by the warm SST anomalies in the North Atlantic. Only
in 1878 are these likely primarily a response to the El
Niño. POGA-ML substantially underestimates the IOD
response (Fig. 9k) because of the lack of ocean dynamics
in the model configuration (Meyers et al. 2007). Con-
sequently, greater and more robustly simulated pre-
cipitation drops over South Africa in 1877 in GOGA
than in POGA-ML could be attributed to the correct
IOD state and magnitude in GOGA.
e. Comparison of the 1877–78 El Niño with other
strong Niño events
We have shown that the occurrence of the record warm
North Atlantic and the strongest positive IOD event
amplified the drying effects of El Niño events in several
regions. Such conditions in the North Atlantic and Indian
Oceans are not always linked to El Niño events (Meyers
et al. 2007). The correlation between the Niño-3.4 and
DMI indices is ;0.5 (pvalue 0.01) and between the
Niño-3.4 and AMO indices is ;0.35 (pvalue 0.01),
suggesting that their covariability is rare. A similar se-
quence of extreme conditions in these three basins has only
occurred one other time in the instrumental record in 1997/
98. While the annual mean SST signal of the 1877–78 El
Niño was comparable to the El Niño of 1997/98 (Fig. 7a),
the impacts were far more severe in many regions in 1877–
78. We identify four main differences in the spatiotemporal
features of these events that explain the differing regional
precipitation impacts between these events (Figs. 1012).
First, the 1877–78 event was stronger and longer
lasting than other notable El Niños, covering two sum-
mer monsoon seasons. SST anomalies exceeding 0.58C
in the Niño-3.4 region lasted for 16 consecutive months
during the 1876–78 period, 3 months longer than in 1997/
98 and 2 months longer than in 1982/83 (Fig. 10a). The
cumulative intensity of the 1877–78 event also exceeds
all other El Niños between 1870 and 2013 (Fig. 10a).
Second, although 1997/98 was the warmest of the three
events in all Niño regions during the monsoon seasons,
the largest warm anomalies were in the Niño-112 re-
gion, the far eastern equatorial Pacific (Figs. 11a–d).
Although the 1877/78 and 1997/98 events were similarly
warm in central to eastern equatorial Pacific during the
monsoon season, the regions of peak anomalies in sum-
mer 1877 were west of those in 1997 (Fig. 11e). The lo-
cation of the peak anomalies is relevant to understanding
FIG. 10. SST characteristics: (a) Cumulative intensity and duration characteristics of all El Niño events since 1870.
Cumulative intensity is calculated as the sum of the monthly temperature anomalies over the duration of the El
Niño event and duration is the number of consecutive months with Niño-3.4 anomalies exceeding 0.58C.
(b) Comparison of the magnitude of the annual mean Niño-3.4 SST anomalies and the July–December seasonal
mean DMI for all years between 1870 and 2015.
FIG. 11. Features of extreme El Niño events: Comparison of the temporal evolution of area-average, detrended SST anomalies (relative
to the 1901–50 climatological mean) over the (a) Niño-112, (b) Niño-3, (c) Niño-3.4, and (d) Niño-4 regions, during the three most
extreme El Nino events: 1877–78, 1982/83, and 1997/98. (e) Global SST anomalies during the monsoon season (June–September) in these
three years. The black contour line indicates the 288C isotherm.
ECEMBER 2018 S I N G H E T A L . 9461
the regional impacts of the individual events. For in-
stance, a stronger drought in South Asia in 1877 than
in 1997 is consistent with previous modeling work
showing a higher likelihood of westward-shifted than
eastward-shifted El Niño events to produce subsidence
and drought over the region (Kumar et al. 2006). Third,
the 1997/98 El Niño developed rapidly in the eastern
Pacific (Niño-112 region) starting in February whereas
FIG. 12. Comparison of 1877–78 and 1997/98 El Niño events: (a),(b) Evolution of Niño-3.4, AMO, and IOD
indices over the duration of each event, (c)–(f) PDSI and 12-month (September–August) average detrended SST
anomalies during the developing and decaying years of the El Niños. (g),(h) Moist static energy and (i),(j) surface
pressure anomalies for the summer monsoon season. Standardized anomalies are calculated using the mean and
standard deviation of the 1901–50 baseline period. Dots represent regions where anomalies are not significant.
Significance is calculated based on the spread (61s
) of a variable in each season exceeding its spread (61s
) over
the climatological period, where s
is the standard deviation between the 56 ensemble members of 20CR averaged
to the seasonal scale.
the 1877 event development in this region started in
June (Fig. 11a). This early and rapid development of
the 1997/98 El Niño likely contributed to the early
basinwide warming of the Indian Ocean (Fig. 11e) that
enhanced the moisture availability and weakened the
suppression of convection typical of El Niño events.
The IOD event during 1997 was weaker than during the
1877–78 El Niño event (Fig. 10b) and peaked later
(Figs. 12a,b). Consequently, rainfall over India was near
normal in 1997 (Ihara et al. 2008) compared to the
greater suppression of rainfall over India in 1877 that
arose from the enhanced tropospheric stability associ-
ated with a warm tropical Pacific and a relatively cooler
Indian Ocean (Ihara et al. 2008). Fourth, the North At-
lantic warming following the 1997/98 event peaked in the
following summer season rather than in the spring as in
1877–78 when it was able to suppress the main rainy
season over northeastern Brazil (Figs. 12a,b). Finally, the
North Atlantic was warm in1877 and 1878, which worked
to weaken both rainy seasons in northeastern Brazil
whereas the North Atlantic was relatively cooler in 1997
leading up to the 1997/98 El Niñoevent(Figs. 12a,b).
The 1877–78 event had substantially more severe and
widespread drought conditions across Asia, northern
Africa, and parts of Europe than the 1997/98 event,
and opposite hydroclimatic conditions over Australia
(Figs. 12c–f). During the 1877 summer monsoon season,
the westward-shifted peak SST anomalies in the Pacific
led to peak positive moist static energy (MSE) anoma-
lies and hence convection, occurring in the western-
central Pacific, farther west than both typical El Niño
events and the 1997/98 event (Figs. 12g,h). Conse-
quently, the surface high pressure anomalies (.2s) over
the Indian continent were substantially larger during the
summer monsoon in the developing phase of the 1877 El
Niño than during 1997 (Figs. 12i,j), leading to a greater
weakening of the MSE (,21.5s) over the peak region
of the Indian monsoon circulation (Boos and Kuang
2010;Cane 2010). Combined with its enormous magni-
tude, this particular SST anomaly pattern in 1877 was
also associated with anomalously high surface pressure
(.2s) across much of central, northern, and eastern Asia
and the Maritime Continent during the summer monsoon
season, substantially larger and more widespread than in
1997 (Figs. 12i,j). Accordingly, these regions experienced
stronger suppression of MSE (,21.5s) and more ex-
treme precipitation deficits in 1877 than in 1997. The
exception is Indonesia, which had stronger drought con-
ditions in 1997. In eastern Australia, drought was severe
and widespread in 1877–78 but largely concentrated in
southeastern Australia in 1997/98 with wetter conditions
across the rest of the region. These differences were as-
sociated with the stronger positive IOD event in 1877 and
the substantially cooler temperatures off the northern and
western coast of Australia, which lead to greater sup-
pression of moisture availability, MSE, and precipitation
across a large part of eastern Australia (Figs. 12c,d,g,f).
