Theor Appl Climatol

DOI 10.1007/s00704-015-1688-3

ORIGINAL PAPER

Climate change in Bangladesh: a spatio-temporal analysis

and simulation of recent temperature and rainfall data using GIS
and time series analysis model
Md. Rejaur Rahman 1,2 & Habibah Lateh 1

Received: 10 April 2014 / Accepted: 27 November 2015

# Springer-Verlag Wien 2015

Abstract In this paper, temperature and rainfall data series temperature exhibited an increase of 0.018 °C per year in
were analysed from 34 meteorological stations distributed 2011–2020, and if this trend continues, this would lead to
throughout Bangladesh over a 40-year period (1971 to 2010) approximately 1.0 °C warmer temperatures in Bangladesh
in order to evaluate the magnitude of these changes statistical- by 2020, compared to that of 1971. A greater rise is projected
ly and spatially. Linear regression, coefficient of variation, for the mean minimum (0.20 °C) than the mean maximum
inverse distance weighted interpolation techniques and geo- (0.16 °C) temperature. Annual rainfall is projected to decline
graphical information systems were performed to analyse the 153 mm from 2011 to 2020, and a drying condition will per-
trends, variability and spatial patterns of temperature and rain- sist in the northwestern, western and southwestern parts of the
fall. Autoregressive integrated moving average time series country during the pre- and post-monsoonal seasons.
model was used to simulate the temperature and rainfall data.
The results confirm a particularly strong and recent climate Keywords Climate change . Temperature . Rainfall . Trends .
change in Bangladesh with a 0.20 °C per decade upward trend Variability . GIS . ARIMA . Spatial analysis
of mean temperature. The highest upward trend in minimum
temperature (range of 0.80–2.4 °C) was observed in the north-
ern, northwestern, northeastern, central and central southern 1 Introduction
parts while greatest warming in the maximum temperature
(range of 1.20–2.48 °C) was found in the southern, southeast- Climate change is the greatest threat that our Earth faces today.
ern and northeastern parts during 1971–2010. An upward A warming planet impacts various aspects, which includes the
trend of annual rainfall (+7.13 mm per year) and downward weather system, hydrology, ecology and environment.Climate
pre-monsoon (−0.75 mm per year) and post-monsoon rainfall change refers to a change in the state of the climate that can be
(−0.55 mm per year) trends were observed during this period. identified (e.g., using statistical tests) by changes in the mean
Rainfall was erratic in pre-monsoon season and even more so or the variability of its properties, which persists for an ex-
during the post-monsoon season (variability of 44.84 and tended period, typically decades or longer (IPCC 2007). The
85.25 % per year, respectively). The mean forecasted classical period for averaging these variables is 30 years, as
defined by the World Meteorological Organization (WMO). It
was observed that, in general, developing countries are more
* Md. Rejaur Rahman vulnerable to the effects of climate change than developed
rejaur2001@yahoo.com countries, mainly because of the low capacity to adapt to cli-
* Habibah Lateh mate change in the developing world. According to the fifth
habibah@usm.my Intergovernmental Panel on Climate Change (IPCC) report,
significant trends in temperature and precipitation were ob-
1
School of Distance Education, Universiti Sains Malaysia (USM), served around the world but with different magnitudes
Gelugor, Malaysia (IPCC 2014). Globally, surface temperatures are rising, but
2
Department of Geography and Environmental Studies, University of warming is not uniform all over the globe (Jones 2001;
Rajshahi, Rajshahi, Bangladesh Jones and Moberg 2003; Solomon et al. 2007; Kerr 2009;
 Rahman M.R., Lateh H.

Nick et al. 2009; Lorentzen 2014). In the fifth assessment of impacts will therefore ultimately depend on the availability of
IPCC, it was stated that the average temperature of the Earth accurate and cost-effective forecasts.
has risen approximately 0.85 °C from 1880 to 2012 and tem- Several studies have been carried out over different parts of
perature of the world would increase between 0.3 and 4.8 °C the world with regard to the changing temperature and rainfall
by the end of the twenty-first century (Hartmann et al. 2013). patterns (Bloomfield 1992; McCarthy et al. 2001; IPCC 2001;
Furthermore, coupled with global warming, there is strong Solomon et al. 2007; IPCC 2007; Reiter et al. 2012; Mair et al.
evidence that changes in rainfall patterns have already taken 2013; IPCC 2014). In recent years, in some parts of the globe,
place on both global (Hulme et al. 1998) and regional scales investigations of climatic parameters were also assessed with
(Yu and Neil 1993; Rodrı′guez-Puebla et al. 1998; Trenberth weather station data to detect temperature and precipitation
et al. 2007). Variations to regional and local climate depend on trends (Domonkos and Tar 2003; Matulla et al. 2003; Franke
regional and local features; therefore, climate change at re- et al. 2004; Founda et al. 2004; Begert et al. 2005; Feidas et al.
gional and local scales often does not match those on a global 2007; Good et al. 2008; Zhang et al. 2008; Brázdil et al. 2009;
scale. In this regard, the assessment of climate variability and Schaefer and Domroes 2009; Tabari and Hosseinzadeh 2011;
change on a smaller scale (country level) is a key issue and this Pérez and Jury 2013). Moreover, recently, the extent of future
will help to improve our understanding of long-term climate temperature and rainfall change has also been estimated by
variability and change and as well as the associated mecha- using different simulation techniques (Hansen et al. 2006;
nisms of forcing of change at country level or local scales. Rahmstorf et al. 2007; Diomede et al. 2008; Lin et al. 2009;
Like global climate change, the climate in Bangladesh is Marengo et al. 2009; Karmalkar et al. 2011; Afshin et al. 2011;
also changing since it is becoming more unpredictable in re- IPCC 2007, Esfahani and Friedel 2014). In these studies, re-
cent years (Rahman 2013). Similar to many other countries, searchers mainly focused on long-term forecasting by apply-
Bangladesh will face tremendous challenges from climate ing intelligent methods, earth models of intermediate com-
change, particularly since it is mostly an agrarian country with plexity (EMICs), energy balance models (EBMs), global/
a high population. Therefore, since its agriculture and water general circulation models (GCMs), providing regional cli-
are the most vulnerable sectors to climate change which can mates for impacts studies (PRESIS) regional model, support
lead to floods, droughts, tornados, cyclones, tidal surges and vector regression, fuzzy inference system, hybrid approach,
soil salinities, climate change scenarios need to be incorporat- the combination of artificial neural network, fuzzy logic and
ed into the development and planning activities of the country wavelet functions. However, these methods/models have their
(Rahman 2013). The IPCC has reported and highlighted the own difficulties, complexity as well as large data require-
need for more detailed information about climate change on ments. Moreover, it is well known that the potential of any
regional and local scales, which is of particular interest to global and regional climate model is limited in simulating
various nations and economic groups (IPCC 2014). local climatic (rainfall and temperature) characteristics (Dash
Therefore, country-level information about climate variability et al. 2013). Furthermore, global and regional models are
and change is needed to adapt with the changes and formulate complicated and not easy to use because of huge data require-
appropriate strategies to overcome these problems. On the ment and current knowledge of the physical climate system
other hand, spatial distribution of meteorological data is be- (IPCC 2001). Therefore, many efforts have been made in the
coming important as inputs to spatially explicit landscape, statistical modelling of temperature and rainfall data using
regional, and global models (Li et al. 2006; Rahman et al. historical records (Hansen et al. 2006; Rahmstorf et al.
2014). Spatial technologies, such as geographic information 2007). Time series models (Lee and Sohn 2007) have become
system (GIS) and numerical modelling techniques, have been quite popular due to its trend-detecting capability. Among
developed as powerful tools for ecological and environmental others, autoregressive integrated moving average (ARIMA)
assessments (Paudyal 1996; Boyle et al. 1998; Krivtsov 2004; time series model is one of the most important to forecast data
Rahman and Saha 2007; Rahman and Saha 2008; Rahman statistically (Romilly 2005). In general, ARIMA processes
and Saha 2009; Rahman et al. 2009; Rahman et al. 2014; have several advantages over other methods including its
Rahman et al. 2015). On the other hand, within the context capability to simulate the path of forecasting, its rich infor-
of the general climate discussion, the evaluation of climate mation on time-related changes and the consideration of
time series is growing in importance. Predicting weather con- serial correlations between observations (Yurekli et al.
ditions using previous data is one of the real uses of simula- 2007). Previously, the ARIMA models were most notably
tion. Most of the weather forecasters use this information pub- used in various scientific and engineering applications but
lished by weather bureaus and these kinds of simulations can not too often for the prediction of temperature and rainfall
help predict weather conditions. Numerical weather prediction trends over tropical areas. Recently, however, the ARIMA
for forecasting involves complicated numeric computer model has been used by Baker-Austin et al. (2012) to pro-
models to predict weather accurately by taking many param- ject sea surface temperature (SST) data in the Baltic Sea up
eters into account. Effective management of climate change to year 2050.
Climate change in Bangladesh: a spatio-temporal analysis

