ArticlePDF Available

Effects of rainfall variation on rice production in the Ganges-Brahmaputra Basin


Abstract and Figures

Rice production in India is highly correlated with monsoon rainfall. The relationships between rainfall variation and rice production have attracted significant interest at a country scale in Asia, but regional differences within a country remain unclear. In this study, we examined the effects of rainfall variation on 'kharif' rice (rainy season rice)-including temporal changes in this relationship-in the Ganges-Brahmaputra Basin, using a statistical model and a district-level data series of rice production and rainfall. Three homogeneous regions were identified within the study area. In the upper Ganges, the drought effect on rice production was dominant; however it became less pronounced due to decreased rainfall variation. In the lower Ganges, the flood effect increased due to increased rainfall. In the Brahmaputra Basin, the drought effect increased due to increased rainfall variation. Non-stationarity in the rainfall-rice production relationship was caused mainly by changes in rainfall patterns; however the impact of other factors, including social factors, should be evaluated on a regional scale.
Content may be subject to copyright.
Clim Res
Vol. 38: 249260, 2009
doi: 10.3354/cr00785
Printed May 2009
Published online April 21, 2009
Rice is the main staple in India and Bangladesh, and
these two countries account for ~28% of the total rice
production in the world (FAO 2008). Although total
rice production in India and Bangladesh is increasing
steadily every year, the weather related year-to-year
variation is still large. At farm level, the price of food-
grain in India rises >10% in the bad monsoon years
(Mooley et al. 1981), and income is unstable for local
farmers. Scarce rainfall even causes a drop (2 to 5% in
dry monsoon years) in gross domestic product, GDP
(the agricultural sector comprises 22% of GDP in India).
(Gadgil 1995, Gadgil & Rupa Kumar 2006, Gadgil et al.
2007). In addition, global warming may threaten food
security in India and Bangladesh.
The impact of future climate change on agricultural
production has been studied with crop simulation mod-
els (Karim et al. 1996, Kumar & Parikh 2001, Aggarwal
& Mall 2002, Mall & Aggarwal 2002). However, there is
a potential for bias of the simulation results, depending
upon the uncertainties in climate change scenarios,
region of study and crop models used for impact
assessment (Mall et al. 2006). Thus, it is important to
understand the past links between climate and agricul-
ture in order to improve the accuracy of crop models.
In order to reveal relationships between climate and
rice cultivation, many studies have been carried out at a
country scale. In India, foodgrain production, including
rice production, is highly correlated with the amount of
summer monsoon rainfall (SMR) from June to September
(Parthasarathy et al. 1988, Parthasarathy et al. 1992,
Selvaraju 2003, Krishna Kumar et al. 2004). Partha-
sarathy et al. (1988) found that the correlation coefficient
between SMR and rice production is 0.82 (19611985).
Parthasarathy et al. (1992) also showed that rice produc-
tion can be predicted statistically by using SMR as a de-
terminant index. El Niño-southern oscillation (ENSO) in-
dices are also correlated with rice production in India.
Krishna Kumar et al. (2004) showed that the sea-surface
temperatures (SSTs) in NINO3 (SSTs averaged over 5° S
to 5° N and 90 to 150° W) during June to August was
© Inter-Research 2009 ·*Email:
Effects of rainfall variation on rice production in
the Ganges-Brahmaputra Basin
Haruhisa Asada
, Jun Matsumoto
Graduate School of Asian and African Area Studies, Kyoto University, Kyoto 606-8501, Japan
Department of Geography, Tokyo Metropolitan University, Hachioji, Tokyo 192-0397, Japan
ABSTRACT: Rice production in India is highly correlated with monsoon rainfall. The relationships
between rainfall variation and rice production have attracted significant interest at a country scale in
Asia, but regional differences within a country remain unclear. In this study, we examined the effects
of rainfall variation on ‘kharif’ rice (rainy season rice) including temporal changes in this relation-
shipin the Ganges-Brahmaputra Basin, using a statistical model and a district-level data series
of rice production and rainfall. Three homogeneous regions were identified within the study area. In
the upper Ganges, the drought effect on rice production was dominant; however it became less pro-
nounced due to decreased rainfall variation. In the lower Ganges, the flood effect increased due to
increased rainfall. In the Brahmaputra Basin, the drought effect increased due to increased rainfall
variation. Non-stationarity in the rainfallrice production relationship was caused mainly by changes
in rainfall patterns; however the impact of other factors, including social factors, should be evaluated
on a regional scale.
KEY WORDS: Rice cultivation · India · Bangladesh · Ganges · Brahmaputra · Non-stationarity
Resale or republication not permitted without written consent of the publisher
Clim Res 38: 249260, 2009
significantly correlated with total rice production in In-
dia (r = –0.40, 19501998). Selvaraju (2003) suggested
that the rice production in India could be predicted
several months in advance from the SSTs in NINO3
during June to August.
However, most previous studies discuss the relation-
ship of climate and rice production in the entire
country, and it is doubtful whether the results are
applicable at the state or district level. In the Indian
subcontinent, rice is generally grown in the lowland of
major river basins or coastal delta areas, and research
at the regional scale is needed to reveal weatheryield
relationships. Krishna Kumar et al. (2004) analyzed
the correlation between state-level rice production and
subdivisional monsoon rainfall and showed that the
correlation is relatively low in some states of the east-
ern part of the country.
In the Ganges-Brahmaputra Basinwhich ranks among
monsoon Asia’s most productive regions in terms of
rice cultivationthe correlation between rice produc-
tion and monsoon rainfall remains unknown (Krishna
Kumar et al. 2004). Tanaka (1976) showed that there is
a high positive correlation (r = 0.54 for 19601975)
between rice yield and rainfall during June to August
in the Ganges Valley, but few studies have examined
this area since this study. In Bangladesh, located in the
lowest part of the Ganges-Brahmaputra Basin, Mowla
(1976) showed that rice production and yield are
highly related to fluctuations in annual rainfall. At the
same time, he pointed out that the recurrent floods
may also have a great influence on rice cultivation and
stressed the importance of researching such a relation-
The effect of floods on rice production in Bangladesh
was investigated by many researchers. Brammer (1990)
investigated the effect of severe floods in 1987 and
1988 on rice cultivation and found that transplanted
aman (monsoon rice), normally planted from mid-July
to mid-September, was the most affected in both years.
Paul & Rasid (1993) showed that average annual loss of
rice production resulting from flooding in Bangladesh
was approximately 4% of the total production, and the
highest proportional loss was almost 14% in the severe
flood year of 1988. Asada et al. (2005) revealed the
dynamic change of the rice-cropping pattern in Bang-
ladesh in the severe flood years of 1988 and 1998.
Although the total rice production in Bangladesh is less
influenced by rainfall and floods, the annual variation
of rainy season rice production is still large.
Another problem is that the relationship between
rainfall and rice production can change over time (i.e.
it exhibits non-stationarity). For example, Parthasa-
rathy et al. (1992) showed that the correlation coeffi-
cient for these two factors changed in different de-
cades. The correlation coefficient between SMR and
rice production in India during 19771988 is higher (r =
0.948) than for 19661976 (r = 0.816). As with other
uncertainties, this uncertainty will also affect the pre-
diction of crop models (Challinor et al. 2005a). Challi-
nor et al. (2005b) suggested that climatic factors (such
as climatic trends and random variability in the system)
and nonclimatic factors (such as yield technology and
changes in data accuracy) contribute to the non-
stationarity of the weatheryield relationship. The
non-stationarity of climate and agriculture has not
been examined in depth, and it is necessary to under-
stand these links to improve model accuracy for future
yield prediction.