4. Discussion and conclusions
Our analysis leads to three main findings. First,
multiple sources of data reveal an intense, global-
scale drought affecting many tropical and subtropical
regions simultaneously between 1875 and 1878, with
record-setting conditions in Asia where there were the
highest number of reported famine victims (Davis 2001).
While single-year droughts might not have similarly se-
vere societal impacts, these severe and prolonged climatic
conditions undoubtedly initiated the Global Famine cri-
sis. Second, this event was associated with the strongest El
Niño event in the instrumental record, which followed
the longest cool period in the tropical Pacific, and
whose early evolution, long duration, and cumulative
intensity relative to other strong El Niños accounts for
the severity of its global impacts. The magnitude of the
1877–78 El Niño SST anomalies was likely more extreme
than in the reconstructed datasets: the paucity of SST
observations in the tropical Pacific in the late nineteenth
century can only lead to underestimating its strength
(Kaplan et al. 1998). Third, this multiyear, global-scale
extreme event was largely orchestrated by the tropical
Pacific via direct atmospheric teleconnections and then
indirectly by influencing pan-tropical SSTs that ad-
ditionally drove the regional droughts. Record warm
conditions in the North Atlantic in 1878 and the re-
cord positive IOD conditions in 1877 resulting from the
cascading influence of this powerful El Niñowerecritical
in shaping its regional impacts, particularly on Nordeste
Brazil, northern and southern Africa, and eastern Aus-
tralia, during and after the 1877 El Niño. However, the
independently warm North Atlantic in 1877 aided the
development of drought in Nordeste Brazil prior to
the evolution of the El Niño.
While data coverage in 1877 was sparse in the Pacific
basin, with availability only at a few points in the central
Pacific, the extreme magnitude of this event, which is
corroborated by multiple reconstructed SST datasets
(Fig. S2), has little uncertainty. Extensive tests conducted
by Kaplan et al. (1998), where the input data coverage for
the reconstructed SST product was withheld to the cov-
erage in the 1870s, shows that this change in coverage does
not substantially influence the magnitude of reconstructed
SSTs in the tropical Pacific, particularly during notable El
Niño events. The Indian and Atlantic Oceans had much
greater data coverage than the Pacific and consequently
lower errors and uncertainties for this time period.
1DECEMBER 2018 S I N G H E T A L . 9463
Exacerbated by prevailing social conditions, famines
followed the occurrence of severe droughts across the
world (Davis 2001). In India, despite agricultural losses
associated with the drought, British colonialists col-
lected harsh taxes, hoarded and exported grain from
India to England, and destroyed common resources that
traditionally buffered societies from climate variability
(Davis 2001;Meena 2015). Food shortages beginning in
1875 depleted local reserves, and high prices made food
inaccessible to the starving local population, who were
ultimately denied labor for being weak (Davis 2001). In
northern China, disruption of the agrarian societies by
imperialist forces and a dysfunctional transportation
system that made relief hard to access led to widespread
death and depopulation of vast communities starting in
1877, following a year of drought (Davis 2001). In the
Brazilian Nordeste, the Great Drought devastated
the cotton and cattle raising important to the regional
economy and subsistence farmers alike. Initially, the
people of the Sert
ao remained but with starvation
spreading and the drought persisting out-migration fol-
lowed, creating social instability across a wider region
(Greenfield 1992;Davis 2001). As in India, the official
response was to create work camps and exchange aid for
labor. Smallpox broke out in the camps, greatly in-
creasing the mortality (Davis 2001). By the end of the
Great Drought in 1880 up to one million were dead, and
it is claimed the Nordeste never fully recovered (Cuniff
1970). In Algeria and Morocco, the drought and failed
crops forced peasants to sell their wealth, cattle and
sheep, for export to France, further impoverishing the
population. As in India and the Nordeste, out-migration
soon followed with concentrations of migrants lead-
ing to cholera and typhoid and increased mortality
(Davis 2001).
The Great Drought and the Global Famine cast a long
shadow on the politics and economy across the tropics.
The demographic disruption cast by the famines often
lasted for generations: in the Chinese province of Shanxi,
for example, it took until 1953 to regain 1875 population
levels (Davis 2001). The decimation of agricultural
workforces, along with the destruction of local means of
production (in northern China starving peasants actually
ate their homes, constructed of sorghum stalks), pros-
trated traditional Asian and African societies in the face
of the colonizing wave of the late nineteenth century.
Starvation among the African population facilitated the
French colonial expansion in North Africa and the even-
tual British defeat of the famine-weakened Zulu Nation in
summer 1879 [see Davis (2001) and references therein].
In a very real sense, the El Niño and climate events of
1876–78 helped create the global inequalities that would
later be characterized as ‘‘first world’’ and ‘‘third world.’’
The severe and widespread 1876–78 drought in mul-
tiple grain-producing regions of the world was induced
by natural SST variability. Therefore, such a global-scale
event might happen again. With the projected intensi-
fication of El Niño–induced hydroclimate anomalies
due to rising greenhouse gas concentrations and global
warming (Seager et al. 2012;Cai et al. 2014), such
widespread droughts could become even more severe.
While the sociopolitical factors that translated the Great
Drought into unprecedented famine (Davis 2001) do not
exist in the current world, such extreme events would
still lead to severe shocks to the global food system with
local food insecurity in vulnerable countries potentially
amplified by today’s highly connected global food trade
network (Puma et al. 2015). Continued improvements
in understanding why this event, and the coupled atmo-
sphere–ocean processes it induced across the tropics, led
to such a devastating global drought should translate into
improved prediction of the consequences of any such
future event and allow effective management of the re-
sulting food crises, so that the next Great Drought does
not trigger another Great Famine.
Acknowledgments. General: We thank the National
Oceanic and Atmospheric Administration (NOAA) and
the Indian Institute of Tropical Meteorology (IITM)
for archiving and enabling public access to their data.
ERSSTv4 data, Twentieth Century Reanalysis version 2c,
and time series of all indices were provided by NOAA
ESRL PSD, Boulder, Colorado. GHCN data were pro-
vided by NOAA’s National Centers for Environmen-
tal Information (NCEI). Support for the NOAA-CIRES
Twentieth Century Reanalysis Project version 2c dataset
is provided by the U.S. Department of Energy, Office of
Science Biological and Environmental Research (BER),
and by the National Oceanic and Atmospheric Admin-
istration Climate Program Office. The rainfall time series
for India, which are derived from the original rainfall
records maintained by the Indian Meteorological De-
partment, were provided by IITM. We acknowledge
NCEI for maintaining the World Data Center for Pa-
leoclimatology archives. We also thank Alexey Kaplan
for assistance with the Dipole Mode Index data and Hun
Baek for valuable discussions on the AMO.