In reviewing the relevant studies in Bangladesh, there is average), respectively. January is the coldest and April and
limited information about past climate change at the national May are the hottest months in Bangladesh (Fig. 2a).
level (Met Office 2011). In the past, studies were carried out The historical average rainfall of the country is 2428 mm
mainly on the trend of climate change over Bangladesh per year (BMD 2013). and the rainfall is very much seasonal
(Divya 1995; Mia 2003; Karmakar and Shrestha 2000; in Bangladesh (Fig. 2b), which varies from 1400 to 4400 mm.
Shahid 2009; Shahid 2010; Shahid 2011; Rahman 2013; The highest rainfall occurs in June, July, and August. More
Hasan et al. 2014; Rahman and Lateh 2015). However, there than 75 % of the total rainfall in Bangladesh occurs during the
is no such detail on spatial temporal analysis of potential cli- monsoon season, caused by winds blowing from the Southern
mate change, particularly on temperature and rainfall, from Hemisphere from mid-May to September, which accumulates
recent data in Bangladesh. Moreover, few studies are available moisture and deposits copious amounts of precipitation over
that predict rainfall and temperature changes over Bangladesh the South Asian continent. In respect to the global warming
using climate or statistical models, and in addition, those are and climate change, Bangladesh is one of the most vulnerable
mostly focused on long-term projections (May 2004; Dash countries in the world due to its least capacity to address the
et al. 2006; Immerzeel 2007; Kripalani et al. 2007; Islam devastating impacts (IPCC 2007). Recently, Bangladesh is
2009; Rahman et al. 2012a). However, impact communities experiencing higher temperatures, more variability in rainfall,
are more interested on shorter time scales prediction, such as more extreme weather events and sea level rise. Bangladesh is
one or two decades as well as at the regional and local scales in highly vulnerable, because it is low-lying, located on the Bay
order to assess the impacts of climate change and to develop of Bengal in the delta of the Ganges, Brahmaputra and
suitable adaptation and mitigation policies (Giorgi 2005). Meghna and also densely populated. Since agriculture is the
Unfortunately, the optimum time and spatial scales for climate mainstay of the economy of Bangladesh, its agriculture and
change are not well studied in Bangladesh with recent data. water sectors are very sensitive to impacts of the climate
Thus, in this paper, the temperature and rainfall data from the change. Thus, detail study about spatio-temporal analysis of
last four decades (1971 to 2010) in Bangladesh were analysed temperature and rainfall over Bangladesh is a key issue and
in order to evaluate the magnitude of their changes statistically short-term prediction of these two main climatic parameters is
and spatially. Specifically, trend, variability and spatial patterns important for adaptation and planning to cope up the problems
of the mean, mean minimum and mean maximum temperatures due to climate change at present and future as well.
and annual, pre-monsoon and post-monsoon rainfalls were
assessed and analysed. A short-term (one decade, 2010–
2020) prediction of temperature and rainfall using an ARIMA 3 Data and methods
time series model was evaluated and analysed spatially using
GIS. This was done since the first phase of research in climate The monthly dataset of the minimum and maximum temper-
change is to use scientific criteria to verify that a ‘change’ has atures and rainfall from 34 stations in Bangladesh during the
actually been taken place (Lorentzen 2014) and later modelling period 1971–2010 that was used and analysed in this study
the variations of climate change in order to make dependable was provided by the Bangladesh Meteorological Department
forecasts for sound environmental policies (Romilly 2005). (BMD 2013). and the locations of the 34 weather stations are
shown in Fig. 1. However, out of 34 weather stations, data
from 29 stations was available from the last 40 years while
2 Study area from the remaining five stations, namely Tangail (1987–
2010), Kutubdia (1985–2010), Chuadanga (1989–2010),
The spatial extent of Bangladesh is between 20° 34′ N to 26° Mongla (1989–2010) and Sayedpur (1991–2010), data was
38′ N latitude and 88° 01′ E to 92° 41′ E longitude (Fig. 1) only available from the past 20 to 26 years. Therefore, a
with an area of 144,000 km2. Bangladesh has a sub-tropical shorter series of measurements (20 to 26 years) was used
humid climate characterized by wide seasonal variations in during analysis and we considered these five stations in our
rainfall, moderately warm temperature and high humidity. study because of no other stations were located in those par-
Three distinct seasons can be recognized in Bangladesh, (i) ticular areas. After processing the data from each station, there
the dry winter/post-monsoon season from November to was some missing data, which amounted to less than 2 %.
February, (ii) the pre-monsoon hot summer season from When missing data was found, this was replaced by the aver-
March to May and (iii) the rainy monsoon season from June age value of the same month but from the previous and sub-
to October (Rashid 1991). The historical average temperature sequent years. The data was visually examined using histo-
of the country is 25.75 °C, with a range of 18.85 to 28.75 °C grams for any potential outliers as well as compared with
(monthly average). The average minimum and maximum neighbouring weather stations. However, no significant anom-
temperatures are 21.18 and 30.33 °C, and varies from 12.5 alies were found. Additionally, subjective double mass curve
to 25.7 °C (monthly average) and 25.2 to 33.2 °C (monthly (Kohler 1949) and the objective student’s T test (Panofsky and
 Rahman M.R., Lateh H.

Fig. 1 Study area with location

of weather stations

Brier 1968; Shahid 2010) were performed to test homogeneity distribution of temperature and rainfall and their trends
of temperature and rainfall time series data. There were no and variability, inverse distance weighted (IDW) interpola-
statistically significant variations existing in the time series tion technique and geographical information systems (GIS)
data. Besides, in 1971, most of the stations have missed large were applied. Integrated land and water information system
number of observed data. Thus, data for the year 1971 was (ILWIS), a GIS and image processing software (ILWIS
replaced by the data from 1970 in the data set. 2005). was used for this purpose. ILWIS and IDW were also
A linear regression analysis using the least square method was used to generate the spatial pattern of the forecasted temper-
applied to detect any trends in temperature and rainfall time series ature and rainfall. In this study, autoregressive integrated
data and the confidence level of 95 % was taken as the threshold. moving average (ARIMA) time series model, which was
Details of the linear regression and least square methods can be popularized by Box and Jenkins (1976), was used and the
found in Moore and McCabe (2003). To measure the tempera- monthly dataset of 34 stations was used to simulate temper-
ture and rainfall variability, a coefficient of variation (Cv) was ature and rainfall during 2011–2020. The details of the
used and calculated for the period of 1971 to 2010. The Cv ARIMA model can be found in Box and Jenkins (1976),
represents the ratio of the standard deviation (σ) to the mean Yurekli et al. (2007) and Brockwell and Davis (1991).
(x
), which is useful for comparing the degree of variation from However, ARIMA models are, in theory, the most general-
one data series to another. Often, the coefficient of variation is ized class of models for forecasting time series. In the
expressed as a percentage and can be represented by Eq. 1. ARIMA (p, d, q) processes, p is an auto regressive (AR)
σ retrospectively weighted series, d is a difference lag (I) to
C v ¼
 100 ð1Þ achieve series stationary, and q is a retrospectively weighted
x
moving average (MA) random error series. The appropriate
For trend and variability analysis, Excel software (Excel values of autoregressive order p and moving average q are
version 2010) was used. For mapping the spatial chosen by examining the autocorrelation function (ACF)