In the present study we investigated the effect of
rainfall on rice production in both the Indian and
Bangladeshi parts of the Ganges-Brahmaputra Basin
over the past 40 yr, in addition to the relationship
between rainfall change and rice production. Among
the climatic factors affecting variation of rice produc-
tion, the impact of rainfall was examined, as it is the
most important limiting factor in South Asia (Gadgil &
Rupa Kumar 2006). The study was carried out for the
period of 19612000, a period for which district-level
statistics of rice production and rainfall are available.
The study area covers Bangladesh and the Indian
states Assam, West Bengal (WB), Bihar and Uttar Pra-
desh (UP) (
Fig. 1). These four states account for about
40% of the total rice production in India (14.6, 13.6, 6.4
and 4.7% in WB, UP, Bihar and Assam, respectively, in
2000/01). Assam and part of Bangladesh lie in the
Brahmaputra Basin. Bihar, WB and most of Bangladesh
lie in the lower Ganges Basin. UP is in the upper
Ganges Basin. The study area amounts to ca. 46%
of the total Ganges-Brahmaputra catchment areas
(1.7 million km
). Although the new states Uttaranchal
and Jharkhand were separated out of UP and Bihar
respectively in 2000, the former administrative units
of UP and Bihar were considered.
The climate of the study area is predominantly a sub-
tropical monsoon climate. In the months from June to
September, the southwestern monsoon flow brings a
humid air mass and causes high precipitation in the
whole study area. The amount of SMR is higher in the
eastern part of the basin and not so high in the western
part. The onset of the summer monsoon begins in April
in eastern Assam and in late May in the southeastern
part of Bangladesh and advances westwards upstream
of the Ganges. SMR amounts to ca. 60 to 70% of the
annual rainfall and causes floods in the study area.
The rice cropping system in the study area corre-
sponds to the local climate and hydrological environ-
Asada & Matsumoto: Rainfall variation and rice production in India/Bangladesh
ment. In Assam, Bangladesh and WB, rice is grown
three times a year as follows: autumn rice (April to
July), winter rice (August to November) and summer
rice (December to May). Autumn and winter rice are
classified as the rainy-season variety (‘kharif’ rice).
Kharif rice is grown mostly in a rain-fed field and is
more vulnerable to water shortage. In Bihar and UP,
only kharif rice is grown during the year, as wheat and
other crops are planted in the dry season. Summer rice
is classified as the dry-season variety (‘rabi’ rice). Rabi
rice is grown in irrigated paddy fields and yield is
much higher than kharif rice owing to more hours of
sunshine and fewer natural disasters than in the rainy
season. As kharif rice is more vulnerable to rainfall
variation than rabi rice (Krishna Kumar et al. 2004) and
is cultivated throughout the study area, the effect of
rainfall was only investigated on kharif rice in this study.
3.1. Rice production data
Rice production and cultivated area data at state and
district level during 19612000 were obtained from the
yearly agricultural bulletins ‘Agricultural Situation in
India’, Dep. of Agriculture, Gov. of India, ‘Yearbook
of Agricultural Statistics of Bangladesh’, Bangladesh
Bureau of Statistics, Hamid (1991) and FAO-RAP
(2008). For years where data was lacking, statistics
published by local governments were used. Yield was
derived by dividing production by cultivated area.
3.2. Climate data
Monthly rainfall data during 19612000 was obtained
from the Indian Institute of Tropical Meteorology (IITM, and the 0.5 × 0.5° gridded dataset
from the Variability Analysis of Surface Climate Obser-
vations (VASClimO) in the Global Precipitation Clima-
tology Centre (GPCC). The details and the location of
the original rain gauges of the IITM dataset are well
documented in Parthasarathy et al. (1994). A brief intro-
duction of the VASClimO dataset is reported in Beck et
al. (2005). The monthly rainfall in every district was cal-
culated from gridded data by averaging the grid value
superimposed over the district area. In order to calcu-
late the monthly value of Bangladesh rainfall, daily
rainfall data at 14 stations from the Bangladesh Meteo-
rological Department (BMD) was analyzed.
Flood-affected area data in Assam, WB, Bihar and
UP during 19612000 were obtained from the Central
Water Commission (CWC), the Ministry of Water Re-
sources, and the Gov. of India. Flood affected area data
in Bangladesh for the same period was obtained from
the Flood Forecasting and Warning Center (FFWC)
and the Bangladesh Water Development Board (BWDB).
The assessment techniques adopted to calculate flood
affected areas are reported in Mirza et al. (2001).
3.3. Methods
The effect of rainfall variation on rice production was
investigated by statistical modeling. Simple correlation
Fig. 1. Study area and the Ganges – Brahmaputra river system. The study area covered Bangladesh and 4 Indian states. Black,
grey and dotted lines represent international borders, state borders and major rivers, respectively. Shading:topographical relief
Clim Res 38: 249260, 2009
was examined after detrending each data series. All
the time series of rainfall were normalized by their
respective standard deviations (SDs) for 19612000.
The long-term trend of technological progress in rice
production, area and yield data were removed by
applying the 5 yr running mean, and the percentile
anomaly from the running means were used as indices
of short-term variation of production, area and yield.
The flood affected ratio was derived by dividing the
flood affected area by the geographical area, and the
correlation with rice production was examined without
detrending the time series.
The temporal correlation between rice production
and SMR was examined by 7 yr moving correlation.
The temporal trend of the correlation coefficient was
examined by Mann-Kendall rank statistics (Kendall
1938). In addition, flood effect and drought effect were
examined by the significance level of the correlation
coefficient (r = 0.76, significant level at 5%). Finally,
the temporal change of the correlation in each district
was classified into 6 classes (Fig. 2). Districts showing
a significant positive trend were defined into three
classes: ‘flood decreasing and drought increasing’
(with both positive r and negative r, Fig. 2a), ‘flood
decreasing’ (with only negative r,
Fig. 2b) and ‘drought
increasing’ (with only positive r,
Fig. 2c). Districts
showing a significantly negative trend were also
defined into three classes: ‘drought decreasing and
flood increasing’ (with both positive r and negative r,
Fig. 2d), ‘drought decreasing’ (with only positive r,
1961 1971 1981 1991 1961 1971 1981 1991
1961 1971 1981 1991 1961 1971 1981 1991
1961 1971 1981 1991 1961 1971 1981 1991
TP (i) TP (iv)
TP (ii) TP (v)
TP (iii) TP (vi)
Fig. 2. Classification of district by correlation coefficient (r): (a) flood decreasing and drought increasing, (b) flood decreasing,
(c) drought increasing, (d) drought decreasing and flood increasing, (e) drought decreasing, (f) flood increasing. Dotted lines:
significance levels at 5% (r = 0.76). TP: Trend pattern (referred to in Fig. 6)
Asada & Matsumoto: Rainfall variation and rice production in India/Bangladesh
Fig. 2e) and ‘flood increasing’ (with
only negative r,
Fig. 2f).
The trends of rainfall amount and
year-to-year rainfall variations were
also examined in each district. The
trend of rainfall amount was examined
by applying Mann-Kendall rank statis-
tics to the 5 yr running average of rain-
fall time series. The trend of rainfall
variations was examined by applying
Mann-Kendall rank statistics to the 5 yr
running average SD of the rainfall time
In this study, the positive (negative)
correlation between rainfall and rice
production which means that scarce
(excess) rainfall causes production
losswas defined as the drought
(flood) effect on rice production. Flood
effect means damage caused by heavy
rainfall in the area, and does not
include the case of river water flooding which is caused
by rainfall in upstream areas. We examined the cases
when the correlation coefficient was statistically signif-
icant. But it should be noted that when the correlation
coefficients were low or not significant, it does not
always mean that there was not any relationship be-
tween rainfall and rice production.