Funding: D. Singh was supported by the Lamont-
Doherty Earth Observatory Fellowship. R. Seager and
M. A. Cane were supported by NSF OCE1657209. B. I.
Cook was supported by the NASA Modeling, Analysis,
and Prediction program. M. Ting was supported by
NOAA Grant NA14OAR4310223 and NSF Grant AGS
16-07348. We acknowledge the support of the World
Surf League P.U.R.E. and the Center for Climate and
Life at Lamont-Doherty Earth Observatory.
Author contributions: DS designed the analysis, ana-
lyzed the data, performed the research, and wrote the
paper. RS, MC, BIC, and MT contributed to the designof
the study. All authors contributed to the interpretation of
results and writing of the manuscript.
Competing interests: The authors declare no competing
Data and materials availability: All data used in the
analysis are from publicly available datasets for which
sources have been provided in section 2.
Aceituno, P., M. del Rosario Prieto, M. E. Solari, A. Martínez,
G. Poveda, and M. Falvey, 2009: The 1877–1878 El Niño ep-
isode: Associated impacts in South America. Climatic Change,
92, 389–416,
Alexander, M. A., I. Bladé, M. Newman, J. R. Lanzante, N.-C. Lau,
and J. D. Scott, 2002: The atmospheric bridge: The influence
of ENSO teleconnections on air–sea interaction over the
global oceans. J. Climate,15, 2205–2231,
Allan, R. J., C. J. C. Reason, P. Carroll, and P. D. Jones, 2002: A
reconstruction of Madras (Chennai) mean sea-level pressure
using instrumental records from the late 18th and early 19th
centuries. Int. J. Climatol.,22, 1119–1142,
Anil, N., M. R. Ramesh Kumar, R. Sajeev, and P. K. Saji, 2016:
Role of distinct flavours of IOD events on Indian summer
monsoon. Nat. Hazards,82, 1317–1326,
Ashok, K., Z. Guan, and T. Yamagata, 2003: Influence of the In-
dian Ocean dipole on the Australian winter rainfall. Geophys.
Res. Lett.,30, 1821,
Baek, S. H., J. E. Smerdon, S. Coats, A. P. Williams, B. I. Cook,E. R.
Cook, and R. Seager, 2017: Precipitation, temperature, and
teleconnection signals across the combined North American,
Monsoon Asia, and Old World Drought Atlases. J. Climate,
30, 7141–7155,
Barnston, A. G., M. Chelliah, and S. B. Goldenberg, 1997: Docu-
mentation of a highly ENSO-related SST region in the equa-
torial Pacific: Research note. Atmos.–Ocean,35, 367–383,
Blanford, H. F., 1887: The eleven-year periodical fluctuation of the
Carnatic rainfall. Nature,36, 227–229,
Boos, W. R., and Z. Kuang, 2010: Dominant control of the South
Asian monsoonby orographic insulation versus plateau heating.
Nature,463, 218–222,
Buchan, A., 1877: Meteorology and the Indian famine. Nature,16,
Cai, W., T. Cowan, and M. Raupach, 2009: Positive Indian
Ocean dipole events precondition southeast Australia
bushfires. Geophys. Res. Lett.,36, L19710,
——, P. van Rensch, T. Cowan, and H. H. Hendon, 2011: Telecon-
nection pathways of ENSO and the IOD and the mechanisms
for impacts on Australian rainfall. J. Climate,24, 3910–3923,
——, and Coauthors, 2014: Increasing frequency of extreme El
Niño events due to greenhouse warming. Nat. Climate Change,
4, 111–116,
Cane, M. A., 2010: Climate: A moist model monsoon. Nature,463,
Clarke, H., 1878: The drought. Nature,19, 53,
Cook, E. R., 2015: Asian monsoon variability over the past
millennium reconstructed from long tree-ring records: The
Monsoon Asia Drought Atlas, version 2 (MADAv2). AGU
Chapman Conf., Hong Kong, China, Amer. Geophys. Un-
——, R. Seager, M. A. Cane, and D. W. Stahle, 2007: North
American drought: Reconstructions, causes, and conse-
quences. Earth Sci. Rev.,81, 93–134,
——, K. J. Anchukaitis, B. M. Buckley, R. D. D’Arrigo, G. C.
Jacoby, and W. E. Wright, 2010a: Asian monsoon failure and
megadrought during the last millennium. Science,328, 486–
C. Woodhouse, 2010b: Megadroughts in North America: Placing
IPCC projections of hydroclimatic change in a long-term palaeo-
climate context. J. Quat. Sci.,25, 48–61,
——, and Coauthors, 2015: Old World megadroughts and plu-
vials during the Common Era. Sci. Adv., 1, e1500561, https://
Cuniff, R. L., 1970: The Great Drought: Northeast Brazil, 1877–
1880. Ph.D. thesis, University of Texas (Austin), 347 pp.
Davis, M., 2001: Late Victorian Holocausts: El Niño Famines and
the Making of the Third World. Verso Books, 470 pp.
Derby, O. A., 1878: The rainfall of Brazil and the sun-spots. Nature,
18, 384–385,
Elliott, J. R., S. P. Jewson, and R. T. Sutton, 2001: The impact of the
1997/98 El Niño event on the Atlantic Ocean. J. Climate,14,
Enfield, D. B., and D. A. Mayer, 1997: Tropical Atlantic sea surface
temperature variability and its relation to El Niño–Southern
Oscillation. J. Geophys. Res.,102, 929–945,
——, A. M. Mestas-Nuñez, and P. J. Trimble, 2001: The Atlantic
multidecadal oscillation and its relation to rainfall and river
flows in the continental U.S. Geophys. Res. Lett.,28, 2077–
Goddard, L., and N. E. Graham, 1999: Importance of the Indian
Ocean for simulating rainfall anomalies over eastern and
southern Africa. J. Geophys. Res.,104, 19 099–19 116, https://
Gráda, C. O., 2009: Famine: A Short History. Princeton University
Press, 327 pp.
Greenfield, G. M., 1992: The Great Drought and elite discourse in
imperial Brazil. Hisp. Amer. Hist. Rev.,72, 375–400, https://
Hao, Z., J. Zheng, G. Wu, X. Zhang, and Q. Ge, 2010: 1876–1878
severe drought in North China: Facts, impacts and climatic
background. Chin. Sci. Bull.,55, 3001–3007,
Harris, I., P. D. Jones, T. J. Osborn, and D. H. Lister, 2014: Up-
dated high-resolution grids of monthly climatic observations—
The CRU TS3.10 dataset. Int. J. Climatol.,34, 623–642, https://
Hasell, J., and M. Roser, 2018: Famines. Our World Data, accessed
January 2018,
ECEMBER 2018 S I N G H E T A L . 9465
Hastenrath, S., and L. Greischar, 1993: Circulation mecha-
nisms related to northeast Brazil rainfall anomalies.