Fig. 2 a Average temperature

(°C) and b rainfall (mm) of
Bangladesh (based on data
provided by BMD, 2013)
Climate change in Bangladesh: a spatio-temporal analysis

and partial autocorrelation function (PACF) of the time se- the temperature and rainfall for the decade of 2001–2010 and
ries. ARIMA model can be represented by Eq. 2. then the forecasted values were compared with the actual ob-
served values for the decade of 2001–2010. After validation
ϕðBÞ∇d Y t ¼ θðBÞεt ð2Þ and acceptance, finally, the temperature and rainfall were fore-
casted for the decade 2011–2020 using the validate model.
where
The overall methodology of the study is shown in Fig. 3.
ϕðBÞ ¼ 1
ϕ1 B
ϕ2 B2
…
ϕp Bp ð2
1Þ
θðBÞ ¼ 1
θ1 B
θ2 B2
…
θq Bq ð2
2Þ
4 Results and discussion
where Yt and εt represent time series and random error terms at
time t, respectively. B is the backward shift operator, ∇d de-
4.1 Trends of temperature and rainfall
scribes differencing operation to data series to make the data
series stationary, d is the number of differencing. ϕp and θq are
The annual mean, mean minimum and mean maximum tem-
the model parameters, and the ϕ(B) and θ(B) are of order p and
peratures and annual, pre-monsoon (MAM months) and post-
q. The time series often contained seasonal effects, to which
monsoon (NDJF months) rainfall were considered for analysis
seasonal differencing is often used to remove these effects.
in this study. From the last 40 years (1971–2010), weather
Thus, ARIMA models have two general forms: non-
data analysis shows that the annual mean, mean minimum
seasonal ARIMA (p, d, q) and multiplicative seasonal
and mean maximum temperatures of Bangladesh were
ARIMA (p, d, q) (P, D, Q)S, where p, d and q are non-
25.83, 21.21 and 30.44 °C, respectively. The obtained annual
seasonal parts of the model, and P, D and Q are seasonal parts
mean, mean minimum and mean maximum temperatures
of the model (Box and Jenkins 1976). which can be represent-
showed a positive trend of about 0.020, 0.018 and 0.022 °C
ed by Eq. 3.
per year, respectively. These results differ with those of Jones
ϕðBÞϕðBs Þð1
BÞd ð1
Bs ÞD Y t ¼ C þ θðBÞθðBs Þεt ð3Þ

where

ϕðBs Þ ¼ 1
ϕ1 Bs
ϕ2 B2s
…
ϕP BPs ð3
1Þ

θðBs Þ ¼ 1
θ1 Bs
θ2 B2s
…
θQ BQs ð3
2Þ

where ϕ(Bs) is the seasonal autoregressive part of order P

(ARs(P)), θ(Bs) is the seasonal moving average part of order
Q (MAs(Q)), (1-B)d is the differencing of order d (I(d)), (1-
Bs)D is the seasonal differencing of order D (Is(D)) and s is the
period of the seasonal pattern appearing. In this study, Expert
Modeller with ARIMA model of time series module in SPSS
statistical analysis software was used to perform the simula-
tion of temperature and rainfall data for each station. In SPSS
software, the Expert Modeller automatically finds the best-
fitting model for each dependent series and considers both
seasonal and non-seasonal models. In the process, if indepen-
dent (predictor) variables are specified, the Expert Modeller
selects those that have a statistically significant relationship
with the dependent series. Model variables are transformed,
where appropriate, using differencing and/or a square root or
natural log transformation. ARIMA model involves explicitly
specifying autoregressive and moving average orders, as well
as the degree of differencing. It is possible to include indepen-
dent (predictor) variables and define transfer functions for any
or all of these in the ARIMA model and can also specify
automatic detection of outliers or specify an explicit set of
outliers (SPSS 2007). To validate the ARIMA model, the first
30 years of observed data (1971–2000) was used to forecast Fig. 3 Flowchart of the methodology
 Rahman M.R., Lateh H.

(1995), Ahmad and Warrick (1996), Shahid (2010), and Met mean maximum temperature (0.03–0.06 °C per year) was no-
Office Hadley Centre (Met Office 2011). For example, Jones ticed in the southern, southeastern and northeastern parts
(1995) found no significant change in the annual mean mini- (Fig. 4c). It was also found that the mean temperature trend
mum and mean maximum temperatures of Bangladesh, while was remarkably positive (0.02–0.042 °C per year) in the
Ahmad and Warrick (1996) mentioned that the trend of the southern, southeastern, northeastern and extreme northwest-
mean temperature of Bangladesh was 0.5 °C over 100 years ern parts of the county (Fig. 4a). The maximum positive trend
(0.005 °C per year). Shahid (2010) found that the mean, mean was noticed at Cox’s Bazar, Sylhet and Dinajpur stations for
minimum and mean maximum temperatures increased by the mean temperature. On the other hand, a maximum upward
0.0097, 0.0091 and 0.0102 °C per year, respectively, in the trend was found at the Rangpur and Dinajpur stations for the
last 50 years (1958–2007). McSweeney et al. (2010) men- mean minimum and Cox’s Bazar and Sitakunda stations for
tioned that the surface air temperature in Bangladesh has the mean maximum (Fig. 4). A downward trend of the mean
warmed at a rate of 0.002, 0.007 and 0.012 °C per year for minimum temperature was observed at the Rangamati station,
the annual, June–August (JJA), and September–November situated in the southeastern part of Bangladesh (Fig. 4b). A
(SON) months, respectively. According to Met Office downward trend of the mean minimum temperature in this
Hadley Centre (Met Office 2011), that since 1960, the mean area is unknown and further studies are needed to investigate
temperature trend was positive during both the summer and the causes and their influences on the seasonal patterns. Thus,
winter seasons in Bangladesh at 0.019 and 0.024 °C per year, it may be said that the northern, northwestern, southern, south-
respectively. The difference between the present and previous eastern and, to some extent, central parts of the country are
findings might be due to the use of long-term (50 years) data warming at an alarming rate. This may be linked to the impact
as well as the absence of recent data in their analysis which of global warming and climate change.
was included in this study. The rate of change of temperature On the other hand, data analysis during the 1971–2010
was more accelerated in the last 30 years (Table 1). The shorter period shows that the average annual, pre-monsoon (MAM
analysis datasets exhibit faster warming and is clearly found in months) and post-monsoon (NDJF months) rainfall were
Table 1 that maximum temperature was warmed dramatically 2387, 446 and 77 mm, respectively. Analysis by least square
over the last 30-year period. Moreover, the changing rate of fitting for the rainfall data reveals the following results: during
the maximum temperature was higher than that of the mini- the investigated period, the annual rainfall rose by 7.13 mm
mum temperature (0.022 vs. 0.018 °C per year for 1971– per year (a total of 14 %), while pre-monsoon and post-
2010). Therefore, recent data analysis is likely a key issue to monsoon rainfall declined by −0.75 mm (total of 17 %) and
find out the valid trends of the climatic parameters. −0.55 mm (total of 39 %) per year, respectively. Thus, the
The spatial pattern of temperature trend is shown in Fig. 4, general trend was positive for the annual and negative for
which was generated using calculated trend statistics for the seasonal rainfall in the country. The greatest decrease in rain-
34 weather stations, IDW and GIS. As seen in Fig. 4, all of the fall over the year occurred during the post-monsoonal season
weather stations presented a positive trend, except at the (Table 1). Analysis further shows that the rate of rainfall
Rangamati and Sitakunda stations for the mean minimum change (annual and seasonal) decelerated in the last 30 years
and the Mymensingh station for the mean maximum temper- (Table 1). Previous studies showed that there was a positive
ature. The spatial pattern of mean minimum temperature indi- trend in the annual (+5.5 mm per year) and pre-monsoonal (+
cates a 0.02 to 0.06 °C upward trend per year in the northern, 2.47 mm per year) rains and changes to rainfall during the
northwestern, northeastern, central and central southern parts post-monsoon season were not significant (Shahid 2010). A
of the country (Fig. 4b), while an extremely high trend for the study by OECD (2003) pointed out a positive trend of precip-
itation annually as well as during the pre-monsoon and post-
monsoon seasons, but there was no appreciable change during
Table 1 Trends of mean, mean minimum, mean maximum the winter in Bangladesh. Meanwhile, Ahmed and Alam
temperatures and annual, pre- and post-monsoon rainfalls of Bangladesh (1999) mentioned that there was little change in winter rainfall
and an upward trend in rainfall during the other seasons in
Temperature (°C/year) 1971–2010 (40 years) 1981–2010 (30 years)
Bangladesh. According to McSweeney et al. (2010), the mean
Mean 0.020 0.024 rainfall over Bangladesh has decreased by 13.2 mm per de-
Mean minimum 0.018 0.021 cade (6 %) between 1960 and 2003, but this was not statisti-
Mean maximum 0.022 0.028 cally significant. However, a positive trend (+3.4 %) was ob-
Rainfall (mm/year) 1971–2010 (40 years) 1981–2010 (30 years) served in the months MAM and a negative trend (−1.7 %) in
Annual average 7.130 −5.616 the months June–August (JJA) between 1960 and 2003.
Pre-monsoon −0.750 −4.769 Conversely, the Met Office Hadley Centre pointed out that
Post-monsoon −0.550 −1.296 there has been a negligible positive trend in total precipitation
over Bangladesh since 1960 (Met Office 2011). However,
Climate change in Bangladesh: a spatio-temporal analysis

