You are on page 1of 12

International Journal of Environmental Studies

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/genv20

Problems with analysing meteorological data


for assessing climate change: examples from
Bangladesh

Hugh Brammer

To cite this article: Hugh Brammer (2020) Problems with analysing meteorological data for
assessing climate change: examples from Bangladesh, International Journal of Environmental
Studies, 77:6, 905-915, DOI: 10.1080/00207233.2020.1848763

To link to this article: https://doi.org/10.1080/00207233.2020.1848763

Published online: 25 Nov 2020.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=genv20
INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES
2020, VOL. 77, NO. 6, 905–915
https://doi.org/10.1080/00207233.2020.1848763

ARTICLE

Problems with analysing meteorological data for assessing


climate change: examples from Bangladesh
Hugh Brammer O.B.E
Formerly FAO Agricultural Development Adviser, Bangladesh

ABSTRACT KEYWORDS
Climate is not changing uniformly within geographical regions, and Bangladesh; climate change;
long-term changes can be cyclical or irregular in some regions. The climate model; cyclone
latter means that the standard statistical methods for analysing frequency; flood frequency;
meteorological data
meteorological data that produce linear outputs can give mislead­
ing trends, with consequent misleading environmental and socio-
economic outputs of models using such trends as inputs.
Additionally, national and regional climate trends do not necessa­
rily match international climate model outputs, so the latter need to
be tested for relevance before being used in national or regional
environment-related studies.

Introduction
Some of the factors that can complicate the assessment of climate change are reviewed
below: irregular rates of change in annual temperatures between geographical regions
and within countries; local factors that can influence temperatures within countries;
problems with the statistical analysis of climate data in areas where changes over time are
non-linear; determination of the begin- and end-dates of the rainy season in countries
where seasonal rainfall determines crop-growing and animal-grazing seasons; and pro­
blems in relating climate model outputs to local changes in rainfall and temperatures.
Recommendations are made for addressing such problems.

Regional climate differences

Regional temperature changes


After a still-stand between ca 1940 and the late-1970s, global warming has continued
irregularly since then (Figure 1). That figure shows how global mean annual tempera­
tures have varied over the past 170 years. However, temperatures have not changed at
a uniform rate throughout the world. While it is widely recognised that temperatures are
increasing most rapidly in arctic regions, it is less widely recognised that temperatures are
changing at different rates within regions, and that they have changed little or not at all in
some parts of the world since global warming resumed over 50 years ago (Figure 2).1

CONTACT Hugh Brammer h.brammer@btinternet.com Formerly FAO Agricultural Development Adviser,


Bangladesh
© 2020 Informa UK Limited, trading as Taylor & Francis Group

Published online 25 Nov 2020


906 H. BRAMMER

0.8 0.8
0
Temperature difference from mean C 0.7 0.7
0.6 0.6
0.5 0.5
0-4 0-4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
-0.1 -0.1
-0.2 -0.2
-0.3 -0.3
-0.4 -0.4
-0.5 = 10-year running mean -0.5
-0.6 -0.6

2020
1850

1860

1900

1950

1960

1970

1990

2010
1870

1880

1910

1920

1940

2000
1890

1930

1980
Source: HadCRUT4 time series dataset (Morice et al.) [1].

Figure 1. Departures of global mean temperatures 1850–2019 from the 1861–1990 mean with 10-year
running mean. Source: HadCRUT4 time series dataset (Morice et al.) [1].

Figure 2. Global temperature trends 1979–2019.

Note that temperatures changed at different rates within the North American, South
American and African continents, but little over the Indian subcontinent.

Irregular temperature changes


Environmentalists need to be aware that temperature data derived from ground observa­
tions can give misleading information. That is because the majority of meteorological
INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 907