J. Geophys. Res.,98, 5093–5102,
Herweijer, C., and R. Seager, 2008: The global footprint of
persistent extra-tropical drought in the instrumental era.
Int. J. Climatol.,28, 1761–1774,
Huang, B., and Coauthors, 2015: Extended Reconstructed Sea
Surface Temperature version 4 (ERSST.v4). Part I: Upgrades
and intercomparisons. J. Climate,28, 911–930,
Hunter, W. W., 1877: Droughts and famines in southern India.
Nature,14, 16,
Hurrell, J. W., 1995: Comparison of NCAR Community Climate
Model (CCM) climates. Climate Dyn.,11, 25–50, https://
Ihara, C., Y. Kushnir, M. A. Cane, and A. Kaplan, 2008: Timing of El
Niño–related warming and Indian summer monsoon rainfall.
J. Climate,21, 2711–2719,
Kang, S., B. Yang, C. Qin, J. Wang, F.Shi, and J. Liu, 2013: Extreme
drought events in the years 1877–1878, and 1928, in the south-
east Qilian Mountains and the air–sea coupling system. Quat.
Int.,283, 85–92,
Kaplan, A., M. A. Cane, Y. Kushnir, A. C. Clement, M. B.
Blumenthal, and B. Rajagopalan, 1998: Analyses of global sea
surface temperature 1856–1991. J. Geophys. Res. Oceans,103,
18 567–18 589,
Kiladis, G. N., and H. F. Diaz, 1986: An analysis of the 1877–78
ENSO episode and comparison with 1982–83. Mon. Wea. Rev.,
114, 1035–1047,
Kumar, K. K., B. Rajagopalan, and M. Cane, 1999: On the weakening
relationship between the Indian monsoon and ENSO. Science,
284, 2156–2159,
——, ——, M. Hoerling, G. Bates, and M. Cane, 2006: Unraveling
the mystery of Indian monsoon failure during El Niño. Sci-
ence,314, 115–119,
Lawrimore, J. H., M. J. Menne, B. E. Gleason, C. N. Williams,
D. B. Wuertz, R. S. Vose, and J. Rennie, 2011: An overview of
the Global Historical Climatology Network monthly mean
temperature data set, version 3. J. Geophys. Res.,116, D19121,
Lee, D. E., and M. Biasutti, 2014: Climatology and variability of
precipitation in the Twentieth-Century Reanalysis. J. Climate,
27, 5964–5981,
Lucena, D. B., J. Servain, and M. F. Gomes Filho, 2011: Rainfall
response in Northeast Brazil from ocean climate variability
during the second half of the twentieth century. J. Climate,24,
Lyon, B., and A. G. Barnston, 2005: ENSO and the spatial ex-
tent of interannual precipitation extremes in tropical
land areas. J. Climate,18, 5095–5109,
Meena, H. K., 2015: Famine in late 19th century India: Natural or
man-made. J. Hum. Soc. Sci. Res.,6, 35–44.
Meyers, G., P. McIntosh, L. Pigot, and M. Pook, 2007: The years
of El Niño, La Niña, and interactions with the tropical In-
dian Ocean. J. Climate,20, 2872–2880,
tive and break events of Indian summer monsoon during
1901–2014. Climate Dyn.,46, 3921–3939,
Palmer,J. G., and Coauthors, 2015: Drought variability in theeastern
Australia and New Zealand summer drought atlas (ANZDA,
CE 1500–2012) modulated by the interdecadal Pacific oscilla-
tion. Environ. Res. Lett.,10,124002,
Parthasarathy, B., N. A. Sontakke, A. A. Monot, and D. R.
Kothawale, 1987: Droughts/floods in the summer monsoon
season over different meteorological subdivisions of India for
the period 1871–1984. J. Climatol.,7, 57–70,
——, K. R. Kumar, and A. A. Munot, 1993: Homogeneous Indian
monsoon rainfall: Variability and prediction. Proc. Indian
Acad. Sci. (Earth Planet. Sci.),102, 121–155,
——, A. A. Munot, and D. R. Kothawale, 1994: All-India monthly
and seasonal rainfall series: 1871–1993. Theor. Appl. Climatol.,
49, 217–224,
——, ——, and ——, 1995: Monthly and seasonal rainfall series
for all-India homogeneous regions and meteorological sub-
divisions: 1871–1994. Indian Institute of Tropical Meteorology
Research Rep. RR-065, 113 pp.
Pokhrel, S., H. S. Chaudhari, S. K. Saha, A. Dhakate, R. K. Yadav,
K. Salunke, S. Mahapatra, and S. A. Rao, 2012: ENSO, IOD
and Indian summer monsoon in NCEP Climate Forecast
System. Climate Dyn.,39, 2143–2165,
Polzin, D., and S. Hastenrath, 2014: Climate of Brazil’s Nordeste
and tropical Atlantic sector: Preferred time scales of vari-
ability. Rev. Bras. Meteor.,29, 153–160,
Puma, M. J., S. Bose, S. Y. Chon, and B. I. Cook, 2015: Assessing
the evolving fragility of the global food system. Environ.
Res. Lett.,10, 024007,
Rajeevan, M., C. K. Unnikrishnan, J. Bhate, K. Niranjan Kumar,
and P. P. Sreekala, 2012: Northeast monsoon over India:
Variability and prediction. Meteor. Appl.,19, 226–236, https://
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V.
Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003:
Global analyses of sea surface temperature, sea ice, and night
marine air temperature since the late nineteenth century.
J. Geophys. Res.,108, 4407,
Reynolds,R. W., and T. M. Smith, 1994: Improved global sea surface
temperature analyses using optimum interpolation. J. Climate,
7, 929–948,,0929:
Ropelewski,C.F.,andM.S.Halpert, 1987: Global and re-
gional scale precipitation patterns associated with the El
Niño/Southern Oscillation. Mon. Wea. Rev.,115, 1606–
Saji, N. H., and T. Yamagata, 2003: Possible impacts of Indian
Ocean dipole mode events on global climate. Climate Res.,25,
——, B. N. Goswami, P. N. Vinayachandran, and T. Yamagata,
1999: A dipole mode in the tropical Indian Ocean. Nature,401,
Schlesinger, M. E., and N. Ramankutty, 1994: An oscillation in the
global climate system of period 65–70 years. Nature,367, 723–
Seager, R., Y. Kushnir, C. Herweijer, N. Naik, and J. Velez, 2005:
Modeling of tropical forcing of persistent droughts and
pluvials over western North America: 1856–2000. J. Climate,
18, 4065–4088,
——, N. Naik, and L. Vogel, 2012: Does global warming cause in-
tensified interannual hydroclimate variability? J. Climate,25,
Singh, D., M. Tsiang, B. Rajaratnam, and N. S. Diffenbaugh, 2014:
Observed changes in extreme wet and dry spells during the
South Asian summer monsoon season. Nat. Climate Change,
4, 456–461,
Slingo, J. M., and H. Annamalai, 2000: 1997: The El Niño of the
century and the response of the Indian summer monsoon.