Fig. 4 Trends of temperature (°C

per year): a mean, b mean
minimum and c mean maximum

significant changes in rainfall have been observed in recent per year) in the northwestern, southern and southwestern parts
decades as well as the overall variability increasing and ex- (Fig. 5c). On the other hand, most of the weather stations also
treme weather events have been observed in Bangladesh re- showed negative trends for the pre-monsoon rainfall, except
cently. Previous studies mostly analysed data for the long- for the stations located at the extreme northwestern, northeast-
term and started from 1958 and did not include very recent ern and southeastern parts of the country (Fig. 5b). The max-
data, which may explain slight differences with the current imum downward trend of pre-monsoon rain (−2.10 to
study. However, the present study corroborates with the −10.88 mm per year) was observed in the central, central
trend in rainfall of South Asia obtained by Solomon et al. western and central southern parts (Fig. 5b). This implies that
(2007) in which it was found that seasonal precipitation sig- a dry weather or drought condition occurred during the pre-
nificantly decreased in southern Asia. and post-monsoon seasons over a wide area of Bangladesh,
The spatial trend of rainfall shows a regional disparity as particularly in the northwestern and southwestern areas.
well as variations of rainfall in Bangladesh (Fig. 5). When Though the findings of the present study agree with those of
comparing the districts, an upward trend of annual rainfall Shahid (2010), in which there was a positive trend in annual
was highest (9 to 43 mm per year) in the hill districts, located rainfall, it does not agree with the spatial pattern of seasonal
in the southeastern part of Bangladesh. However, a negative rainfall trends in Bangladesh. Shahid (2010) found a signifi-
trend of annual rainfall was found at the Rajshahi, Ishurdi, cant upward trend of annual rainfall in the western part; how-
Faridpur, Madaripur and Patuakhali stations (Fig. 5a). The ever, in the present study, a significant upward trend was
post-monsoon rainfall presented a negative trend at all of the mostly observed in the southeastern and northern parts of
weather stations, and the rate was highest (−1.70 to −3.4 mm Bangladesh. In the last decade, low rainfall was recorded in

Fig. 5 Trends of rainfall (mm per

year): a annual, b pre–monsoon
and c post-monsoon
 Rahman M.R., Lateh H.

the northwestern, western, central and central southern parts Spatial pattern depicted that the annual rainfall variability was
and high rainfall was recorded in the southeastern and north- the highest (between 22.25 and 32.80 % per year) in the north-
eastern parts (BMD 2013). The present study agrees with the western, northern and eastern parts (Fig. 7a). On the other
findings of the pre-monsoon spatial trend of rainfall obtained hand, the maximum pre-monsoon (between 46 to 60 % per
by Shahid (2010) in which there was an increase to the pre- year) and post-monsoon (between 83 to 100 % per year) rain-
monsoon rainfall in the extreme northwestern and southeast- fall variability was highest in the southern coastal and south-
ern parts of Bangladesh. eastern hill districts of Bangladesh. The pre-monsoon rainfall
variability was also remarkable in the northwestern and south-
western parts (Fig. 7b). The maximum rainfall variability, par-
4.2 Variability of temperature and rainfall ticularly in pre- and post-monsoon seasons, may be attributed
to the relative dryness observed in those parts of the country
The variability of the temperature shows that the mean, mean during the period, since generally, low rainfall areas experi-
minimum and mean maximum variabilities were 0.081, 0.025 ence greater variability (Rahman and Lateh 2015). Of course,
and 0.017 °C per year, respectively. With the exception of the the southern coastal and southeastern hill districts have early
mean temperature variability, the variability of the mean min- post-monsoon depressions or cyclone rain effects, which have
imum and mean maximum was not remarkable in Bangladesh been increasing in recent decades that could also explain the
during the investigated period of 1971–2010. However, the high rainfall variability in the area. It may be noted here that in
spatial pattern of the mean minimum temperature variability Bangladesh, tropical cyclones occur during the pre- and post-
shows that a higher variability (>0.027 to 0.051 °C per year) monsoon seasons. A study on Bangladesh rainfall has shown
was observed mainly in the northwestern, northern and north- that, in general, there was decrease in rainfall during El-Niño
eastern parts (Fig. 6b). The spatial variability in the mean years in all the seasons (the pre-monsoon, the monsoon and
maximum temperature was higher (between 0.017 and the post-monsoon) (Ahmed et al. 1996), which may be linked
0.026 °C per year) at the northeastern, eastern and southeast- to high rainfall variability. Further, in a study by Rajeevan
ern parts of the country (Fig. 6c) compared to other parts of the et al. (2008), it was shown that the inter-decadal and seasonal
country. In the case of the spatial pattern of the mean temper- variability of rainfall might be linked to the variations and
ature variability, maximum variability (0.11 to 0.25 °C per anomalies of sea surface temperatures (SST) over the equato-
year) was found in the northeastern, southern and southeastern rial Indian Ocean, particularly the East Indian Ocean, which is
parts (Fig. 6a). Indeed, some parts of the investigated area associated with global warming.
showed very low variability of the mean minimum and mean
maximum temperatures (<0.02 °C per year) [Fig. 6b, c].
Further, variability analysis showed that rainfall was erratic 4.3 Forecasting of temperature and rainfall (2011–2020)
during the pre-monsoon season (44.84 % per year) and be-
came much more erratic during the post-monsoon season It was mentioned earlier in this study that we forecasted (short-
(85.25 % per year) during the last 40 years in Bangladesh. term) temperature and rainfall using ARIMA time series anal-
The annual rainfall variability was found to be 22.59 % per ysis model. To do so, ARIMA Expert Modeller of SPSS sta-
year (range of 15.20–32.80 %) during the investigated period. tistical analysis software was used to perform the automatic

Fig. 6 Variability of temperature

(°C per year): a mean, b mean
minimum and c mean maximum
Climate change in Bangladesh: a spatio-temporal analysis