4.1. Rainfall effect at the state level
Kharif rice accounts for a large fraction of total rice
production in the study area (
Table 1). In Bangladesh
and WB (India), rabi production and area are compar-
atively larger. In other Indian states, rabi rice is culti-
vated marginally and is less important. The SDs of
kharif rice in the 1990s are high, especially in Bihar,
where >10% of kharif rice production is variable every
year. The variation of kharif production is largely
caused by variation in yield. As the irrigated ratio in
the kharif area is small, kharif rice depends on rainfall
through its growing period, and yield becomes unsta-
ble. The variation of rabi rice is also large except for in
Bangladesh. Yield, however, is more stable than that of
kharif rice, as most of the rabi area is irrigated. Varia-
tion of rabi production is largely due to variation of
cultivated area. Rabi rice is cultivated after harvest of
kharif rice, and the variation in cultivated area seems
to be less influenced by climatic factors.
The correlation coefficient between rainfall and kharif
rice shows different characteristics of rainfall effect
by state (Table 2). Rice production in Assam and
Table 1. Production, cultivated area and yield of kharif and rabi rice in 2000
in Bangladesh and 4 Indian states (from west to east). Parentheses: SDs of
detrended percentile anomaly during 19912000. WB: West Bengal; UP: Uttar
Production Area Yield Irrigated
(1000 t) (1000 ha) (t ha
) area (%)
UP 11540.1 (5.6) 5838.8 (1.7) 1.98 (5.5) 67.4
Bihar 5229.4 (14.6)0 35310. (4.9) 1.48 (13.4)0 36.6
WB 7886.7 (5.4) 4033.5 (2.4) 1.96 (4.8) 16
Bangladesh 13164.5 (8.5) 7033.8 (3.8) 1.87 (5.5) 5.6
Assam 3317.4 (3.4) 2316.9 (1.8) 1.43 (2.1) 4.3
UP 5.1 (20.0)0 2.2 (19.0)0 2.29 (8.5) 96.3
Bihar 213.2 (8.4) 125.3 (3.4) 1.7 (5.0) 76.7
WB 4541.3 (9.6) 1401.8 (7.1) 3.24 (4.3) 110.7
Bangladesh 11920.9 (5.0) 3761.1 (4.0) 3.17 (1.9) 83.2
Assam 6810. (11.3)0 329.3 (8.9) 2.07 (3.9) 3.7
Table 2. Correlation coefficient between monthly rainfall and
kharif production, area and yield for the period 19612000 in
Bangladesh and 4 Indian states (from west to east). *, ** p <
0.05, 0.01, respectively. Correlation with the flood-affected
area also shown. WB: West Bengal; UP: Uttar Pradesh; SMR:
summer monsoon rainfall
UP Bihar WB Bangladesh Assam
Apr 0.02 0.20 0.05 0.01 0.40**
May 0.01 0.26 0.16 0.12 0.01
Jun 0.22 0.31* 0.32* 0.04 0.12
Jul 0.51** 0.36* 0.36* 0.04 0.08
Aug 0.32* 0.13 0.17 0.30 0.17
Sep 0.55** 0.08 0.05 0.14 0.15
SMR (JJAS) 0.73** 0.45** 0.32* 0.10 0.08
Flood area 0.43** 0.03 0.03 0.52** 0.29
Apr 0.01 0.15 0.02 0.03 0.13
May 0.02 0.44** 0.45** 0.13 0.00
Jun 0.44** 0.30 0.04 0.20 0.17
Jul 0.56** 0.57** 0.39* 0.04 0.03
Aug 0.16 0.13 0.16 0.14 0.16
Sep 0.22 0.13 0.13 0.21 0.02
SMR (JJAS) 0.62** 0.58** 0.17 0.12 0.01
Flood area 0.29 0.06 0.16 0.53** 0.28
Apr 0.02 0.20 0.06 0.04 0.46**
May 0.00 0.19 0.05 0.26 0.02
Jun 0.15 0.28 0.36* 0.20 0.10
Jul 0.43** 0.28 0.30 0.03 0.10
Aug 0.33* 0.12 0.15 0.35* 0.13
Sep 0.56** 0.08 0.00 0.07 0.18
SMR (JJAS) 0.68** 0.38* 0.33* 0.06 0.13
Flood area 0.41** 0.01 0.09 0.42** 0.24
Clim Res 38: 249260, 2009
Bangladesh has little relation to the amount of SMR.
The correlation becomes higher in Bihar and UP,
where the rainy season is relatively shorter and there is
less SMR. Both kharif area and yield are correlated
with monthly rainfall within the growing period. The
correlation coefficient is significant at the 1% level in
Assam (associated with April rainfall) and Bihar and
UP (associated with July and September rainfall). The
timing coincides with the western migration of mon-
soon onset, which starts in April from Assam. This is
consistent with the results of Gadgil et al. (2002) that
the monsoon onset has significant influence on kharif
rice production. In Bangladesh, a sig-
nificant negative correlation is seen
between rainfall in August and kharif
yield. A negative correlation is also
seen in Assam, though the correlation
coefficient is not significant. Rainfall in
August can easily damage rice pro-
duction because of flooding, as it is
the peak period of river water level.
The correlation coefficients between the
flood affected area and kharif produc-
tion, area and yield in Bangladesh and
Assam all show a significant negative
4.2. Rainfall effect at district level
Fig. 3 shows the district-level correla-
tion between SMR and kharif produc-
tion, area and yield during 19612000.
The positive correlation between SMR
and kharif production is highest in the
Gaya district of central Bihar (r = 0.67;
p < 0.01), and the negative correlation is
highest in the Pabna district of western
Bangladesh (r = –0.38; p = 0.02). The
significant positive correlation is espe-
cially dominant in Bihar and UP. The
number of districts showing positive
correlation is largest in UP, and de-
creases in the order of Bihar, WB and
Assam. In WB, Bangladesh and Assam,
both positive and negative correlations
are recognized, and a distinct pattern
cannot be seen. In western Bangladesh,
the negative correlation is shown in the
kharif growing area. The drought effect
on kharif production is mainly due to
yield loss resulting from water shortage.
On the other hand, the flood effect
on kharif production is mainly caused
from inundation of the cultivated area.
In Bangladesh, the districts along the major rivers are
more vulnerable to floods.
The period and timing of rainfall as well as the total
amount of SMR during the summer monsoon season is
important for rice cultivation (Gadgil et al. 2002). The
correlation of monthly rainfall from April to September
with kharif production during 19612000 was investi-
gated at the district level (
Fig. 4). From April to June,
during the initial stage of kharif rice production, the
number of districts showing a significant correlation
was relatively low compared to other periods. Rainfall
from July to September shows a positive correlation
Fig. 3. Correlation coefficient between summer monsoon rainfall and kharif (a) pro-
duction, (b) growing area, (c) yield for the period 19612000 in Bangladesh and 4
Indian states. r = 0.26 and 0.31 for significance levels of 10 and 5%, respectively
Asada & Matsumoto: Rainfall variation and rice production in India/Bangladesh
with rice production in districts in UP and Bihar. Rain-
fall in August, however, is not correlated with rice pro-
duction in many districts. On the contrary, a negative
correlation between rice production and rainfall is
seen in the wide area of the lower Ganges. In July, dis-
tricts in southern Bangladesh show a negative correla-
tion, and more districts from Bangladesh to north Bihar
show a negative correlation in August. In Assam, dis-
tricts with a significant correlation decreased during
the whole monsoon season.