stations are in urban areas where temperatures can have increased due to an urban heat
island effect and they can have done so at different rates within countries because of
different rates of urban growth and different contributors to urban warming (such as
building type and density, industrial effluents, traffic type). Changes in land use can also
have affected temperatures in different parts of a country over time, as for example by the
expansion of irrigated rice cultivation in Bangladesh in recent decades which apparently
decreased maximum temperatures and increased minimum temperatures in the dry
season [2].
Changes in the sites of meteorological stations can also affect apparent trends. Figure 3
illustrates a situation where local factors apparently caused differences in mean annual
temperature trends between four stations in the central and north-east regions of
Bangladesh. Two of the stations (Dhaka, Sylhet) are in urban environments; the other
two are in rural environments. However, the sites of three of these stations were moved
during the study period: Dhaka from an urban park to the airport and then to a built-up
area with high-rise buildings; Mymensingh from the town centre to a university farm site
outside the town (but where the built-up area on the campus has expanded over time);
and Srimangal from a former airfield on a floodplain to a tea research station in the
adjoining forested hills. The sudden increase in temperature trends at Sylhet in the late-
1980s suggests that there was a change in the adjoining environment (which the author
has not been able to check). Note that temperatures are higher at Dhaka than at the other
stations, suggesting a longer urban warming effect in the country’s long-established
capital city than at the other places.
Environmentalists should examine temperature data to see if changes in trends, or
trends differing from those of neighbouring stations, suggest that local factors may have
contributed to the recorded temperature changes. In the case of site change, only data
obtained since the last change should be used in assessing the possible influence of
climate change. To the extent possible, the use of data from urban sites should be
avoided; preference should be given to data from meteorological stations in rural areas

OC OC
27 27
Mymensingh Sylhet

26 26

25 25

24 xx x xxxxxxxxx xxx x 24 xx x xxx x x x


1960

1980

1990

2000

2010

1960

1980

1990

2000

2010
1970

1970

OC OC
28 28
Dhaka Srimangal

27 27

26 26

25 25

24 x x xx 24 x x xx xxxxxxx xx xxx xx
10-year running mean Short or long-term mean x No or incomplete data

Figure 3. Mean annual temperatures at stations in the centre and north-east of Bangladesh
1959–2016 with 10-year running means.
908 H. BRAMMER

such as in national parks and on agricultural research stations where temperatures are
less likely to have been affected by urban growth or by major changes in vegetation or
land use during the reporting period.

Cyclical and irregular regimes


Figure 4 illustrates annual rainfall at Kolkata (Alipore station) over the 190-year period
1829–2018. It shows the considerable interannual rainfall variability and also, via the 10-year
running mean line, a ca 60-year cyclical pattern. Figure 5 illustrates that more tropical
cyclones and severe cyclonic storms affected Bangladesh in the four decades between 1960
and 2000 than occurred in the previous six decades and in the subsequent two decades,
a distribution that appears neither to be cyclical nor related to the rainfall pattern shown in
Figure 4. Environmentalists need to look at long-term data in areas where they are working to

Kolkata
mm mm
2500 2500

2000 2000

1500 1500

1000 1000

500 500

0 X X 0
1980
1940
1910

1930

1970

2000
1890
1830

1840

1850

1870

1950

1960

1990

2010

2020
1860

1880

1900

1920

Source. Based on data provided by the Climate Research Unit, University of East Anglia, U.K., and the Indian Meteorological Department, Kolkata
Legend
10-year running mean Mean annual rainfall X No data

Figure 4. Annual rainfall at Kolkata 1829–2018 with 10-year running mean.

12 12
10 10
8 8
6 6
4 4
2 2
0 0
1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s 2010s

Figure 5. Frequency of tropical cyclones and severe cyclonic storms affecting Bangladesh 1900–2018
by decade. Sources: 1. 1900-1959: Bangladesh Bureau of Statistics Yearbooks;
2. 1960-2013: Table 2 in Brammer 2016 [3];
3. 2014-2018: personal records
INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 909

see if cyclical or other patterns occur before assuming that any recent change is due to global
warming. Kolkata lies within the Indian subcontinent where Figure 2 shows that there was
little change in annual temperatures over the 60-year period, so the irregular rainfall pattern
illustrated in Figure 4 is unlikely to be related to global climate change.
Researchers particularly need to check factual information sources. Table 1 shows that
there was no increase in the intensity of cyclones affecting Bangladesh during the period
of global warming since the 1950s.2 Similarly, Figure 6 shows that, contrary to popular
assumptions made in newspaper articles in Bangladesh, damaging floods – say those
affecting more than 30% of the country’s total area – have not increased in frequency or
extent in recent decades; presentation of the data in percentage classes in Table 2
indicates that the distribution was erratic. Assessment of the flood hazard in