Fig. 7 Variability of rainfall (%

per year): a annual, b pre-
monsoon and c post-monsoon

best fitted simulation of temperature and rainfall data for each indicting validation of forecasting by the model. For other
station. Therefore, in a continuous sequence from January to stations, we also found similar statistical comparisons. It
December (30 years), observed temperature and rainfall data may be noted here that in Fig. 8d, three horizontal clusters
were used separately for each station as a single time series were noticed mainly because of seasonal variations in the data
variable (360). We validated the forecasted values for the de- set. Thus, it is apparent that the ARIMA Expert Modeller,
cade 2001–2010 with observed values of the same decade first which automatically fitted the best ARIMA model for simu-
and then, finally forecasted the temperature and rainfall for the lation for the specific time series data set, can be applied with
decade of 2011–2020. When comparing the observed values adequate accuracy to forecast temperature and rainfall in the
and the forecasted values for the 2001–2010 period and visu- area.
ally inspecting the plot, it was quite evident that the forecasted Here, temperature (mean, mean minimum and mean
values were very close to the observed values (Figs. 8 and 9). maximum) and rainfall (annual, pre-monsoon and post-
In Fig. 8, the observed mean maximum and mean minimum monsoon) of Bangladesh during the second decade (2011–
temperatures from the Rajshahi station were compared (as a 2020) of the twenty-first century was forecasted and yearly
sample/evident) with ARIMA model forecasted values statistics were calculated from the monthly data set and then
(Fig. 8a, c). Between these two data sets, of observed and decadal statistics were calculated from the yearly statistics.
forecasted, the coefficient of determination (R2) was found Afterwards, a spatial interpolation technique was followed
to be +0.956 and +0.970 (highly positive) for the mean max- using IDW and GIS to generate the spatial pattern of the
imum and mean minimum temperatures, respectively, at the forecasted temperature and rainfall. The spatial variation
Rajshahi station (Fig. 8b, d). Again, when comparing the ob- of the mean, mean minimum and mean maximum tempera-
served and forecasted values of the annual average rainfall tures in 2011–2020 are predicted to be 25.11–27.30, 20.04–
during the 2001–2010 period, this also showed a highly pos- 23.06 and 29.78–32.08 °C, respectively (Fig. 10). Within
itive (+0.904) coefficient of determination (R2) (Fig. 9b), this decade (2011–2020), the temperature is predicted to

Fig. 8 Comparison between the

observed and forecasted
temperature of Rajshahi station,
January 2001–December 2010. a
Time plot and b scatter plot of the
observed monthly mean
maximum temperature and
ARIMA forecasted value (°C); c
time plot and d scatter plot of the
observed monthly mean
minimum temperature and
ARIMA forecasted value (°C)
 Rahman M.R., Lateh H.

Fig. 9 Comparison between the observed and forecasted rainfall of Rajshahi station, January 2001–December 2010. a Time plot and b scatter plot of the
observed monthly average rainfall and ARIMA forecasted value (mm)

rise by about 0.18, 0.20 and 0.16 °C for the mean, mean will receive low rainfall, which is predicted to continually
minimum and mean maximum temperatures, respectively, decrease during the 2011–2020 period (Fig. 11b). On the
compared to previous decade of 2001–2010. The forecasted other hand, during the post-monsoon season, the north-
warming of 0.18 °C per decade (mean temperature) during western, western and northern districts of the country will
2011–2020 is consistent with the fourth and fifth IPCC’s receive very low rainfall and this will likely continue to
predictions of 0.2 and 0.17 °C per decade, respectively decrease (Fig. 11c). By 2020, the post-monsoon rainfall
(IPCC 2007; IPCC 2014). It is also comparable in magni- will decrease by about 13 %, compared to 1971 (Table 2).
tude to other forecasts (Lean and Rind 2009). Comparative However, the data predicts that during the post-monsoon
statistics of climate change predictions in Bangladesh are season, the southwestern and central parts of the country
given in Table 2. Our study indicates faster warming than will receive slightly more rainfall than other parts
others and more of a decrease to post-monsoon rainfall. The (Fig. 11c). This is mainly because of high variability of
results obtained in this study may be an indication of the post-monsoonal rains in these parts of the country.
global change to the seasonal rainfall response. In summary, the trends of temperature and rainfall during
Prediction of rainfall shows a decrease of about the 1971–2010 period enhanced the attention of the region
−153 mm (−15.3 mm per year) rainfall in the decade of by climate change, particularly with regard to the trend of
2011–2020, compared to the previous decade 2001–2010. increased temperature. The results indicate that climate
Moreover, during the period of 2011–2020, a remarkable change apparently shows much stronger warming in
low rainfall is projected for the Rajshahi, Bogra, Jessore Bangladesh than what was mentioned by the global view
and Kustia districts, compared to the others parts of the published in the IPCC report (IPCC 2007; IPCC 2014). In
country (Fig. 11a). However, by 2020, the annual rainfall Bangladesh, starting from 1971, the mean temperature in-
is projected to increase by about 5.5 % (Table 2), com- creased by about 0.20 °C per decade in the last 40 years.
pared to the rainfall in 1971 which is consistent with the This is in contrast with the global mean temperature, which
IPCC’s prediction of a 5–6 % increase of rainfall by 2030 increased by 0.074 °C ± 0.02 °C per decade in the last
(IPCC 2007). During the pre-monsoon season, the north- 100 years (1906–2005) and 0.13 °C ± 0.03 °C per decade
western, southwestern and western parts of the country in the last 50 years (from 1956 to 2005) (IPCC 2007). Based

Fig. 10 Spatial pattern of

forecasted temperature (°C):
2011–2020. a Mean, b mean
minimum and c mean maximum
Climate change in Bangladesh: a spatio-temporal analysis

Table 2 Projected change

scenarios in temperature and Year Mean temperature (°C) Annual Post-monsoon Reference
rainfall of Bangladesh rainfall (%) rainfall (%)

2020 +0.98 (base 1971–2010) +5.5 −13 Present study

2020 +0.9 to +1.0 (base 1960–2000) +5 to7 +3 Tanner et al. 2007
2020 +0.09 (base 1951–1990) +5.3 +1.3 Islam 2009
2030 +1.0 (base 1960–1995) +3.8 −1.2 Agrawala et al. 2003
2030 +1.0 (base 1951–1990) +5 −2 GOB 2005
2030 +0.8 (base 1960–2003) +1 −2 McSweeney et al. 2010
2050 +0.5 to +2.1 (base 1961–1990) − −12 Rahman et al. 2012a
2075 +2.5 (base 1979–2006) +1.4 +5.7 Rahman and Ferdousi 2012
2100 +3.0 to +3.5 (base 1960–1990) +5 to +10 − Met Office 2011

on the Global Historical Climatology Network (GHCN) da- temperatures increased between 0.80 and 2.4 °C over the
tabase, global linear trends of 0.10 °C per decade in the 1971–2010 period. Conversely, this study exhibited a
twentieth century have accelerated since 1950 to 0.16 °C 14 % (+7.13 mm per year) upward trend in the annual rain-
p er de c ad e ( I P C C 2 0 07 ) . M o r e o v e r, w a r m i n g i n fall and 17 % (−0.75 mm per year) and 39 % (−0.55 mm per
Bangladesh was even faster than the 1979–2012 period with year) downward trend in pre-monsoon and post-monsoon
global warming rates of 0.16 °C per decade (Morice et al. rainfall, respectively, during 1971–2010. The rainfall was
2012). Furthermore, the fifth assessment of the IPCC stated also erratic during the pre-monsoon season, which then in-
that the average temperature of the Earth has risen by creased during the post-monsoon season. Furthermore, dry
0.12 °C per decade during the period 1951–2012 (IPCC weather or drought conditions were noticed during the pre-
2014). Again, the investigated temperature trends in this and post-monsoon seasons over a wide area of Bangladesh,
study appear to be higher than most other studies in particularly in the northwestern and southwestern parts of
Bangladesh so far (Section 4.1). This supports the observa- the country during the post-monsoon season. In these areas,
tion that an increase of temperature has taken place more in rainfall decreased by about 50–70 % during the 1971–2010
recent decades than previous ones in Bangladesh. In addi- period. The areas under a faster warming trend coincided
tion, the highest positive trend in minimum temperature was with the areas having this drying trend. The ARIMA time
observed in the northern, northwestern, central and central series model predicted that the mean temperature exhibited
southern parts, while the highest positive trend was noticed a 0.18 °C elevation during the 2011–2020 period compared
in the southern, southeastern and northeastern parts in the with the previous decade, indicating an approximately
case of the maximum temperature. In these areas, 0.98 °C warmer temperature in Bangladesh by 2020