4.3. Non-stationarity in the rainfallrice production
The relationship between rainfall and rice produc-
tion did not remain constant for the whole period. The
temporal change of the correlation coefficient between
SMR and rice production in each state is shown in
Fig. 5. A 7 yr moving correlation was used to examine
the temporal change. There were regional differences
in the temporal variation of the correlation coefficient.
In UP, the correlation coefficient remained high, and
exceeded the 5% significance level. In Bihar, the cor-
relation coefficient was more variable than in UP, but
remained positive over the 40 yr period.
In Assam, Bangladesh and WB, the correlation coef-
ficient was not constant, and the fluctuations were
much larger than UP and Bihar. In Assam, the correla-
tion coefficient was negative until 1975, but after that it
turned positive, which means that the drought effect
became more prominent than the flood effect. In con-
trast, in Bangladesh the correlation coefficient turned
negative in 1975, and the flood effect increased. Simi-
larly, the flood effect increased in WB recently, but the
correlation coefficient turned negative in the early
Fig. 4. Correlation coefficient between kharif production and rainfall in Bangladesh and 4 Indian states in (a) April, (b) May, (c) June,
(d) July, (e) August, (f) September for the period 19612000. r = 0.26 and 0.31 for significance level of 10 and 5%, respectively
Clim Res 38: 249260, 2009
1990s. The change in the correlation coefficient of pro-
duction was largely due to the change in the correla-
tion coefficient of yield. In Table 2, the correlation
coefficient between rice production and SMR is very
low in Bangladesh and Assam, and this is because the
correlation coefficient changed adversely from positive
to negative and vice versa over the 40 yr period.
The investigation of temporal change of the correla-
tion coefficient showed regional differences in non-
stationarity; however, homogeneous regions can be
recognized within the study area (
Fig. 6). The first
region includes the districts in eastern UP and south-
ern Bihar, where the drought effect on rice production
decreased. The decreasing effect of drought on yield is
prominent. The second region includes
the districts extending from northern
Bihar to northern Bangladesh, where the
flood effect on rice production increased.
Some districts are adjacent to the
Ganges. Both rice area and yield are
influenced by an increasing flood effect.
The third region includes the districts in
Assam and part of northern WB, where
the drought effect on rice production in-
creased. The increasing drought effect is
prominent in the districts located on the
northern side of the Brahmaputra. The
yield was more affected by drought than
other regions. It is interesting that both
northern Bihar and Assam are in the
southern foothills of the Himalayas, but
nevertheless had opposing rainfall trends.
The trends of rainfall amount and rain-
fall variation in each district are shown
in Fig. 7. Three regions show distinctive
changes of rainfall, and they correspond
to the area showing significant non-
stationarity in
Fig. 6. In eastern UP and
southern Bihar, variation in rainfall de-
creased, indicating that year-to-year
variations stabilized. The SD of SMR
in these districts decreased 119.6 mm
40 yr
(13.7% of SMR) on average. Rain-
fall changes in April, June, August and
September were responsible for the
change in SMR. From north Bihar to
north Bangladesh, the rainfall amount
showed an increasing trend. The SMR
increased 208.0 mm 40 yr
(17.0% of
SMR) on average, and the Malda district
in central WB showed the highest rain-
fall increase, 324 mm 40 yr
(27.7% of
SMR). Rainfall changes in June, July and
August are responsible. In Assam, rain-
fall variation increased, indicating that
more extreme events like severe floods or droughts are
likely to occur. The SD of rainfall increased 184.3 mm
40 yr
(10.2% of SMR). Rainfall change in May, June
and September was responsible.
On the basis of the pattern of rainfall effect on rice
production in the study area, 3 homogeneous regions
can be identified: (1) eastern UP and southern Bihar in
the upper Ganges Basin, (2) the northern part of Bihar,
WB and Bangladesh in the lower Ganges Basin, and
(3) northern Assam in the Brahmaputra Basin.
1961 1971 1981 1991
1961 1971 1981 1991
1961 1971 1981 1991
Fig. 5. Correlation coefficient between summer monsoon rainfall and kharif (a)
production, (b) growing area, (c) yield in a 7 yr moving window in Bangladesh and
4 Indian states in Bangladesh and 4 Indian states. Dashed lines: significant at 5%
level (r = 0.76). WB: West Bengal; UP: Uttar Pradesh
Asada & Matsumoto: Rainfall variation and rice production in India/Bangladesh
In the upper Ganges Basin, rice production is
strongly affected by rainfall fluctuation, and is vulner-
able to rainfall shortage, which is similar to the rela-
tionship throughout all of India as revealed in previous
studies (Parthasarathy et al. 1988, 1992, Selvaraju
2003, Krishna Kumar et al. 2004). The period of the
rainy season is shorter, and the rainfall amount avail-
able for rice cultivation is limited. The scarce rainfall
from July to September caused the yield loss, but the
drought effect on rice production de-
creased recently. Rainfall change is
favorable, as stable rainfall can de-
crease the drought effect and help to
stabilize rice production variation.
In the lower Ganges Basin, districts
along the major rivers show a flood
effect on rice production, and this
effect has recently been increasing. A
trend of increasing rainfall may be
partly responsible for the change in
the relationship. Heavy rainfall in
the transplanting period causes cul-
tivated area loss, and deficiency of the
growth period causes yield loss.
In the Brahmaputra Basin, district
analysis shows the increasing effect
of drought on rice production due
to increased rainfall variation. An in-
crease of either severe floods or
droughts can cause rice production
loss, but the drought effect on rice
production is more prominent. At
the state-level analysis, however, the
correlation coefficient between the
flood-affected area and rice produc-
tion shows a significantly negative
correlation, and thus kharif rice in
Assam is vulnerable to both floods and
droughts. In addition to the effect of
rainwater flooding, the effect of river-
water flooding should also be used to
more accurately estimate the flood
effect on rice production.
All of the 3 regions are smaller than
state (in India) or country (Bangladesh)
level, and some extend over state bor-
ders. Analysis at state level provides
insufficient resolution. It is important
to analyse the rainfallrice production
relationship at a district scale, and
consider this measure for amelioration
of future climate changes. In the re-
gions where the rainfall effect is in-
creasing, measures to control the local
hydrological environment such as irri-
gation and the construction of flood embankments may
be required. The results could also contribute towards
improving the accuracy of crop models, if the local
mechanism of the relationship is revealed.
Changes in the relationship between rainfall and
rice production were detected over a period of 40 yr.
The non-stationarity could be partly due to the rainfall
change in the same period, but the mechanism still
remains unclear. Flood effect increased both in Bang-
Fig. 6. Classification of district based on trend patterns (i) to (vi) (see Fig. 2).
Trend of correlation coefficient between summer monsoon rainfall and kharif
(a) production, (b) area, (c) yield
Clim Res 38: 249260, 2009
ladesh and WB, but the timing of the prominence of
the effect differed by region. Challinor et al. (2005b)
suggested that social factors also contribute to non-
stationarity along with climate factors. Different social
factors such as introduction of irrigation or changes in
cropping systems may have contributed to the differ-
ent timing of temporal change.
Non-stationarity in monsoon rainfall is another
important problem. SMR in the Ganges-Brahmaputra
Basin showed a decreasing trend over the last century
(Gregory 1989, Rupa Kumar et al. 1992). SMR in the
eastern Ganges plain (Bihar and WB) showed a de-
creasing trend from 19001984, but began to increase
after 1984 (Singh & Sontakke 2002). The number of
extreme rainfall events also showed a decreasing trend
in the Ganges Basin. Extreme rainfall events of 1 to
3 d duration showed a significant decreasing trend
(>10%) in Bihar during 19011980 (Rakhecha &
Soman 1994), and extreme daily rainfall showed a
decreasing trend in Bihar and WB during 19102000
(Roy & Balling 2004). Rainfall changes in this study
period are not consistent with the results of the previ-
Rainfall amount
Rainfall variation
Fig. 7. Trends of rainfall amount and rainfall variation in
Bangladesh and 4 Indian states for the period 19612000. (a)
Summer monsoon rainfall, (b) April, (c) May, (d) June, (e) July,
(f) August, (g) September. + (–): increasing (decreasing) trend.