Table 1. Number of cyclones and severe cyclonic storms affecting Bangladesh 1960–2018 in Saffir-
Simpson hurricane classes.
Decade 0 87–118 kmph I 119–153 kmph II 154–177 kmph III 178–208 kmph IV 209–251 kmph V
1960–69 1 2 6 2 0 0
1970–79 3 1 2 0 1 0
1980–89 2 2 2 0 0 0
1990–99 1 4 1 0 2 0
2000–09 1 0 0 0 1 0
2010–18 3 0 0 0 0 0
Total 11 9 11 2 4 0
a
0 = Storms classed as Severe Cyclonic Storms with sustained wind speeds of 87–118 kmph, below the lowest limit of 119
kmph for hurricanes (tropical cyclones) in the Saffir-Simpson scale.
b
Wind speeds were not recorded for two storms, in late May 1965 and October 1988, but storm surge heights >5 m
suggest that the storms were of cyclone dimensions. These cyclones were not included in the table.
c
Source. Compiled from data in Table 2 in Brammer [3], but information obtained personally from BGD for two cyclonic
storms reported after the last date reported in that table.

Percent of Area
floodplain 000 km2
70
100
60

80
50

40 60

30
40
20

20
10

XX X X X 0
1991

2001

2011
1971

1981
1962

1972

1982

1992

2002

2012
1960

1964

1974

1984
1954
1955
1966
1956

1963

1966
1867

1970
1968
1969

1973

1976

1983

1986

1990

2000
1987
1988
1989

1997
1998
1999

2003
2004

2010
2008
2009

2013
1961

1965

1975

1977
1978
1979
1980

1993
1994
1095
1996

2005
2006
1985

2007

X = Not reported

Figure 6. Variations in areas flooded in Bangladesh 1954–2013.


Adapted from Figure 3.11 in Baseline Study 3: Water Resources, Bangladesh Delta Plan 2100 [4]
910 H. BRAMMER

Table 2. Proportions of Bangladesh’s floodplain area flooded 1954–2013 in percentage classes by


decade.
Percentage Decade and number of years with records
class 1950s 3 1960s 10 1970s 9 1980s 9 1990s 9 2000s 10 2010s 4
60–69 0 0 0 1 1 0 0
50–59 0 0 0 0 0 0 0
40–49 0 0 0 0 0 1 0
30–39 1 0 1 1 0 1 0
20–29 2 6 3 1 4 2 1
10-19 0 4 3 1 2 5 3
1–9 0 0 2 5 2 1 0

Bangladesh is a complex subject [5]. Flood risk and damage are not only determined by
the area covered; they are also influenced by time of occurrence in relation to crop sowing
and harvesting dates. The wide expansion of irrigated boro rice cultivation in the dry
season has greatly reduced the impact of floods on total national rice production over
time: aus plus aman produced ca 80% of total rice production at the time of Bangladesh’s
Independence in 1971, but only ca 44% in 2014–15 [6].
The standard period for analysing meteorological data to indicate long-term trends is
30 years. The cyclical and irregular rainfall patterns illustrated in Figure 4 indicate that a 30-
year period of analysis can be too short to provide reliable long-term trends. That such
a period can be too short is illustrated by the decrease of 25.6% in annual rainfall in the north-
west region of Bangladesh in the period 1981–2014 reported by Dey et al. [7]. The annual
rainfall data shown in Figure 7 for the five main meteorological stations in that region
indicate that linear statistical analyses starting ca 1961 would have given significantly different
projections from those given by Dey et al.: for example, that there was no significant long-
term decrease in annual rainfall within the region. Figure 7 suggests that there is a 15-year
cycle at Dinajpur and Rangpur in the north of the region which is barely visible at Bogra in
the centre and at Ishurdi and Rajshahi in the south. In relation to the assessment of climate
change, note that both rainfall in individual years as well as long-term trends shown by the
running mean lines illustrated in Figure 7 differed between stations in the north-west region:
it can therefore be misleading to generalise trends for a region or a whole country when there
can be different trends within them. These examples illustrate that, in regions where climate
patterns are cyclical or irregular, linear statistical projections can differ depending on the
starting date used, whether that be for a fixed-term period or determined by the start and end
dates of the data available, and they can lead to misleading interpretations.
The danger of using too short a period for determining trends is illustrated in a paper by
Timsina et al. [8] who used the Dey et al. rainfall trend data quoted above to model boro
(winter) rice production in the north-west region of Bangladesh in the next 10 and 20 years.
They concluded that boro rice production could fall by 50% in the next two decades because
of reduced recharge of groundwater used for irrigation in the region due to decreasing
rainfall. Groundwater is undoubtedly being depleted in the drier western part of the Barind
Tract in north-west [3], but that is largely due to farmers’ extravagant method of flood
irrigating rice; more efficient methods of irrigating rice are known – Alternate Wetting and
Drying (AWD) and the System of Rice Intensification (SRI) – that could greatly reduce water
use [9]. In fact, Table 3 shows that rice production in the NW region increased substantially
throughout the period of supposedly declining rainfall after 1981 reported by Dey et al.
INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 911

mm
4000
Dinajpur Rangpur

3000

2000

1000

xx xxxxxxx 0 x xx x x
1960

1980

1990

2000

2010

1960

1980

1990

2000

2010
1970

1970
mm
3000
Bogra
Legend
10-year running mean 2000
Short or long-term mean
x No or incomplete data
1000
Source. Bangladesh Meteorological Department
0 x x xx