Mon. Wea. Rev.,128, 1778–1797,
Trenberth, K., and D. P. Stepaniak, 2001: Indices of El Niño
evolution. J. Climate,14, 1697–1701,
——, and Coauthors, Eds., 2016: The Climate Data Guide:
Southern Oscillation Indices: Signal, Noise and Tahiti/Darwin
SLP (SOI). Accessed 2 March 2016, https://climatedataguide.
Ummenhofer, C. C., R. D. D’Arrigo, K. J. Anchukaitis, B. M.
Buckley, and E. R. Cook, 2013: Links between Indo-Pacific
climate variability and drought in the Monsoon Asia Drought
Atlas. Climate Dyn.,40, 1319–1334,
Uvo, C. B., C. A. Repelli, S. E. Zebiak, and Y. Kushnir, 1998:
The relationships between tropical Pacific and Atlantic SST
and Northeast Brazil monthly precipitation. J. Climate,11,
Wang, B., J. Li, and Q. He, 2017: Variable and robust East Asian
monsoon rainfall response to El Niño over the past 60 years
(1957–2016). Adv. Atmos. Sci.,34, 1235–1248,
Whitcombe, E., 1993: Famine mortality. Econ. Polit. Wkly.,28,
Xu, Z. X., K. Takeuchi, and H. Ishidaira, 2004: Correlation
between El Niño–Southern Oscillation (ENSO) and
precipitation in South-east Asia and the Pacific region.
Hydrol. Processes,18, 107–123,
ECEMBER 2018 S I N G H E T A L . 9467
... Here, I present a precipitation case study for Africa during the particularly dry period 1877-1878 CE, sometimes referred to as the Global Drought (Brönnimann, 2015) or the Late Victorian Great Drought (Davis, 2001). During this period, concurring prolonged droughts in South Asia and East Asia, Africa, South America, and the Mediterranean region caused substantial crop failures, catalyzing some of the most extreme and widespread famines in modern times (Davis, 2001;Cook et al., 2010;Singh et al., 2018). It is widely acknowledged that the record-breaking El Niño event of 1877-1878 CE is the driving force behind these extraordinary drought conditions, affecting much of the tropics (Kripalani and Kulkarni, 1997;Aceituno et al., 2009;Brönni-mann, 2015). ...
... It is widely acknowledged that the record-breaking El Niño event of 1877-1878 CE is the driving force behind these extraordinary drought conditions, affecting much of the tropics (Kripalani and Kulkarni, 1997;Aceituno et al., 2009;Brönni-mann, 2015). However, Singh et al. (2018) suggest that an extraordinary combination of preceding cool tropical Pacific conditions in 1870-1876 CE, a record strong Indian Ocean Dipole (IOD) in 1877 CE, a record warm northern Atlantic Ocean in 1878 CE, and the strong El Niño event in 1877-1878 CE led to these extreme drought events and the socalled Global Famine. On the African continent, northeastern and southern Africa were particularly affected by drought conditions. ...
Full-text available
Climatic variations have impacted societies since the very beginning of human history. In order to keep track of climatic changes over time, humans have thus often closely monitored the weather and natural phenomena influencing everyday life. Resulting documentary evidence from archives of societies enables invaluable insights into the past climate beyond the timescale of instrumental and early instrumental measurements. This information complements other proxies from archives of nature, such as tree rings in climate reconstructions, as documentary evidence often covers seasons (e.g., winter) and regions (e.g., Africa, eastern Russia, Siberia, China) that are not well covered with natural proxies. While a mature body of research on detecting climate signals from historical documents exists, the large majority of studies is confined to a local or regional scale and thus lacks a global perspective. Moreover, many studies from before the 1980s have not made the transition into the digital age and hence are essentially forgotten. Here, I attempt to compile the first-ever systematic global inventory of quantitative documentary evidence related to climate extending back to the Late Medieval Period. It combines information on past climate from all around the world, retrieved from many studies of documentary (i.e., written) sources. Historical evidence ranges from personal diaries, chronicles, and administrative and clerical documents to ship logbooks and newspaper articles. They include records of many sorts, e.g., tithe records, rogation ceremonies, extreme events like droughts and floods, and weather and phenological observations. The inventory, published as an electronic Supplement, is comprised of detailed event chronologies, time series, proxy indices, and calibrated reconstructions, with the majority of the documentary records providing indications on past temperature and precipitation anomalies. The overall focus is on document-based time series with significant potential for climate reconstruction. For each of the almost 700 records, extensive meta-information and directions to the data (if available) are given. To highlight the potential of documentary data for climate science, three case studies are presented and evaluated with different global reanalysis products. This comprehensive inventory promotes the first ever global perspective on quantitative documentary climate records and thus lays the foundation for incorporating documentary evidence into climate reconstruction on a global scale, complementing (early) instrumental measurements and natural climate proxies.
... In the second example, the value of additional marine data is studied for the case of the climate anomalies over Africa during the El Niño year of 1877/1878. This was one of the most momentous climate events in history (Singh et al., 2018). Using a period in the 1920s, I then address the value of assimilating upper air data and even total column ozone data (the period was chosen based on ozone data availability). ...
... In the second case I tested whether assimilating more marine pressure data could help to improve historical reanalyses. The case selected for this covers July 1877 to June 1878, when during a strong El Niño event, droughts were observed globally (Singh et al., 2018). Data from six ships were assimilated; a map of the ship tracks is shown in Figure 4. ...
Full-text available
Historical reanalyses have become a widely used resource for analyzing weather and climate processes and their changes over time. In this article I explore how further historical observations could support reanalyses and lead to products that reach further back in time or have a better quality. Using an off-line Ensemble Kalman Filter I estimate improvements arising from assimilating additional observations into the ensemble of the “Twentieth Century Reanalysis” Version 3 (20CRv3). I demonstrate this for three case studies and evaluate them using independent data and a leave-one-out approach. For Europe in 1807, assimilating additional pressure data improves the skill for pressure but slightly decreases it for temperature, while assimilating temperature data, a variable that is not assimilated in 20CRv3, improves the skill for temperature but slightly decreases it for pressure. Assimilating both leads to substantially increased skill in a leave-one-out approach. For Sub-Saharan Africa in 1877/78, assimilating ship-based pressure observations as well as land station data, albeit extremely sparse, leads to a slight improvement over the entire domain. Finally, for Europe in 1926/27, assimilating upper air and total column ozone observations both lead to improvements in geopotential height and wind in the middle troposphere and in total column ozone, but with little or no effect in the lower troposphere. This is because 20CRv3 is already close to perfect over Europe in this period. The article shows how additional observations could improve historical reanalyses. A backward extension to the 1780s seems possible, but further data rescue efforts are necessary. For some applications, improved fields as generated by the offline assimilation presented in this study could be useful.