Significance level is at 5%
Asada & Matsumoto: Rainfall variation and rice production in India/Bangladesh
ous studies, and illustrate the interdecadal variation of
Indian SMR (ISMR) (Parthasarathy et al. 1991). ISMR
has a long-term oscillation within a period of about
30 yr (Parthasarathy et al. 1991, Krishnamurthy &
Goswami 2000), and rainfall changes for the study
period in this study can be considered as a part of the
cycle. The teleconnection between ISMR and ENSO
changes at a decadal scale, and contributes to the
interdecadal variations of ISMR (Krishnamurthy &
Goswami 2000). Turner et al. (2007) showed that the
teleconnection between ISMR and ENSO is likely to
remain robust with future climate change. If inter-
decadal variations of rainfall can be predicted on a
time scale of 30 to 40 yr, it could help improve future
prediction of rice production.
This study revealed the non-stationary effect of rain-
fall variation on rice cultivation in the Ganges-Brahma-
putra Basin, one of the most populated and most
important rice producing areas in South Asia. The
majority of previous studies have analyzed the rela-
tionship between rainfall and rice production at whole
country or state level. The district-level analysis used
in this study more effectively revealed the area (i.e.
across several districts, both interstate and inter-
country) where rice production is particularly vulnera-
ble to rainfall variation. We also found that the effect of
rainfall variation on rice production changes over time
in the study area, and that this non-stationarity can
be explained by long-term rainfall change. However,
the mechanism causing this change remains unclear,
and there are still many uncertainties in the non-
stationarity of the rainfallrice production relationship.
This study provides basic information for under-
standing the rainfallrice production relationship in
the Ganges-Brahmaputra Basin. The effect of rainfall
on rice production differs by region, so research should
be carried out at a regional scale, and local mecha-
nisms (including social factors) need to be analysed to
help stabilize year to year variation of rice production.
Regional planning measures including for irrigation
and river-embankment workare necessary to mini-
mize the influence of future rainfall change on rice cul-
tivation. The regional and temporal scales of the non-
stationarity revealed in this study should help reduce
uncertainty for simulations of future climate impacts on
rice production. The mechanism of non-stationarity
needs to be further studied to improve the predictive
accuracy of crop models.
Acknowledgements. The authors thank Prof. K. Ando, Center
for Southeast Asian Studies, Kyoto University, and Prof. A. K.
Bhagabati, Department of Geography, Gauhati University,
for providing helpful advice. Comments from anonymous
reviewers were also appreciated. This research was sup-
ported by the Japan Society for the Promotion of Science.
Aggarwal PK, Mall RK (2002) Climate change and rice yields
in diverse agro-environments of India. II. Effect of uncer-
tainties in scenarios and crop models on impact assess-
ment. Clim Change 52:331343
Asada H, Matsumoto J, Rahman R (2005) Impact of recent
severe floods on rice production in Bangladesh. Geogr
Rev Jpn 78:783793
Beck C, Grieser J, Rudolf B (2005) A new monthly precipita-
tion climatology for the global land areas for the period
1951 to 2000. In: Climate Status Report, 2004. German
Meteorological Service, Offenbach, p181-190
Brammer H (1990) Floods in Bangladesh. I. Geographical
background to the 1987 and 1988 floods. Geogr J 156:
Challinor AJ, Wheeler TR, Slingo JM, Hemming D (2005a)
Quantification of physical and biological uncertainty in
the simulation of the yield of a tropical crop using present-
day and doubled CO2 climates. Philos Trans R Soc B 360:
Challinor AJ, Wheeler TR, Slingo JM, Craufurd PQ, Grimes
DIF (2005b) Simulation of crop yields using ERA-40: limits
to skill and nonstationarity in weatheryield relationships.
J Appl Meteorol 44:516531
FAO (2008) Statistical databases & data-sets.
waicent/portal/statistics_en.asp (accessed 1 October 2008)
FAO-RAP (2008) Regional data exchange system. www.faorap- (accessed 1 October 2008)
Gadgil S (1995) Climate change and agriculture: an Indian
perspective. Curr Sci 69:649659
Gadgil S, Rupa Kumar K (2006) The Asian monsoon, agricul-
ture and economy. In: Wang B (ed) The Asian monsoon.
Springer, Berlin, p 651–683
Gadgil S, Seshagiri Rao PR, Rao KN (2002) Use of climate
information for farm-level decision making: rainfed ground-
nut in southern India. Agric Syst 74:431457
Gadgil S, Rajeevan M, Francis PA (2007) Monsoon variability:
links to major oscillations over the equatorial Pacific and
Indian oceans. Curr Sci 93:182194
Gregory S (1989) Macro-regional definition and characteris-
tics of Indian summer monsoon rainfall, 18711985. Int J
Climatol 9:465483
Hamid HA (1991) A database on agricultural and foodgrains
in Bangladesh. Binimoy Printers, Dhaka
Karim Z, Hussain SG, Ahmed M (1996) Assessing impacts of
climatic variations on foodgrain production in Bangladesh.
Water Air Soil Pollut 92:5362
Kendall MG (1938) A new measure of rank correlation.
Biometrika 30:8193
Krishna Kumar K, Rupa Kumar K, Ashrit RG, Desphpande
NR, Hansen JW (2004) Climate impacts on Indian agri-
culture. Int J Climatol 24:13751393
Krishnamurthy V, Goswami BN (2000) Indian monsoon-
ENSO relationship on interdecadal timescale. J Clim 13:
Kumar KSK, Parikh J (2001) Indian agriculture and climate
sensitivity. Global Environ Change 11:147154
Mall RK, Aggarwal PK (2002) Climate change and rice yields
in diverse agro-environments of India. I. Evaluation of
impact assessment models. Clim Change 52:315330
Clim Res 38: 249260, 2009260
Mall RK, Singh R, Gupta A, Srinivasan G, Rathore LS (2006)
Impact of climate change on Indian agriculture: a review.
Clim Change 78:445478
Mirza MMQ, Warrick RA, Ericksen NJ, Kenny GJ (2001) Are
floods getting worse in the Ganges, Brahmaputra and
Meghna basins? Environ Hazard 3:3748
Mooley DA, Parthasarathy B, Sontakke NA, Munot AA (1981)
Annual rain water over India, its variability and impact on
the economy. J Clim 1:167186
Mowla KG (1976) Relation between climatic fluctuation and
rice production in Bangladesh. In: Takahashi K, Yoshino
MM (eds) Climatic change and food production. Univer-
sity of Tokyo Press, Tokyo, p 137146
Parthasarathy B, Munot AA, Kothawale DR (1988) Regression
model for estimation of Indian foodgrain production from
summer monsoon rainfall. Agric Meteorol 42:167182
Parthasarathy B, Rupa Kumar K, Munot AA (1991) Evidence
of secular variations in Indian monsoon rainfall-circulation
relationships. J Clim 4:927938
Parthasarathy B, Rupa Kumar K, Munot AA (1992) Forecast of
rainy-season foodgrain production based on monsoon
rainfall. Indian J Agric Sci 62:18
Parthasarathy B, Munot AA, Kothawale DR (1994) All-India
monthly and seasonal rainfall series 18711993. Theor
Appl Climatol 49:217224
Paul BK, Rasid H (1993) Flood damage to rice crop in Bangla-
desh. Geogr Rev 83:150159
Rakhecha PR, Soman MK (1994) Trends in the annual extreme
rainfall events of 1 to 3 days duration over India. Theor
Appl Climatol 48:227237
Roy SS, Balling JRC (2004) Trends in extreme daily precipita-
tion indices in India. Int J Climatol 24:457466
Rupa Kumar K, Pant GB, Parthasarathy B, Sontakke NA
(1992) Spatial and subseasonal patterns of the long-term
trend of Indian summer monsoon rainfall. Int J Climatol
Selvaraju R (2003) Impact of El Nino-Southern Oscillation on
Indian foodgrain production. Int J Climatol 23:187206
Singh N, Sontakke NA (2002) On climatic fluctuations and
environmental changes of the Indo-Gangetic plains, India.