1960

1980

1990

2000

2010
1970
mm
3000
Rajshahi Ishurdi

2000

1000

xxx x x x xx 0 xxx x xx x x x
1960

1980

1990

2000

2010
1970

1960

1980

1990

2000

2010
1970

Figure 7. (a) Annual rainfall at five stations in North-west Bangladesh with 10-year running mean.
(b) Combined 10-year running mean lines of annual rainfall at the five stations shown in.

Table 3. Rice production in North-west Bangladesh 1980–2015 by decade.


Year Aus Metric tons Aman Metric tons Boro Metric tons Total Metric tons
1980–81 854,410 1,552,454 302,925 2,709,785
1990–91 446,480 2,306,670 1,885,820 4,638,970
2000–01 179,990 3,667,740 4,069,320 7,917,050
2010–11 422,756 4,420,645 4,723,831 9,567,222
2014–15 505,296 4,907,322 6,638,862 11,046,480
a
The table shows the three seasonal rice crops: aus = pre-monsoon; aman = late-monsoon; boro = dry-season.
b
Source: Bangladesh Bureau of Statistics [6].
t the strong wind.

Seasonal variability
In countries and regions where the climate is markedly seasonal, the use of monthly
rainfall data may be inadequate for assessing possible changes in the actual begin and end
dates of the rainy season that are important for farmers practicing rainfed cultivation and
for pastoralists dependent on seasonal grazing. Where daily rainfall data are available, it
is preferable to use such data to illustrate variability in the dates when significant
amounts are reached and to identify possible changes in such dates over time. The
selected rainfall amounts will vary between climates depending on crops grown and
grazing practices.
Figure 8 illustrates the great interannual variability of dates when specific rainfall
amounts were reached in the pre-monsoon and post-monsoon seasons at two
912 H. BRAMMER

Rajshahi Sylhet

2005

2015
2000

2010
1980

1985

1995

2005

2015
2000

2010
1985

1995

1980

1990
1990
0 0
X X X
10 10
March
20 20
31 31
1 1
10 10
April
20 20
30 30
1 1
10 10
May
20 20
31 31
1 1
10 10
June
20 20
30 30
1 1
10 10
July
20 20
31 31
1 1
10 10
August
20 20
31 31
1 1
10 10
September
20 20
30 30
1 1
10 10
October
20 20
31 31
1 1
10 10
November
20 20
30 30
1985

1995

2005

2015

1985

1995

2005

2015
2000

2010

2000

2010
1980

1990

1980

1990

X No data Date when 50 mm rainfall received from 1 March


Date when 250 mm rainfall received from 1 March
10-year running mean Date when 250 mm rainfall received before end-November
Date when 50 mm rainfall received before end-November

Figure 8. Variability of the start and end dates of the periods with >50 mm and >250 mm rainfall at
Rajshahi and Sylhet 1980–2016.

stations in Bangladesh where rice is the main rainfed crop grown: Rajshahi in the
dry mid-west, where the mean annual rainfall is 1,516 mm, and Sylhet in the wet
north-east (mean annual rainfall 4,098 mm). Aus and deepwater aman rice plus jute
are the main crops sown in the pre-monsoon season; photoperiod-sensitive aman
rice maturing in October–November is the major crop grown in the post-monsoon
season. Note the later start and earlier end dates of the rains as well as the greater
variability of the pre-monsoon season start dates at Rajshahi versus Sylhet, and also
the different long-term trends at the two stations indicated by the 10-year running
mean lines.
The presentation of the data by pentad, as in Figure 9, indicates that there was no long-
term change in the range of variability over time at either of the stations. However, the
mean lines for Sylhet suggest that the rains have started earlier during the 35-year study
period and that the post-monsoon rainy season may be ending earlier. No consistent
change is apparent at Rajshahi. The analysis of 50 years of data for 31 of Bangladesh’s
meteorological stations for the period 1959–2008 did not show any significant changes in
annual or seasonal rainfall over time [2].
INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 913