... The risk of multiple breadbasket failures is elevated during the simultaneous physical hazards imposed by the large-scale natural climate variability modes such as El Niño-Southern Oscillation (ENSO), Indian Ocean Dipole and Atlantic Niño 11,12,19 . ENSO is one of the predominant drivers of hydroclimate variability across tropical regions and El Niño events are associated with several major historical synchronous droughts and extreme temperatures across Asia, Africa and South America 6,15,20,21 . For instance, the strong El Niño event in 1983 caused extreme heatwaves and droughts across multiple maize-producing regions that resulted in the most extensive simultaneous crop failures in recent records 12,16 . ...
... These regions also include important breadbaskets and vulnerable populations that depend on rainfed agriculture for their livelihood 23,24 . Given the importance of ENSO for hydroclimate variability over many of these regi ons 12,15,[25][26][27][28] (Extended Data Fig. 1c), we investigate the influence of El Niño and La Niña events on compound drought characteristics in the historical and late twenty-first century climate. We also quantify future changes in the exposure of population and agricultural areas to compound droughts to evaluate the societal implications of projected changes. ...
Full-text available
Spatially compounding extremes pose substantial threats to globally interconnected socio-economic systems. Here we use multiple large ensemble simulations of the high-emissions scenario to show increased risk of compound droughts during the boreal summer over ten global regions. Relative to the late twentieth century, the probability of compound droughts increases by ~40% and ~60% by the middle and late twenty-first century, respectively, with a disproportionate increase in risk across North America and the Amazon. These changes contribute to an approximately ninefold increase in agricultural area and population exposure to severe compound droughts with continued fossil-fuel dependence. ENSO is the predominant large-scale driver of compound droughts with 68% of historical events occurring during El Niño or La Niña conditions. With ENSO teleconnections remaining largely stationary in the future, a ~22% increase in frequency of ENSO events combined with projected warming drives the elevated risk of compound droughts.
... The 1877-1878 El Niño is among the strongest in the observational record (Huang et al., 2020;Sanchez et al., 2020). It has been associated with prolific drought and famine that affected India and China from 1876 to 1877 (Lin et al., 2020) when an estimated 3% of the world's population died, comparable to the impact of the 1918-1919 influenza pandemic (Singh et al., 2018). Furthermore, coral records from the central equatorial Pacific found evidence of significant warming in the northern Line Islands 12-18 months before the peak of the 1876-1878 El Niño event in late 1877 (Sanchez et al., 2020), analogous to the well-documented 2015-2016 El Niño (Levine and McPhaden, 2016). ...
... Figure 6 highlights this discrepancy, with particular attention on the 1877-1878 El Niño (Figure 6b). The 2-sigma uncertainty range in 1877 suggests that there was potential for a developing La Niña as late as June or July, which strongly contradicts evidence (Kiladis and Diaz, 1986;Singh et al., 2018;Lin et al., 2020;Sanchez et al., 2020) that one of the strongest El Niños on record (Huang et al., 2020) had already begun developing. Figure 6c shows the 1888-1889 El Niño and is a case where the 1-sigma and 2-sigma uncertainty ranges both suggest at least a positive Niño3.4 ...
Full-text available
We describe a new El Nino Southern Oscillation (ENSO) index that we call the Ensemble Oceanic NINO Index (Ensemble ONI). Ensemble ONI uses monthly sea surface temperature (SST) anomalies in the NINO3.4 region from 32 input SST datasets, which form the ensemble, and our new monthly ENSO index extends from the present back to the year 1850. We use the input datasets to quantify the uncertainty in the monthly index values. The uncertainty is calculated from datasets that are not completely independent from each other, but comparing our findings to other estimates of uncertainty in SST, well‐documented historical ENSO events, a detailed consideration of literature‐based definitions of ENSO events, and proxy‐based determinations of past ENSO events, we show that our uncertainty estimates from the ensemble of input datasets is comparable both in terms of average magnitude and time‐varying trend with other studies. Similar to past research, we find that ENSO events occur every 4‐5 years on average, and there have been six “Super” El Ninos (1877‐78, 1888‐89, 1972‐73, 1982‐83, 1997‐98, 2015‐16) that statistically rise above all other El Ninos since 1850. Finally, the time span of our work shows that El Nino events were most intense at ends of both the 19th Century and the 20th Century, with a lull in the mid‐1900s, corroborating previous instrumental, written, and proxy records.
... Moreover, a universal definition of drought does not exist due to the involvement of multiple hydroclimatic variables and stakeholders [9,44,43,65,106]. Since drought characteristics (onset, termination, intensity, and areal extent) are difficult to be measured, drought occurrence poses several challenges to water and food availability [46,63,57], leading to water scarcity, crop failure, forest fire, economic and financial challenges, migration, and loss of life [21,28,40,63,75,82,93]. For instance, recent droughts in India have significantly affected water and food availability, population, and ecology [6,28,63]). ...
Full-text available
Drought is one of the complex and deleterious natural hazards that poses severe challenges to water security, food production, ecosystem, and socio-economic condition in India. Using efficient drought monitoring and assessment, the severe impacts of drought can be reduced. However, drought monitoring and assessment are associated with large uncertainty due to different datasets, methods, drought indices, and modeling approaches. Here, we examine the sources of uncertainty in drought assessment using multiple observational and future projections datasets, methods, and hydrological models. Moreover, we discuss potential ways to overcome the challenges associated with drought assessment in India. The drought assessment without considering uncertainty may cause overestimation or underestimation of risk in the observed and projected future climate, which further affects the planning and management of water resources. Therefore, a thorough understanding of these challenges is essential to improve the existing drought monitoring and assessment approaches.
This chapter brings a climatic perspective to the study of Singaporean history by exploring the impacts of the strong El Niño inspired droughts of 1877, 1902 and 1911. The narrative focuses on unpacking the nexus of nature-inspired versus human-induced vulnerability to drought within the contexts of colonial urbanisation and looks at the short-to medium-term impacts of the events on society. It also explores how such events inspired new questions about the climate and regional teleconnections, as a wealth of evidence became available due to the increasingly connected nature of scientific institutions, scientific literature, and communications systems across the Indian Ocean World (IOW). By exploring the region climatically, this chapter connects with the others collated here to show how, despite the regional and national differences, the experience of climate-induced environmental disaster can provide a shared narrative across the IOW.