Clim Change 52:287313
Tanaka M (1976) Synoptic study on the recent climatic
change in Monsoon Asia and its influence on agricultural
production. University of Tokyo Press, Tokyo, p 81100
Turner AG, Innes PM, Slingo JM (2007) The effect of doubled
CO2 and model basic satte biases on the monsoon-ENSO
system. I. Mean response and interannual variability. Q J
R Meteorol Soc 133:11431157
Editorial responsibility: Mikhail Semenov,
Harpenden, UK
Submitted: July 27, 2007; Accepted: December 2, 2008
Proofs received from author(s): March 30, 2009
... Rice production substantially influences global food security as it is a crucial food for the daily life of the people (Ara et al. 2017). It is currently the staple food for over 4.7 billion people globally, and rice demand is expected to continue more in the future (Asada and Matsumoto 2009;Homma et al. 2014;Akinbile et al. 2020). Alternatively, climate variables, such as rainfall and temperature, may largely influence rice production and food security worldwide Page 2 of 26 Shrestha et al. ...
... To reveal relationships between climate factors and rice cultivation, numerous studies have been conducted at country or regional level in various locations (e.g., Saseendran et al. 2000;Sarker et al. 2012;Mainuddin et al. 2013;Zakaria et al. 2014;Chung et al. 2015;Le 2016;Pheakdey et al. 2017;Rene et al. 2016;Ara et al. 2017;Rahman et al. 2017;Ali 2018;Baliarsingh et al. 2018;Nath and Mandal 2018;Kingra et al. 2018;Stuecker et al. 2018;Panda et al. 2019;Abbas and Mayo 2021). Some used temperature and rainfall factors in their analyses (Lansigan et al. 2000;Alam et al. 2011;Rowhani et al. 2011;Masud et al. 2014;Bhatt et al. 2019;Ratnasiri et al. 2019;Rayamajhee et al. 2021), and some only used rainfall as a primary climate factor that affects the rice yield (Asada and Matsumoto 2009;Tiamiyu et al. 2015;Kunimitsu and Kudo 2015;Riad and Peter 2017;Ferreira et al. 2018;Rokonuzzaman et al. 2018;Singh et al. 2018;Chandrasiri et al. 2020;Molla et al. 2020). Most of the previous studies found that rainfall is one of the most essential climatic variables that affect the rice production in the tropics, because of its two-sided effects: (i) a lacking resource such as droughts and (ii) a catastrophic driver such as floods (Alam et al. 2011;Koudahe et al. 2017;Rokonuzzaman et al. 2018;Singh et al. 2018;Panda et al. 2019;Chandrasiri et al. 2020;Molla et al. 2020). ...
... Furthermore, we observed variability in spatial rainfall distribution patterns over the affected rice production areas. Accordingly, recent studies highlighted that rice production depended on rainfall patterns in the Ganges-Brahmaputra-Meghna River basin (Asada and Matsumoto 2009;Matsumoto and Asada 2020) and the Odisha region, India (Panda et al. 2019). ...
Full-text available
Climate is one of the main factors for rice crop growth. Understanding the relationship between climate variability and rice production during the period from 1901 to 1939 in Burma can give a clear picture of the impact of climate variability on rice yield since there were fewer human interventions on the catchment and almost no use of chemical fertilizer or high-yielding rice varieties at that time. However, the quantitative analysis of climate variability and its impact on rice production has not yet been paid sufficient scientific attention for the historic period. First, the changing trends of rainfall and rice yield between 1901 and 1939 were analyzed, including the effect of rainfall variability on rice production from multiple perspectives regarding rainfall characteristics, such as seasonal rainfall, various rainfall indices, rainfall anomalies, and monthly rainfall variability. Then, the relationship between rice yield and rainfall was investigated using multiple regression analysis to show how rainfall spatial and temporal variabilities have influenced rice yield and production, including essential factors that affected rice yield in each Burma district. The historical development of rice production in Burma during the period was also explored. Our findings indicate that not only the annual variability of rainfall, but also its monthly variability within a particular year likely influenced rice production. Excessive rainfall in the early or middle stage of crop growth or less during the early-middle or latter half of crop growth possibly caused the rice yield reduction in Burma during the colonial period. Furthermore, the results indicated that although rainfall anomalies widely differed from period to period, rice yield anomalies clearly showed the distinction of periods with higher or lower rice yields than average rice yield. Mostly higher than average rice yield was observed before 1910 in the Coastal Zone and before 1918 in the Delta, Dry, and Hilly Zones. The results of this study imply that selected rainfall indices could affect rice yield, positively or negatively, including the varied magnitude of their effects from one district to another, depending on climatic zones and agricultural ecosystems.
... Indian agriculture is highly dependent on the spatial and temporal distribution of monsoon rainfall (Wollenweber et al, 2003). Asada & Matsumoto (2009) analysed the relationship between district level crop yield data (rainy season 'kharif' rice) and precipitation for 1960-2000. It was shown that different regions were sensitive to precipitation extremes in different ways. ...
Full-text available
Evidence of changes in weather parameters like ambient temperature, precipitation, wind flow, etc., are prominently visible across the world. These changes have been reported to effect global crop yield. This review compiles both direct and indirect effects of climate change on global crop productivity with highlights on existing local and global scenarios. As a conclusion, it may be stated that thorough understanding of agricultural techniques and analysis of global change factors is highly essential for achieving sustainable agricultural yield over the upcoming years
... Moreover, the potential impacts of climate change and extreme climatic events are diverse and vary in scale (Chavez et al. 2015). For instance, paddy loss due to floods in Bangladesh and India is estimated at 4 MMt/year, which can feed 30 million people (Asada and Matsumoto 2009). ...
... The irony of flood and drought occurring concurrently almost every year, makes agriculture highly vulnerable and unstable. Many studies earlier have verified that scanty ISMR reflects low rice production (Webster et al. 1999;Selvaraju 2003;Krishna Kumar et al. 2004;Asada and Matsumoto 2009;Kumar et al. 2013;Singh et al. 2014;Nath et al. 2017;Mall et al. 2018). Some recent studies also pointed out the impact of anthropogenic emissions, as a possible mechanism of triggering drought events (Parth Sarthi et al. 2021). ...