Fig. 11 Spatial pattern of

forecasted rainfall (mm per year):
2011–2020. a Annual average, b
pre-monsoon and c post-monsoon
 Rahman M.R., Lateh H.

compared to 1971. On the other hand, there was a prediction temperature warmed more in the northern, northwestern,
for a 153-mm decline of annual rainfall over the 2011–2020 northeastern, central and central southern parts while the
period. Moreover, it is predicted that the pre- and post- maximum temperature warmed more in the southern,
monsoonal rainfall would decrease at a small rate (3 and southeastern and northeastern parts during the 1971–
5 mm, respectively) which might lead to a drying condition 2010 period. In some of these parts, the mean minimum
that would persist in the northwestern, western and south- and mean maximum temperatures rose by more than
western parts of the country during 2011–2020. The general 2.0 °C (0.50 °C per decade), which will likely pose chal-
tendency derived by the IPCC from analysing 20 global lenges to the population in those parts of the country. On
climate models indicates that winter precipitation in the re- the other hand, the most significant results were related to
gion will increase and summer precipitation will decrease the dry conditions, since during the investigated period, a
(Reiter et al. 2012), however, the analysis of rainfall data remarkable decrease to the pre- and post-monsoonal rain-
from our study depicts decreases of both post-monsoon falls was noticed (−0.75 and −0.55 mm per year, respec-
(winter) and pre-monsoon rainfall which do not match with tively) with very high variability (44.84 and 85.25 % per
the IPCC’s general seasonal tendency. This is mainly due to year, respectively). High rainfall variability is an indicator
the high spatial and temporal variability of rainfall from of drought, and therefore, the areas under very high vari-
region to region. Even in our study, because of high spatial ability with low rainfall, particularly the north western
and temporal variability, we observed a slightly positive districts, are prone to drought hazard. Predictions of rain-
trend of pre-monsoon rainfall at the extreme northwestern, fall reveal that declining rainfall will continue and a drying
northeastern and southeastern parts of the country (Fig. 5b). condition will persist during 2011–2020 (153 mm decrease
of annual rainfall), especially during the pre- and post-
monsoon seasons. Spatial patterns of trend and variability
5 Conclusions of temperature and rainfall indicate that the northwestern,
western and southwestern parts of the country are more
In this paper, we discussed the last 40 years (1971–2010) susceptible to climate change with respect to rising tem-
of temporal-spatial climate change in Bangladesh based on perature, high variability and rain shortfalls, particularly
temperature and rainfall data. The ARIMA time series for the pre- and post-monsoon rains. It is expected that
model-based future prediction of climate change for the this study will not only help to delineate appropriate pol-
2011–2020 period was evaluated and confirmed a particu- icies and planning to combat the impact of climate change
larly strong recent climate change in Bangladesh based on in Bangladesh but also help to understand the regional
temperature and rainfall changes. This study also validates climate change in this part of the South Asia. The trend
the ARIMA time series model for shorter time scale cli- directions, magnitude and spatial patterns identified for
mate simulations and can easily be applied to more local- both temperature and rainfall may also provide helpful
ized climate data since the ARIMA modelling focuses information on global warming on a regional/country-
purely on the data rather than data generating processes. level scale. Improved understanding of recent climate
However, since this model is solely based on a statistical change helps to elucidate the impacts and vulnerability
approach, it is limited in terms of extreme and irregular of the local population in order to implement the most
events since this model cannot predict extreme events due appropriate practices to cope with climate change and
to any external force or natural events. For example, if a manage the changing situation in a better way.
strong global external force of the atmosphere, like a ma-
jor volcanic eruption occurs, this might be a sound reason Acknowledgments First author would like to acknowledge the finan-
to invalidate the forecast for that particular time. cial support from the USM-TWAS post-doctoral fellowship programme
for the research work. Authors are also thankful to the Bangladesh Me-
Therefore, this should be considered as a limitation of teorological Department (BMD) for providing historical data, and the
the model since the type of situation is not a regular event. anonymous reviewers for their valuable comments and suggestions.
For the climate of Bangladesh, the changes of temperature
reflect a warming as a whole and since 1971, the climate
of Bangladesh is warming at a much higher rate of global References
average warming (0.20 versus 0.13 °C per decade).
Prediction of the temperature shows that in the decade of Afshin S, Fahmi H, Alizadeh A, Sedghi H, Kaveh F (2011) Long term
2011–2020, the mean temperature will increase by about rainfall forecasting by integrated artificial neural network-fuzzy log-
0.18 °C, indicating the temperature will be 1.0 °C warmer ic-wavelet model in Karoon basin. Sci Res Essays 6(6):1200–1208
Agrawala S, Ota T, Ahmed AU, Smith J, Van M (2003) Development and
in Bangladesh by 2020 compared to 1971. An even greater climate change in Bangladesh: focus on coastal flooding and the
rise is projected for the mean minimum (0.20 °C) than the Sundarbans. Organisation for Economic Co-operation and
mean maximum (0.16 °C) temperature. The minimum Development (OECD). Paris
Climate change in Bangladesh: a spatio-temporal analysis