This introductory chapter sets out the thematic and methodological approaches taken in the remainder of the book. It argues that droughts and floods, triggered by global climatic anomalies associated with, for example, the El Niño Southern Oscillation, the Indian Ocean Dipole, volcanism, and sunspot activity, are crucial to the conception of the Indian Ocean World. It does so by engaging with the Braudelian concept of ‘deep structure.’ In Indian Ocean World Studies, this deep structure is the Indian Ocean monsoon system, which underpinned agriculture and thus the economy until at least c.1900 across the macro-region. But, while Braudel conceived of ‘deep structure’ as an almost unchanging environmental context in his study of the Mediterranean World, in the Indian Ocean World, changes occur regularly owing to the effects of global climatic anomalies on the monsoon system. The potential rapidity of ‘deep structure’ in Indian Ocean World studies, partly visible through an analysis of drought and flood events, represents a core rationale for this book.
This chapter investigates the effects of the 1876–1878 El Niño and positive Indian Ocean Dipole on equatorial eastern Africa. The region under review comprises mainland regions of Tanzania, Kenya, and Uganda, broadly corresponding to the central caravan route that linked inland regions to the wider Indian Ocean World through the nineteenth-century global ivory trade. It begins by using missionary and limnological sources to reconstruct climate in the region during the event. Notwithstanding some regional variations, the sources suggest that widespread drought occurred in 1876, with floods occurring in 1877–1878. Such an assessment is in line with climatological models that project El Niño’s effect on the region’s climate. It then examines how this drought and subsequent floods affected the region’s history. In so doing, it links this global climatic anomaly to disrupted agriculture, an epidemic of smallpox, an epizootic of bovine trypanosomiasis, and political instability.
The Intertropical Convergence Zone annually moves south from equatorial latitudes, bringing rain to the Upper Zambezi Basin, though it adopts a puzzling northwest to southeast configuration. Long-term average rainfall has been roughly stable for centuries, but conditions in the Indian Ocean, rather than in the Atlantic Ocean, affect variations between years and decades. Written sources exist 1680s–1830s for Angola, whose rivers supply most of the waters reaching the floodplains of the Upper Zambezi. The quality and coverage of documents improve markedly from the 1840s, and further information comes from tree-rings from the mid-1790s and from oral traditions and testimonies. This chapter charts variations in rainfall and floods, but without attempting to link such changes closely to climatic drivers. A better understanding of weather patterns possibly sheds new light on other historical phenomena, and it might also assist in making better predictions for future economic planning and international negotiations.
Given the short span of instrumental precipitation records in the South American Altiplano, long-term hydroclimatic records are needed to understand the nature of climate variability and to improve the predictability of 25 precipitation, a key natural resource for the socioeconomic development in the Altiplano and adjacent arid lowlands. In this region grows Polylepis tarapacana, a long-lived tree species that is very sensitive to hydroclimatic changes and have been widely used for tree-ring studies in the central and southern Altiplano. However, in the northern sector of the Peruvian and Chilean Altiplano (16º-19º S) still exist a gap of hydroclimatic tree-ring records. Our study provides an overview of the temporal evolution of annual precipitation for the period 1625-2013 CE at the northern South American Altiplano, allowing 30 for the identification of wet or dry periods based on a regional reconstruction composed by three P. tarapacana chronologies. An increase in the occurrence rate of extreme dry events, together with a decreasing trend in the reconstructed precipitation, have been recorded since the 1970s decade in the northern Altiplano within the context of the last ~four centuries. The average precipitation of the last 17-year stands out as the driest in our 389-years reconstruction. We revealed a temporal and spatial synchrony across the Altiplano region of wet conditions during the first half of the 19th century and the 35 drought conditions since mid 1970s recorded by independent tree-ring based hydroclimate reconstructions and several 2 paleoclimatic records based on other proxies available for the tropical Andes. The rainfall reconstruction provides also valuable information about the ENSO influences in the northern Altiplano precipitation. The spectral properties of the rainfall reconstruction showed strong imprints of ENSO variability at decadal, sub-decadal and inter-annual timescale , in particular from the Pacific N3 sector. Overall, the remarkable recent reduction in precipitation in comparison with previous 40 centuries, the increase in extreme dry events and the coupling between precipitation and ENSO variability reported by this work is essential information in the context of the growing demand for water resources in the Altiplano that will contribute to a better understanding of the vulnerability/resilience of the region to the projected evapotranspiration increase for the 21st century associated to global warming.
Full-text available
The tree-ring-based North American Drought Atlas (NADA), Monsoon Asia Drought Atlas (MADA), and Old World Drought Atlas (OWDA) collectively yield a near-hemispheric gridded reconstruction of hydroclimate variability over the last millennium. To test the robustness of the large-scale representation of hydroclimate variability across the drought atlases, the joint expression of seasonal climate variability and teleconnections in the NADA, MADA, and OWDA are compared against two global, observation-based PDSI products. Predominantly positive (negative) correlations are determined between seasonal precipitation (surface air temperature) and collocated tree-ring-based PDSI, with average Pearson's correlation coefficients increasing in magnitude from boreal winter to summer. For precipitation, these correlations tend to be stronger in the boreal winter and summer when calculated for the observed PDSI record, while remaining similar for temperature. Notwithstanding these differences, the drought atlases robustly express teleconnection patterns associated with El Niño-Southern Oscillation (ENSO), the North Atlantic Oscillation (NAO), the Pacific decadal oscillation (PDO), and the Atlantic multidecadal oscillation (AMO). These expressions exist in the drought atlas estimates of boreal summer PDSI despite the fact that these modes of climate variability are dominant in boreal winter, with the exception of the AMO. ENSO and NAO teleconnection patterns in the drought atlases are particularly consistent with their well-known dominant expressions in boreal winter and over theOWDAdomain, respectively. Collectively, the findings herein confirm that the joint Northern Hemisphere drought atlases robustly reflect large-scale patterns of hydroclimate variability on seasonal to multidecadal time scales over the twentieth century and are likely to provide similarly robust estimates of hydroclimate variability prior to the existence of widespread instrumental data.
Full-text available
The summer monsoon contributes to about 75 % of mean annual rainfall over the various meteorological subdivisions of India. The role of ocean–atmosphere phenomena such as Indian Ocean Dipole (IOD) and El Nino–Southern Oscillation (ENSO) on the Indian monsoon activity is intriguing. The impacts of ENSO, IOD and Equatorial Indian Ocean Oscillation on monsoon are distinct. The ENSO (IOD) in general affects the monsoon negatively (positively). The present study aims to understand the role of different types of IOD such as early IOD (EIOD), normal IOD and prolonged IOD (PIOD) on Indian Summer Monsoon Rainfall (ISMR). We find that an EIOD, which peaks in the mid-monsoon months (July and August), plays a significant role like PIOD in enhancing ISMR even though it has a medium Dipole Mode Index amplitude value compared to other IODs. During an EIOD, the combined effect of excess evaporation from Arabian Sea and the stronger cross-equatorial flow leads to the enhanced monsoon activity. In addition, there is a substantial decrease in the number of break spells during EIOD years.