The meteorological drought dynamics and its impacts on rice productivity has been evaluated for the Indian Summer Monsoon Rainfall (ISMR) season using the standardized precipitation index (SPI) over the middle Gangetic plains (MGP) of Bihar. The meteorological drought over the ISMR period was found to be a recurring phenomenon coinciding with the rice growing season over Bihar. The rice crop has an intensive water requirement; therefore, it is significantly impacted by the meteorological droughts. In the present study, spatiotemporal characteristics viz. intensity, frequency, and probability of meteorological drought has been assessed along with an investigation for significant trends and detection of regime shift points to identify the impact of drought on rice production. For the purpose, SPI-4 derived from high resolution gridded daily rainfall data (0.25° × 0.25°) from India Meteorological Department (IMD) has been considered to analyse the meteorological drought episodes over agro-climatic zones of Bihar from 1961 to 2019. The regime shifts were determined using the Rodionov test for the drought dynamics and production of rice in Bihar. A moderate to severe drought-prone zone was found over the zone BRZ3B; while zone BRZ2 and BRZ3A showed comparatively a greater number of mild drought events persisting with more than 70% probability of occurrence. An inkling of increasing dependency on groundwater is found, which is in turn governing the rice production regime. The present study shows there is a substantial need for climate resilience and food security policies incorporating the subtle linkage between SPI variability and crop production, especially over rice producing regions of the globe.
... The variation in the monsoon rainfall from year to year has a strong and significant impact on the variability of the wheat crop all over the world (Webster et al. 1998). The variation in the rainfall during the summer season from year to year has a strong and significant impact on the variability of crops in India (Asada and Matsumoto 2009). ...
Full-text available
Climate change is bringing drastic changes to the food availability, accessibility, quality, and stability in the world. Pakistan heavily relies on production of wheat for food security. This study investigates the role of climate change on food security over Punjab regions, Pakistan, from 1979–2020. The study utilized the total production of wheat to measure food security. Moreover, the study utilized mean, maximum, minimum temperature, wind speed, and rainfall as a measurement of climate change. This study utilized auto-regressive distributed lag (ARDL) bound test cointegration approach to test the long-run cointegration, while ARDL model is applied to investigate the short-run relationship among modeled variables. This study adopted Augmented Dicky Fuller (ADF) and Phillips Parron (PP) test to check the stationarity of the data. Moreover, the study ensures reliability and validity of the model by utilizing Breusch-Godfrey serial correlation, Breusch-Pagan-Godfry HSK, and Ramsey RESET test. This study found that wheat cultivated area (0.46*), total irrigated area (2.67***), total un-irrigated area (1.93***), and total area sown ([1.0809**) have a positive and significant long-run impact on food production in all regions of Punjab Pakistan. However, while average, the maximum and minimum temperature is negatively and significantly associated with food production in all regions of Punjab (− 1.07**; − 3.33***; − 1.84**), except northern Punjab, where maximum temperature affects positively food production Punjab. Rainfall negatively and significantly affects food production in northern (− 0.34**) and central and southern (− 0.13***) Punjab. Furthermore, wind speed negatively affects food production in all regions of Punjab (− 0.22**; − 0.21*; − 0.11**), except northern Punjab. This study implies that government should develop policies to increase irrigation facilities and loan facilities to increase the total area sown that will help to increase wheat yield and ensure food security. Moreover, the government should devise policies for large-scale plantations to minimize climate change impacts. The study also suggests new improved varieties of the wheat crop that can survive and flourish in the presence of adverse climatic changes, high temperature, and high wind speed.
... Moreover, the potential impacts of climate change and extreme climatic events are diverse and vary in scale (Chavez et al. 2015). For instance, paddy loss due to floods in Bangladesh and India is estimated at 4 MMt/year, which can feed 30 million people (Asada and Matsumoto 2009). ...
Full-text available
Climate change and extreme climatic events cause massive food grain loss, which adversely impacts food availability, food import, and other economic factors. Bangladesh has been combating climate variability and extreme climatic events to abate agricultural production loss for a long time. In this article, we investigated the relationship between food grain loss and food security with reference to extreme climatic events in Bangladesh, based on data from 1984 to 2017. We used the vector auto-regression (VAR) model and derivative analyses, and suggested policy implications related to the existing national agricultural policy. Five time-series variables were judiciously considered: food availability, food loss, food import, gross domestic product (GDP) growth rate, and inflation rate. The results show that the variable, food grain loss has a reverse association with food security—it escalates import from the world food market, causing import-dependence. Moreover, food loss significantly instigates inflation. However, the GDP growth rate was found to be a weak provocateur. Overall, climate change and climatic extremes jeopardize the country’s food security and hinder its pursuit of the sustainable development goals, especially stand-alone goals 1 and 2. Therefore, it is concluded that changes in climate and their correlations are detrimental to Bangladesh’s food security and economy.
... The monsoon rainfall pattern, both geographically and temporally, is critical for Indian agriculture (Kumar et al., 2004). Asada and Matsumoto (2009) looked at the relationship between district-level crop output statistics (rainy season 'kharif' rice) and precipitation from 1960 to 2000.In a number of scenarios, different locations were shown to be sensitive to precipitation extremes. Crop productivity in the upper Ganges basin is linked to total precipitation throughout the relatively short growing season, rendering it sensitive to drought. ...
... The effect of climate is heterogeneous in both spatial and temporal dimensions (Gornall et al. 2010;Leng and Huang 2017;Zhao et al. 2017;Kukal and Irmak, 2018;Kumar and Kaur 2019;Chang et al. 2019;Kurths et al. 2019;Ray et al. 2019). It is well documented in literature that Indian agricultural growth is highly dependent on the spatial and temporal distribution of monsoon rainfall and temperature (Kumar et al. 2004;Asada and Matsumoto 2009;Shukla et al. 2018). In view of the above, we analyze the heterogeneous impact of climate change on crop yields in the undivided state of Andhra Pradesh, India over different time horizons. ...
Full-text available
Climate change is a major concern the world over more so for a predominantly agrarian country like India. In this paper we analyze the time horizon dynamics of crop and climate variables at the regional level in India. We also analyze the co-movements of crop yields with temperature and rainfall to observe the coherence across heterogenous time horizons. We employ Bai-Perron structural break and Continuous wavelet transform methods on yearly data of seven crop yields and climate variables. Observed variables are analyzed from 1956 to 2010 for the un-divided state of Andhra Pradesh, India. Breakpoint analysis shows an increase of around 1.0° temperature with two observed break points. Rainfall depicts no systematic change with fluctuations being largely random. The framework of wavelets-based time–frequency analysis employed in this study captures climate and crop dynamics at heterogeneous time horizons, allowing one to study the impact of climate and crop yields at both short and longer time-horizons. Wavelet based coherence analysis exhibited significant co-movement between climatic and crop variables. Given shifts in climate patterns and subsequent shifts in co-movements across time horizons at the regional level, policy makers and crop scientists should design time specific and locally viable adaption and mitigation policies to tackle the impact of climate change on crops and livelihoods.
Combining seven years of household data from an original field experiment in villages of Jharkand, East India, with meteorological data, this paper investigates how Indian Self-Help Groups (SHGs) enable households to withstand rainfall shocks. I show that SHGs operate remarkably well under large covariate shocks. While credit access dries up in control villages one year after a bad monsoon, reflecting strong credit rationing from informal lenders, credit flows are counter-cyclical in treated villages. Treated households experience substantially higher food security during the lean season following a drought and increase their seasonal migration to mitigate expected income shocks. Credit access plays an important role, together with other SHG aspects such as peer networks. These findings indicate that local self-help and financial associations can help poor farmers to cope with climatic shocks and to implement risk management strategies.