Ahmad QK, Warrick RA (1996) The implications of climate and sea level the Aegean during 1961–2002. Int J Climatol 28:1735–1749. doi:
change for Bangladesh. Kluwer Academic Publisher, Dordrecht 10.1002/joc.1660
Ahmed ASMS, Munim AA, Begum QN, Choudhury AM (1996) El Niño Hansen J, Sato M, Ruedy R, Lo K, Lea DW, Medina-Elizade M
southern oscillation and rainfall variation over Bangladesh. Mausam (2006) Global temperature change. Proc Natl Acad Sci USA
47(2):157–162 103:14288–14293
Ahmed AU, Alam M (1999) Development of climate change scenarios Hartmann DL, Klein Tank AMG, Rusticucci M, Alexander LV,
with general circulation models. In: Huq S, Karim Z, Asaduzzaman Brönnimann S, Charabi Y, Dentener FJ, Dlugokencky EJ,
M, Mahtab F (eds) Vulnerability and adaptation to climate change for Easterling DR, Kaplan A, Soden BJ, Thorne PW, Wild M, Zhai PM
Bangladesh. Kluwer Academic Publishers, Dordrecht, pp. 125–143 (2013) Observations: atmosphere and surface. In: Stocker TF, Qin D,
Baker-Austin C, Trinanes JA, Taylor NGH, Hartnell R, Siitonen A, Plattner GK, Tignor M, Allen SK, Boschung J, Nauels A, Xia Y, Bex
Martinez-Urtaza J (2012) Emerging Vibrio risk at high latitudes in V, Midgley PM (eds) Climate change 2013: the physical science
response to ocean warming. Nat Clim Chang. doi:10.1038/ basis. Contribution of working group I to the fifth assessment report
nclimate1628 of the intergovernmental panel on climate change. Cambridge
Begert M, Schlegel T, Kirchhofer W (2005) Homogeneous temperature University Press, Cambridge
and precipitation series of Switzerland from 1864 to 2000. Int J Hasan Z, Akhter S, Islam M (2014) Climate change and trend of rainfall
Climatol 25:65–80. doi:10.1002/joc.1118 in the south-east part of coastal Bangladesh. European Sci J 10(2)
Bloomfield P (1992) Trends in global temperature. Clim Chang 21:1–16 Hulme M, Osborn TJ, Johns TC (1998) Precipitation sensitivity to global
BMD (2013) Bangladesh Meteorological Department. Agargoan, Dhaka warming: comparison of observations with HADCM2 simulations.
Box GEP, Jenkins GM (1976) Time series analysis: forecasting and con- Geophys Res Lett 25:3379–3382
trol. Holden-Day, Boca Raton ILWIS (2005). Integrated land and water information system. version 3.3,
Boyle SJ, Tsanis IK, Kanaroglou PS (1998) Developing geographic in- ITC, Netherlands
formation systems for land use impact assessment in flooding con- Immerzeel W (2007) Historical trends and future predictions of climate
dition. J Wat Resour Plan Manage ASCE 124(2):89–98 variability in the Brahmaputra basin. Int J of Clim 28(2):243–254
Brázdil R, Chromá K, Dobrovolný P, Tolasz R (2009) Climate fluctua- IPCC (2001) Climate change: synthesis report. An assessment of inter
tions in the Czech Republic during the period 1961–2005. Int J governmental panel on climate change. Cambridge University
Climatol 29:223–242. doi:10.1002/joc.1718 Press, Cambridge
Brockwell PJ, Davis RA (1991) Time series: theory and methods. 2ed., IPCC (2007) Technical summary of Climate Change 2007: the physical
Springer science basis. Contribution of Working Group I to the fourth assess-
Dash SK, Mamgain A, Pattnayak KC, Giorgi F (2013) Spatial and tem- ment report of the Intergovernmental Panel on Climate Change, S.
poral variations in Indian summer monsoon rainfall and tempera- Solomon et al., (Eds.), Cambridge University Press, Cambridge
ture: an analysis based on RegCM3 simulations. Pure Appl Geophys
IPCC (2014) Climate Change 2014: IPCC fifth assessment synthesis
170:655–674. doi:10.1007/s00024-012-0567-4
report-summary for policymakers. An assessment of inter govern-
Dash SK, Shekhar MS, Singh GP (2006) Simulation of Indian summer
mental panel on climate change. Cambridge University Press,
monsoon circulation and rainfall using RegCM3. Theor App Clim
Cambridge
86(1–4):161–172
Islam MN (2009) Rainfall and temperature scenario for Bangladesh.
Diomede T, Davolio S, Marsigli C, Miglietta MM, Moscatello A, Papetti
Open Atmos Sci J 3:93–103
P, Paccagnella T, Buzzi A, Malguzzi P (2008) Discharge prediction
Jones PD (1995) Maximum and minimum temperature trends in Ireland,
based on multi model precipitation forecasts. Meteorog Atmos Phys
Italy. Thailand Atmos Res 37(1–3):67–78
101(3–4):245–265
Divya MR (1995) Climate change and hydrology with emphasis on the Jones PD (2001) Instrumental temperature change in the context of the
Indian subcontinent. Hydrol Sci J 40:231–241 last 1000 years. In: Brunet M, Lo’pez D (eds) Detecting and model-
Domonkos P, Tar K (2003) Long-term changes in observed temperature ling regional climate change. Springer, New York, pp. 55–68
and precipitation series 1901–1998 from Hungary and their relations Jones PD, Moberg A (2003) A hemispheric and large-scale surface air
to larger scale changes. Theor Appl Climatol 75:131–147. doi:10. temperature variations: an extensive revision and an update to 2001.
1007/s00704-002-0716-2 J Clim 16:206–223
Esfahani AA, Friedel MJ (2014) Forecasting conditional climate-change Karmakar S, Shrestha ML (2000) Recent climate change in Bangladesh.
using a hybrid approach. Environ Model Softw 52:83–97 SMRC No.4, SMRC, Dhaka
Feidas H, Noulopoulou CH, Makrogiannis T, Bora-Senta E (2007) Trend Karmalkar AV, Bradley RS, Diaz HF (2011) Climate change in Central
analysis of precipitation time series in Greece and their relationship America and Mexico: regional climate model validation and climate
with circulation using surface and satellite data: 1955–2001. Theor change projections. Clim Dyn 37:605–629. doi:10.1007/s00382-
Appl Climatol 87:155–177. doi:10.1007/s00704-006-0200-5 011-1099-9
Founda D, Papadopoulos KH, Petrakis M, Giannakopoulos C, Good P Kerr RA (2009) Galloping glaciers of Greenland have reined themselves
(2004) Analysis of mean, maximum, and minimum temperature in in. Science 323:458
Athens from 1897 to 2001 with emphasis on the last decade: trends, Kohler MA (1949) Double-mass analysis for testing the consis-
warm events, and cold events. Glob Planet Chang 44:27–38. doi:10. tency of records and for making adjustments. Bull Amer
1016/j.gloplacha.2004.06.003 Meteor Soc 30:188–189
Franke J, Goldberg V, Eichelmann U, Freydank E, Bernhofer C (2004) Kripalani RH, Oh JH, Kulkarni A, Sabade SS, Chaudhari HS (2007)
Statistical analysis of regional climate trends in Saxony, Germany. South Asian summer monsoon precipitation variability: coupled cli-
Clim Res 27:145–150. doi:10.3354/cr027145 mate model simulations and projections under IPCC AR4. Theor
Giorgi F (2005) Climate Change Prediction. Clim Chang 73:239–265. App Clim 90(3–4):133–159
doi:10.1007/s10584-005-6857-4 Krivtsov V (2004) Investigations of indirect relationships in ecology and
GOB (2005) National Adaptation Programme of Action (NAPA), final environmental sciences: a review and the implications for compara-
report. Ministry of Environment and Forest. Government of the tive theoretical ecosystem analysis. Ecol Model 174:37–54
People’s Republic of Bangladesh, Dhaka Lean JL, Rind DH (2009) How will earth’s surface temperature change in
Good P, Giannakopoulos C, Flocas H, Tolika K, Anagnostopoulou C, future decades. Geophys Res Lett 36:L15708. doi:10.1029/
Maherasd P (2008) Significant changes in the regional climate of 2009GL038932
 Rahman M.R., Lateh H.