Rajshahi Sylhet
50 mm Pre-monsoon 250 mm 50 mm Pre-monsoon 250 mm
10 10 10 10
July July
1 1 1 1
30 30 30 30
20 20 20 20
June June
10 10 10 10
1 1 1 1
31 31 31 31
20 20 20 20
May May
10 10 10 10
1 1 1 1
30 30 30 30
20 20 20 20
April April
10 10 10 10
1 1 1 1
31 31 31 31
20 20 20 20
March March
10 10 10 10
1 1 1 1
1981-85

1986-90

1991-95

1996-00

2001-05

2006-10

1981-85

1986-90

1991-95

1996-00

2001-05

2006-10

1981-85

1986-90

1991-95

1996-00

2001-05

2006-10

1981-85

1986-90

1991-95

1996-00

2001-05

2006-10
2011-15

2011-15

2011-15

2011-15
Post-monsoon 250 mm
Post-monsoon 250 mm
50 mm 50 mm
1 1 1 1
10 10 10 10
Aug Aug
20 20 20 20
31 31 31 31
1 1 1 1
10 10 10 10
Sep Sep
20 20 20 20
30 30 30 30
1 1 1 1
10 10 10 10
Oct Oct
20 20 20 20

31 31 31 31
1 1 1 1
10 10 10 10
Nov Nov
20 20 20 20
1 1 30 30
1981-85

1986-90

1991-95

1996-00

2001-05

2006-10

1981-85

1986-90

1991-95

1996-00

2001-05

2006-10
1981-85

1986-90

1991-95

1996-00

2001-05

2006-10

1981-85

1986-90

1991-95

1996-00

2001-05

2006-10

2011-15

2011-15
2011-15

2011-15

Earliest in pentad Pentad mean Latest in pentad

Figure 9. Comparison of the variability of the start and end dates of the periods with >50 mm and
>250 mm rainfall at Rajshahi and Sylhet 1980–2016 by pentad.

Climate modelling
It is now ca 50 years since global warming resumed after the still-stand in the 1940s to the
1970s shown in Figure 1. That enables environmentalists to check climate model outputs
against actual trends over the past 50 years, using meteorological data for non-urban
areas in the case of temperature trends as indicated above. However, it needs to be
recognised that international model outputs can differ significantly from local reality.
That was illustrated in a comparison of four U.K. Hadley model outputs against north
Indian rainfall and temperature data for 1979–99 [10], summarised in Brammer [2]. In
that comparison, the four model outputs not only differed greatly from recorded data but
they also differed greatly from each other: the four rainfall model outputs differed from
recorded data by between near-zero and +50 mm in January and by between –150 and
–350 mm in July, and the temperature outputs differed from actual monthly data by
between –8 and +13°C in January and between –8 and +5 °C in July.
Climate model outputs show linear trends, but the data presented in Figures 1, 4 and 7
show that historical trends can be cyclical or irregular, and the trends shown in Figures 3
and 7 show that trends are not always uniform within a country. The use of international
climate change models, as in the Bangladesh Delta Plan [4], can exaggerate projected
climate change impacts in countries and regions where Figure 2 shows that temperature
914 H. BRAMMER

changes differed from the global average; regional differences in vegetation and mon­
soon-season water cover can also generate different trends within countries/regions. Such
limitations indicate the need for caution in using climate models to predict potential
future environmental, agricultural production or related economic trends.

Conclusions
Research is needed to seek practical solutions to the limitations of climate data analysis
described above. Environmentalists need to examine long-term climate data for the
country or region in which they are working to see if trends are linear, cyclical or irregular
and apply appropriate methods of trend analysis. To the extent possible, data should be
used from sites least likely to have been affected by urban warming or by changes in
vegetation or land use, and governments need to site more meteorological stations in non-
urban environments. The teaching of geography, at all levels, needs to provide students
with practical experience in analysing meteorological data that will increase their under­
standing of the complexity of assessing climate change and thereby eventually enhance
public knowledge of actual and potential environmental impacts of climate change.
The outputs of international climate models need to be checked against the historical
meteorological data of the country in which climate change studies are being made in
order to find out how closely such models match actual trends since global warming
resumed in the 1970s, taking into account the limiting factors of urban warming and
changes in land use that may have affected recorded temperatures within countries and
may increasingly do so with increasing urbanisation and intensification of land use. The
output of climate models based on unreliable assumptions needs to be revised where
found necessary, and models and decisions based on model outputs need to be checked
and revised as experience is gained. Climatologists making projections for future decades
should note that, as shown in Figure 1, global temperatures have changed erratically in
the past and they could do so in the future; they should therefore usefully make a range of
projections taking into account historical changes in trends over time and they should
periodically update model outputs.