Full-text available
Agricultural production across eastern Australia and New Zealand is highly vulnerable to drought, but there is a dearth of observational drought information prior to CE 1850. Using a comprehensive network of 176 drought-sensitive tree-ring chronologies and one coral series, we report the first Southern Hemisphere gridded drought atlas extending back to CE 1500. The austral summer (December–February) Palmer drought sensitivity index reconstruction accurately reproduces historically documented drought events associated with the first European settlement of Australia in CE 1788, and the leading principal component explains over 50% of the underlying variance. This leading mode of variability is strongly related to the Interdecadal Pacific Oscillation tripole index (IPO), with a strong and robust antiphase correlation between (1) eastern Australia and the New Zealand North Island and (2) the South Island. Reported positive, negative, and neutral phases of the IPO are consistently reconstructed by the drought atlas although the relationship since CE 1976 appears to have weakened.
Full-text available
Resuming earlier research, this study explores rainfall variability in Brazil's Nordeste and underlying circulation mechanisms. The semi-arid northern Nordeste has its short rainy season centered around March-April-May, when temperature maximum, low pressure trough and wind confluence reach their southernmost position. Interannual variability can be understood as departures from the average annual cycle. Based on novel long-term datasets, the present study explores the preferred time scales of variability. In Nordeste rainfall and pertinent circulation indices in the tropical Atlantic sector most prominent are frequencies of 13.2, 9.9 and 5.6 years. Frequency peak of 13.1 years appears also in the record of Southern Oscillation, and of 5.6 years in North Atlantic Oscillation, indicative of causality chain.
Severe flooding occurred in southern and northern China during the summer of 2016 when the 2015 super El Niño decayed to a normal condition. However, the mean precipitation during summer (June–July-August) 2016 does not show significant anomalies, suggesting that—over East Asia (EA)—seasonal mean anomalies have limited value in representing hydrological hazards. Scrutinizing season-evolving precipitation anomalies associated with 16 El Niño episodes during 1957–2016 reveals that, over EA, the spatiotemporal patterns among the four categories of El Niño events are quite variable, due to a large range of variability in the intensity and evolution of El Niño events and remarkable subseasonal migration of the rainfall anomalies. The only robust seasonal signal is the dry anomalies over central North China during the El Niño developing summer. Distinguishing strong and weak El Niño impacts is important. Only strong El Niño events can persistently enhance EA subtropical frontal precipitation from the peak season of El Niño to the ensuing summer, by stimulating intense interaction between the anomalous western Pacific anticyclone (WPAC) and underlying dipolar sea surface temperature anomalies in the Indo-Pacific warm pool, thereby maintaining the WPAC and leading to a prolonged El Niño impact on EA. A weak El Niño may also enhance the post-El Niño summer rainfall over EA, but through a different physical process: the WPAC re-emerges as a forced response to the rapid cooling in the eastern Pacific. The results suggest that the skillful prediction of rainfall over continental EA requires the accurate prediction of not only the strength and evolution of El Niño, but also the subseasonal migration of EA rainfall anomalies.
The study lists active and break monsoon events over India over a very long period (1901–2014) identified using criteria based on a rainfall index derived over a critical high rainfall region called core monsoon zone. The break and active spells identified in this study were mostly comparable with that identified in the earlier studies based on similar rainfall criteria during the common data period (1951–2007). However, some noticeable differences were observed in the rainfall anomaly pattern associated with the break monsoon spells identified in this study and that identified based on the synoptic criteria in the earlier studies. The stringent rainfall criteria used in this study seems to be better criteria for identifying the breaks. During the study period, both the active and break spells of short duration were more frequent than the long duration with about 63.4 % of the break spells and 94.3 % of the active spells falling in the range of 3–6 days. There were no active spells of duration ≥13 days. Whereas, about 8 % of the break spells were of duration ≥13 days. During both the halves of the data period (1901–1957 and 1958–2014), there was no change in the distribution of the break events. However, the number of active spells showed an increase of about 12 % in the in the second half, which was mainly in the short duration (3–6 days) spells. During the data period, decadal variations of break days showed an out phase of relationship with the number of days of monsoon depression (MD). Relatively stronger in phase relationship was observed between the decadal variation of MD days and that of the active days till around early 1980s which reversed later due to sudden decrease in the MD days. During the same period, both the active and break days were in the increasing phase. This was also coincided with the sudden and significant increase in the number of days of monsoon lows (LOW). The LOWs, which generally have short life helped in the occurrence of active spells of short duration. Thus, post early 1980s, the increase in the active days covering short duration active spells caused by the significant increase in the LOW days compensated the decrease in the active days covering relatively long duration active spells caused by the MD days. This lead to the out of phase relationship between MD days and the active days post early 1980s.
The South Asian summer monsoon directly affects the lives of more than 1/6th of the world's population. There is substantial variability within the monsoon season, including fluctuations between periods of heavy rainfall (wet spells) and low rainfall (dry spells)(1). These fluctuations can cause extreme wet and dry regional conditions that adversely impact agricultural yields, water resources, infrastructure and human systems(2,3). Through a comprehensive statistical analysis of precipitation observations (1951-2011), we show that statistically significant decreases in peak-season precipitation over the core-monsoon region have co-occurred with statistically significant increases in daily-scale precipitation variability. Further, we find statistically significant increases in the frequency of dry spells and intensity of wet spells, and statistically significant decreases in the intensity of dry spells. These changes in extreme wet and dry spell characteristics are supported by increases in convective available potential energy and low-level moisture convergence, along with changes to the large-scale circulation aloft in the atmosphere. The observed changes in wet and dry extremes during the monsoon season are relevant for managing climate-related risks, with particular relevance for water resources, agriculture, disaster preparedness and infrastructure planning.
The performance of the Twentieth-Century Reanalysis (20CR) in reproducing observed monthly mean precipitation over the global domain is compared to that of comprehensive reanalyses that also assimilate upper-air and satellite observations [the Climate Forecast System Reanalysis (CFSR), ECMWF Interim Re-Analysis (ERA-Interim), and NCEP-U.S. Department of Energy reanalysis (NCEP2)] and to that of an atmospheric general circulation model (GCM) ensemble simulation [Global Ocean Global Atmosphere (GOGA)] that is forced with observed sea surface temperature (SST). Wintertime rainfall variability in the midlatitude continents and storm tracks is captured with great accuracy, similar to the comprehensive reanalyses, but summertime rainfall is not, probably in consequence of the greater importance of convection in the summer season. Over the tropics, the accuracy of all reanalyses is much less than over the midlatitudes. Over tropical land, the performance of 20CR is better than NCEP2 and similar to ERA-Interim and CFSR, but over the tropical oceans the most recent reanalyses perform significantly better. Across the twentieth century, the clearest gain from the assimilation of a denser observational dataset is the expansion of the area of good skill. A comparison of the accuracy and ensemble spread in the 20CR and GOGA ensembles highlights regions where SST forcing is a stronger source of skill than data assimilation for 20CR. In contrast, for some tropical regions such as the Sahel, the assimilation of sea level pressure is effective in constraining precipitation values-but model biases in the teleconnections with global SST limit the performance of 20CR.