Lee JH, Sohn KT (2007) Prediction of monthly mean surface air temper- Rahman MR (2013) Agro-spatial diversity in Bangladesh—a special ref-
ature in a region of China. Adv Atmos Sci 24:503–508. doi:10. erence to climate change and crop diversification in Rajshahi divi-
1007/s00376-007-0503-1 sion. J Geo-Environment 10
Li J, Huang JF, Wang XZ (2006) A GIS based approach for estimating Rahman MR, Lateh H (2015) Spatio-temporal analysis of warming in
spatial distribution of seasonal temperature in Zhejiang Province, Bangladesh using recent observed temperature data and GIS. Clim
China. J Univ Sci A 7(4):647–656 Dyn. doi:10.1007/s00382-015-2742-7
Lin GF, Chen GR, Wu MC, Chou YC (2009) Effective forecasting of Rahman MR, Saha SK (2007) Flood hazard zonation-a GIS aided multi-
hourly typhoon rainfall using support vector machines. Water criteria evaluation approach (MCE) with remotely sensed data. Int J
Resour Res 45:W08440 Geoinformat 3(3):25–37
Lorentzen T (2014) Statistical analysis of temperature data sampled at Rahman MR, Saha SK (2008) Remote sensing, spatial multi criteria eval-
Station-M in the Norwegian Sea. J Mar Syst 130:31–45 uation (SMCE) and analytical hierarchy process (AHP) in optimal
Mair A, El-Kadi AA, Ha K, Koh GW (2013) Temporal and spatial var- cropping pattern planning for a flood prone area. J Spat Sci 53(2):
iability of rainfall and climate trend on Jeju Island. Geosci J 17(1): 161–177
75–85. doi:10.1007/s12303-013-0002-6 Rahman MR, Saha SK (2009) Spatial dynamics of cropland and cropping
Marengo JA, Jones R, Alvesa LM, Valverde MC (2009) Future change of pattern change analysis using Landsat TM and IRS P6 LISS III
temperature and precipitation extremes in South America as derived satellite images with GIS. Geo-Spatial Informat Sci 12(2)
from the PRECIS regional climate modeling system. Int J Climatol Rahman MR, Shi ZH, Chongfa C (2014) Assessing regional environmen-
29(15):2241–2255. doi:10.1002/joc.1863 tal quality by integrated use of remote sensing, GIS, and spatial
Matulla C, Penlap EK, Haas P, Formayer H (2003) Comparative analysis multi-criteria evaluation for prioritization of environmental restora-
of spatial and seasonal variability: Austrian precipitation during the tion. Environ Monit Assess 186(11):6993–7009. doi:10.1007/
20th century. Int J Climatol 23:1577–1588. doi:10.1002/joc.960 s10661-014-3905-4
May W (2004) Simulation of the variability and extremes of daily rainfall Rahman MR, Shi ZH, Chongfa C, Dun Z (2015) Assessing soil erosion
during the Indian summer monsoon for present and future times in a hazard—a raster based GIS approach with spatial principal compo-
global time-slice experiment. Clim Dyn 22(2–3):183–204 nent analysis (SPCA). Earth Science Informatics DOI:10.1007/
McCarthy JJ, Canziani OF, Leary NA, Dokken DJ, White KS (eds) s12145-015-0219-1
(2001) Climate Change 2001: impacts, adaptation and vulnerability. Rahman MR, Shi ZH, Chongfa C (2009) Soil erosion hazard evaluation-
Cambridge University Press, Cambridge an integrated use of remote sensing, GIS and statistical approaches
with biophysical parameters towards management strategies. Ecol
McSweeney C, New M, Lizcano G (2010) UNDP climate change country
Model 220(13–14):1724–1734
profiles: Bangladesh, Available: http://country-profiles.geog.ox.ac.uk/.
Rahmstorf SA, Cazenave JA, Church JE, Hansen RF, Keeling DE, Parker
Accessed 10 May 2013
RC, Somerville J (2007) Recent climate observations compared to
Met Office (2011) Climate: observations, projections and impacts:
projections. Science 316:709. doi:10.1126/science.1136843
Bangladesh, Met Office, Hadley Centre, FitzRoy Road, Exeter
Rajeevan M, Bhate J, Jaswal AK (2008) Analysis of variability and trends
Devon, EX1 3PB United Kingdom
of extreme rainfall events over India using 104 years of gridded
Mia NM (2003) Variations of temperature of Bangladesh. Proc. SAARC
daily rainfall data. Geophys Res Lett 35:L18707. doi:10.1029/
seminars on climate variability in the South Asian region and its
2008GL035143
impacts, SMRC, Dhaka
Rashid HE (1991) Geography of Bangladesh. University Press Ltd,
Moore D, McCabe G (2003) Introduction to the practice of statistics. WH Dhaka
Freeman and Co., London Reiter A, Weidinger R, Mauser W (2012) Recent climate change at the
Morice CP, Kennedy JJ, Rayner NA, Jones PD (2012) Quantifying un- upper Danube—a temporal and spatial analysis of temperature and
certainties in global and regional temperature change using an en- precipitation time series. Clim Chang 111(3–4):665–696. doi:10.
semble of observational estimates: the HadCRUT4 data set. J 1007/s10584-011-0173-y
Geophys Res 117, D08101. doi:10.1029/2011JD017187 Rodrı′guez-Puebla C, Encinas AH, Nieto S, Garmendia J (1998) Spatial
Nick FM, Vieli A, Howat IM, Joughin I (2009) Large scale changes in and temporal patterns of annual precipitation variability over the
Greenland outlet glacier dynamics triggered at the terminus. Nat Iberian Peninsula. Int J Climatol 18:299–316
Geosci 2:110–114 Romilly P (2005) Time series modeling of global mean temperature for
OECD (2003) Development and climate change in Bangladesh: focus on managerial decision-making. J Environ Manag 76:61–70
coastal flooding and the Sundarbans. In: Agrawala S, Ota T, Ahmed Schaefer D, Domroes M (2009) Recent climate change in Japan—spatial
AU (ed) Organization for Economic Co-operation and Development and temporal characteristics of trends of temperature. Clim Past
(OECD) Report, Paris, France. www.oecd.org/dataoecd/46/55/ (Climate of the Past) 5:13–19
21055658.pdf. Accessed on 25 March 2009 Shahid S (2009) Spatio-temporal variability of rainfall over
Panofsky HA, Brier GW (1968) Some applications of statistics to mete- Bangladesh during the time period 1969–2003. J Atmos Sci
orology. Pennsylvania State University, Pennsylvania 45(3):375–389
Paudyal GN (1996) An integrated GIS-numerical modelling system for Shahid S (2010) Recent trends in the climate of Bangladesh. Clim Res
advanced flood management. In Proc. Int. Conf. on Water Resources 42(3):185–193
& Environment Research: Towards the 21st Century, Water Shahid S (2011) Trends in extreme rainfall events of Bangladesh. Theor
Resources Research Center, Kyoto Univ, Kyoto, Japan, 555–562 Appl Climatol 104:489–499. doi:10.1007/s00704-010-0363-y
Pérez CR, Jury MR (2013) Spatial and temporal analysis of climate Solomon S, Qin D, Manning M, Alley RB, Berntsen T, Bindoff NL, Chen Z,
change in Hispañola. Theor Appl Climatol 113:213–224. doi:10. Chidthaisong A, Gregory JM, Hegerl GC, Heimann M, Hewitson B,
1007/s00704-012-0781-0 Hoskins BJ, Joos F, Jouzel J, Kattsov V, Lohmann U, Matsuno T,
Rahman MM, Ferdousi N (2012) Rainfall and temperature scenario for Molina M, Nicholls N, Overpeck J, Raga G, Ramaswamy V, Ren J,
Bangladesh using 20 km mesh AGCM. Int J Climate Change Rusticucci M, Somerville R, Stocker TF, Whetton P, Wood RA,
Strateg Manag 4(1):66–80 Wratt D (2007) Technical summary. In: Solomon S, Qin D,
Rahman MM, Islam MN, Ahmed AU, Georgi F (2012a) Rainfall and Manning M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL
temperature scenarios for Bangladesh for the middle of 21st century (eds) Climate Change 2007: the physical science basis. Contribution
using RegCM. J Earth Syst Sci 121(2):287–295 of Working Group I to the fourth assessment report of the
Climate change in Bangladesh: a spatio-temporal analysis

Intergovernmental Panel on Climate Change. Cambridge University change, In Climate Change 2007: the physical science basis, eds.
Press, Cambridge, pp. 21–91 Solomon S, Qin D, Manning M, Chen Z., Marquis M., Averyt
SPSS (2007) SPSS software version 16, IBM Corporation, New Orchard K.B., Tignor M. and Miller H.L., Cambridge University Press,
Road, Armonk, New York, USA Cambridge, United Kingdom and New York, NY, USA
Tabari H, Hosseinzadeh TP (2011) Recent trends of mean, maximum and Yu B, Neil DT (1993) Long-term variations in regional rainfall in the
minimum air temperatures in the western half of Iran. Meteor Atmos south-west of Western Australia and the difference between average
Phys 111:121–131 and high intensity rainfalls. Int. J Climatol 13:77–88
Tanner TM, Hassan A, Islam KMN, Conway D, Mechler R, Ahmed AU, Yurekli K, Simsek H, Cemek B, Karaman S (2007) Simulating climatic
Alam M (2007) ORCHID: piloting climate risk screening in DFID variables by using stochastic approach. Build Environ 42:3493–3499
Bangladesh. Detailed Research Report. Institute of Development Zhang Q, Xu C-Y, Zhang Z, Ren G, Chen YD (2008) Climate change or
Studies, University of Sussex, UK variability? The case of yellow river as indicated by extreme maxi-
Trenberth KE, Jones PD, Ambenje P, Bojariu R, Easterling D, Klein mum and minimum air temperature during 1960–2004. Theor Appl
TA, Parker D, Rahimzadeh F, Renwick JA, Rusticucci M, Soden Climatol 93:35–43. doi:10.1007/s00704-007-0328-y
B, Zhai P (2007) Observations: surface and atmospheric climate