Notes
1. NASA sites change periodically. Figure 2 [11] was downloaded in September 2020. An
earlier NASA site [12] shows that regional temperatures changed irregularly, both in time
and place, throughout the past 135 years, including in the past ten years.
2. Cyclone Amphan in April 2020 intensified to Category V in the Bay of Bengal, but it had
weakened to Category III by the time that it crossed the West Bengal coast adjacent to
Bangladesh. Severe cyclonic storms are included in Table 3 because, like cyclones, they can
be accompanied by a storm surge which often causes more crop damage and loss of human
and livestock lives than Figure titles.

Acknowledgments
The author is grateful to Dr David Lister for assistance in obtaining Hadley global mean
temperature data and, together with Mr Kalyan Rudra, for providing Kolkata rainfall data; to
Dr Md Asaduzzaman for assistance in obtaining data on rice production in Bangladesh; to
INTERNATIONAL JOURNAL OF ENVIRONMENTAL STUDIES 915

Dr C. Terwisscha van Scheltinga for providing a copy of the Bangladesh Delta Plan; to Ms M.
Porter for her kind assistance in the publication of this paper; and to Dr Amir Kassam for valuable
comments made on an earlier draft of the paper.

Disclosure statement
No potential conflict of interest was reported by the author.

References
[1] Morice, C.P., Kennedy, J.J., Rayner, N.A, and Jones, P.D., No date, Quantifying uncertainties
in global and regional temperature change using an ensemble of observational estimates: The
HadCRUT4 data set. (Exeter, UK: Meteorological Office, Hadley Centre).
[2] Brammer, H., 2014, Climate Change, Sea-level Rise and Development in Bangladesh (Dhaka:
University Press Ltd).
[3] Brammer, H., 2016, Floods, cyclones, drought and climate change in Bangladesh: A reality
check. International Journal of Environmental Studies 73(6), 865–886. doi: 10.1080/
00207233.2016.1220713
[4] Government of Bangladesh, 2018, Bangladesh Delta Plan 2100 (Planning Commission,
Government of Bangladesh).
[5] Brammer, H., 2004, Can Bangladesh Be Protected from Floods? (Dhaka: University Press
Ltd).
[6] Bangladesh Bureau of Statistics, 2018, 45 year agricultural statistics of major crops (Aus,
Aman, Boro, Jute, Potato and Wheat).
[7] Dey, N.C., Saha, R., Parvez, M., Bala, S.K., Islam, A.S., Paul, J.K., and Hossain, M., 2017,
Sustainability of groundwater use for irrigation of dry-season crops in northwest
Bangladesh. Groundwater for Sustainable Development 4, 66–77. doi:10.1016/j.
gsd.2017.02.001
[8] Timsina, J., Wolf, J., Guilpart, N., Van Bussel, L.G.J., Grassini, P., Van Wart, J., Hossain, A.,
Rashid, H., Islam, S. and Van Ittersum, M.K., 2018, Can Bangladesh produce enough cereals
to meet future demand? Agricultural Systems 163, 36–44. doi:10.1016/j.agsy.2016.11.003
[9] Kassam, A. and Brammer, H., 2016, Environmental implications of three modern agricul­
tural practices: Conservation agriculture, the system of rice intensification and precision
agriculture. International Journal of Environmental Studies 73(5), 702–718. doi: 10.1080/
00207233.2016.1185329
[10] Farquharson, F., Fung, F., Chowdhury, J.U., Hassan, A., Horsburgh, K., and Lowe, J., 2007,
Impact of Climate and Sea-level Rise in Part of the Indian Subcontinent. (CLASSIC)
(Wallingford, England: Centre for Ecology & Hydrology). Final report, February 2007.
[11] NASA, 2020, United States National Aeronautics and Space Administration, Washington,
DC. www.nasa.gov/blog/3017/making-sense-of-climate-sensitivity/
[12] NASA, No date, United States National Aeronautics and Space Administration,
Washington, DC. www.climate.nasa.gov/interactives/climate-time-machine.

You might also like