Global wetlands and floodplains offer benefits and perils alike for human society. For example, humans rely on natural flood cycles for fisheries and agriculture, yet flooding also caused nearly one trillion USD in damage in the past 30 years and impacts millions of people every year. Looking forward, altered flow regimes or increased drought conditions are expected to affect the natural inundation cycle and its ecosystem services. The current and potential future impacts of flooding and drying events warrant increasing efforts to quantify our dependence and exposure within flooded areas, since any change from current inundation patterns is expected to have consequences for those who rely on regular flood occurrences. This paper provides a baseline global assessment of the dependence and exposure of human populations, urban areas, roads, and agriculture on current inundation patterns. The analysis uses a spatially explicit inundation map at ∼500 m resolution (GIEMS-D15) derived from satellite remote sensing to represent flooding extents and overlays it with current population and land use maps. We find that 35% of the analyzed population, or 2.0 billion people, live inside areas that are prone to inland flooding, which comprise only 12% of the land surface area (excluding marine coastal areas), confirming that population densities within inundation zones are about three-times above global average. Likewise, 35% of urban areas potentially experience regular, seasonal, or infrequent flooding. Agriculture shows a similar pattern with 24% of the world’s cropland in areas of recurring inundation. Finally, we estimate that 18% of the global road network is exposed to inundation during high water periods. These global estimates demonstrate a preferential tendency of human populations, infrastructure, and agriculture to be co-located within inundation areas, making related anthropogenic activities highly susceptible to future changes in flood regimes. The results are intended to offer a suite of first-order estimates as partial input to more holistic risk and vulnerability assessments and to ultimately improve environmental planning and policy at large scales.
Full-text available
During the recent decade, with the growing recognition of the possibility of climate change and clear evidence of observed changes in climate during 20 th century, an increasing emphasis on food security and its regional impacts has come to forefront of the scientific community. In recent times, the crop simulation models have been used extensively to study the impact of climate change on agricultural production and food security. The output provided by the simulation models can be used to make appropriate crop management decisions and to provide farmers and others with alternative options for their farming system. It is expected that in the coming decades with the increased use of computers, the use of simulation models by farmers and professionals as well as policy and decision makers will increase. In India, substantial work has been done in last decade aimed at understanding the nature and magnitude of change in yield of different crops due to projected climate change. This paper presents an overview of the state of the knowledge of possible effect of the climate variability and change on food grain production in India.
Full-text available
The long-term variations of rice production in Bangladesh are examined in the latter half of the 20th century and the impact of recent severe floods is discussed. Unprecedented severe floods, which submerged nearly two thirds of the country, occurred recently in Bangladesh in 1988 and 1998, and the effects of these severe floods to rice cropping are revealed. In these severe flood years, the rainy season crop aman drastically declined in production due to the decrease of the cultivated area and yield. On the other hand, the dry season crop boro, which is planted after the flood withdrawal, increased its production much higher than that of the previous year, mainly because of utilization of the residual flood water. In terms of total annual rice production, it turns out that severe floods have even a positive effect on rice production in Bangladesh. Furthermore, rice production after these severe flood years is higher than that of the previous years, suggesting that severe floods may act as a trigger for increasing rice production level through the change of the hydrological environment and farmers' reaction to it.
Full-text available
Empirical evidence is presented to support a hypothesis that the interdecadal variation of the Indian summer monsoon and that of the tropical SST are parts of a tropical coupled ocean-atmosphere mode. The interdecadal variation of the Indian monsoon rainfall (IMR) is strongly correlated with the interdecadal variations of various indices of El Niño-Southern Oscillation (ENSO). It is also shown that the interannual variances of both IMR and ENSO indices vary in phase and follow a common interdecadal variation. However, the correlation between IMR and eastern Pacific SST or between IMR and Southern Oscillation index (SOI) on the interannual timescale does not follow the interdecadal oscillation. The spatial patterns of SST and sea level pressure (SLP) associated with the interdecadal variation of IMR are nearly identical to those associated with the interdecadal variations of ENSO indices. As has been shown earlier in the case of ENSO, the global patterns associated with the interdecadal and interannual variability of the Indian monsoon are quite similar.The physical link through which ENSO is related to decreased monsoon rainfall on both interannual and interdecadal timescales has been investigated using National Centers for Environmental Prediction-National Center for Atmospheric Research reanalysis products. The decrease in the Indian monsoon rainfall associated with the warm phases of ENSO is due to an anomalous regional Hadley circulation with descending motion over the Indian continent and ascending motion near the equator sustained by the ascending phase of the anomalous Walker circulation in the equatorial Indian Ocean. It is shown that, to a large extent, both the regional Hadley circulation anomalies and Walker circulation anomalies over the monsoon region associated with the strong (weak) phases of the interdecadal oscillation are similar to those associated with the strong (weak) phases of the interannual variability. However, within a particular phase of the interdecadal oscillation, there are several strong and weak phases of the interannual variation. During a warm eastern Pacific phase of the interdecadal variation, the regional Hadley circulation associated with El Niño reinforces the prevailing anomalous interdecadal Hadley circulation while that associated with La Niña opposes the prevailing interdecadal Hadley circulation. During the warm phase of the interdecadal oscillation, El Niño events are expected to be strongly related to monsoon droughts while La Niña events may not have significant relation. On the other hand, during the cold eastern Pacific phase of the interdecadal SST oscillation, La Niña events are more likely to be strongly related to monsoon floods while El Niño events are unlikely to have a significant relation with the Indian monsoon. This picture explains the observation that the correlations between IMR and ENSO indices on the interannual timescale do not follow the interdecadal oscillation as neither phase of the interdecadal oscillation favors a stronger (or weaker) correlation between monsoon and ENSO indices.
Temporal and spatial patterns of damage to rice crops in Bangladesh resulting from river flooding are analyzed countrywide for the period 1962 to 1988 and at the district level from 1967 to 1988. Floods annually damaged approximately 4 percent of total rice production, but districts varied markedly in flood-related crop losses. Absolute magnitude of these losses over time may not justify expensive and controversial flood-control initiatives currently being implemented in Bangladesh.
The disastrous floods in Bangladesh in 1987 and 1988 captured world-wide attention. That country is particularly prone to natural disasters which constantly undermine government and international efforts to improve social and economic conditions. The floodplains which occupy 80 per cent of Bangladesh have diverse characteristics and are affected by flash floods, river floods and rainwater floods to different extents. The 1987 floods were predominantly rainwater floods caused by exceptionally heavy monsoon rainfall over northern parts of the country. The 1988 floods were mainly river floods caused by heavy monsoon rainfall over a wider area of the Ganges and Brahmaputra river catchments (more than 90 per cent of which lie outside Bangladesh). In both years, breaching or cutting of embankments aggravated flooding. Despite considerable crop damage, there were compensatory increases in production in areas not affected by the floods and in the following dry season.
Long-term global gridded datasets of observed precipitation are essential for the analysis of the global water and energy cycle, its variability, and possible changes. Several institutions provide those datasets. In 2005 the Global Precipitation Climatology Centre (GPCC) published the so-called Variability Analysis of Surface Climate Observations (VASClimO) dataset. This dataset is especially designed for the investigation of temporal change and variability. To date, however, the GPCC has not published how this dataset has been produced. This paper aims to fill this gap. It provides detailed information on how stations are selected and how data are quality controlled and interpolated. The dataset is based only on station records covering at least 90% of the period 1951–2000. The time series of 9343 stations were used. However, these stations are distributed very inhomogeneously around the globe; 4094 of these stations are within Germany and France. The VASClimO dataset is interpolated from relative deviations of observed monthly precipitation, leading to considerably lower interpolation errors than direct interpolation or the interpolation of absolute deviations. The retransformation from interpolated relative deviations to precipitation is done with local long-term averages of precipitation interpolated from data of the Food and Agriculture Organization of the United Nations. The VASClimO dataset has been interpolated with a method that is based on local station correlations (LSC) that is introduced here. It is compared with ordinary kriging and three versions of Shepard's method. LSC outperforms these methods, especially with respect to the spatial maxima of interpolation errors.