Over the last several decades, global warming has been observed on local, regional and global scales. According to the Intergovernmental Panel on Climate Change (IPCC) global averaged combined land and ocean surface temperature show a warming of 0.85 °C (ranging from 0.65 to 1.06 °C) over the period 1880 to 2012 [1
]. Similar to other parts of the world, Bangladesh has experienced an increase in average temperature in all parts of the country [2
], including the northwest region [5
]. This observed trend is also generally true for other climate variables, such as maximum and minimum temperatures. However, the studies on rainfall do not yield a consistent picture, with some studies showing an increase in rainfall over recent decades [7
], whereas others show a decrease [10
]. Several studies agree that rainfall has increased in the southern coastal regions, and possibly also in the north, but may have declined in the central parts of the country [13
]. The differences in the observed rainfall trends may result from the substantial variability of rainfall, coupled with differences in the periods and areas studied.
The impacts of global climate change on agricultural production is very large across the world [15
] and the impacts have already begun to be visible in Bangladesh [18
], which is a small south Asian nation with a land area of approximately 147,570 km2
, and home to about 160 million people. The country greatly relies on an agriculture-based economy, where water resources are highly critical. It is ranked sixth among 170 countries in the Global Climate Risk Index for the period 1996–2015 [20
]. The Asian Development Bank estimated that Bangladesh may experience a 2% GDP loss (annually) by 2050 because of climate change [21
]. As water impacts practically all sectors (people, agriculture, industries and ecosystems), there have been considerable research efforts into predicting or projecting water availability under future climates [22
] to inform the development of effective adaptation options. Therefore, it is important to understand the changes in rainfall that determine variations in water resources. It is anticipated that food and water security in Bangladesh will be under increasing pressure due to socio-economic growth and global climate change. As per recent estimates, the population in Bangladesh will be more than 230 million by 2050 [25
]. The growth of population and expanding economy will result in an increase in water demand. Moreover, the warmer future climate will increase evapotranspiration and hence increase demand for water in irrigated agriculture, urban centers and water-dependent ecosystems. The northwest region of Bangladesh is one of the major food hubs in the country. Therefore, any change in future rainfall and potential evapotranspiration (PET) could have significant implications for socio-economic and agricultural perspectives [5
Presently, global climate models (GCMs) are one of the most used tools for projecting future climates and potential changes in precipitation and PET. The Coupled Model Intercomparison Project (CMIPs) of IPCC has made available GCM outputs for the Program for Climate Model Diagnosis and Intercomparison (PCMDI), and these products are freely available for research. The climate model outputs from Phase 3 of the CMIP (CMIP3) were broadly used in IPCC’s 4th Assessment Report (AR4). In 2013, IPCC released its 5th Assessment Report (AR5), which was based on climate models from Phase 5 of the CMIP (CMIP5), along with greenhouse gas concentration scenarios termed as representative concentration pathways (RCPs). The models from CMIP5 joined with the RCP scenarios have delivered more precise representations of climate outputs than the CMIP3 model results, because corrections were made in regard to some key assumptions of climate that were previously overlooked by the model developers. The CMIP5 models are considered more competent for capturing numerous features of the Asian monsoon climate than the CMIP3 models [27
It is important to note that GCMS are typically run at coarse resolutions (250 to 600 km). Therefore, the GCM outputs are inherently unable to represent regional or local climate features and dynamics at the necessary spatial resolutions for detailed analyses [28
]. Therefore, a hydrological response to climate change is generally projected, using downscaled future climate projections to drive a hydrological model [29
]. One of the key challenges in factoring climate change into water resources management lies in the uncertainty in the projections [32
]. The sources of the projection uncertainties could be from the GCMs, the downscaling approaches, or the hydrological models [34
]. The performance of GCMs over the South Asia region, including Bangladesh, have been investigated by many researchers [36
]. For example, Saha et al. [38
] found that the majority of the CMIP5 GCMs fail to simulate the post-1950 decreasing trend of Indian summer monsoon rainfall, as they did not capture the weakening monsoon associated with the warming of southern Indian Ocean and strengthening of cyclonic formation in the tropical western Pacific Ocean. Some studies have suggested placing more weight on or using only projections from the better performing GCMs. However, it is challenging to select better performing GCMs for a region or country as none of GCMs can reproduce all salient features of global climate [40
]. The uncertainty in climate projections from GCMs and from downscaling approaches must be adequately represented within the specific context and objectives of any water resources management study.
In the past few years, several studies have estimated the changes in precipitation and/or evapotranspiration for Bangladesh [9
], and the majority of the studies suggested an overall increase in monsoon rainfall and decrease in post-monsoon rainfall [41
]. However, these studies used the climate model output from CMIP3 and were driven by the IPCC AR4 scenarios. In recent years, some studies [44
] estimated the future climate and associated extremes for Bangladesh based on the IPCC’s AR5 report. These studies revealed that overall precipitation and temperature are likely to increase in the future over this region. However, these studies are not specific to the northwest region and uncertainties in GCMs projections were not considered. Moreover, neither study examined projected changes in potential evapotranspiration (PET), which is an important input for irrigation demand estimation.
This paper aims to estimate the changes in future rainfall and PET based on IPCC’s AR5 under RCP4.5 and RCP8.5 emission scenarios [1
]. The novelty of this study is that we aimed to examine rainfall and PET jointly, and to encompass the range of projections to capture the full range of uncertainty. At the same time, this avoids artificially inflating the uncertainties by treating the uncertainties in rainfall and PET projections as independent. This was best done by identifying the two GCMs that best showed the highest and lowest projected rainfall, the two that showed the highest and lowest projected PET, and then using the rainfall and PET projections for those GCMs. We also used the projections for the GCM which was closest to the average rainfall and PET, making five sets of projections in all. We used the GCMs which satisfied the criteria across the region, rather than per grid cell, which could potentially introduce discontinuities in projections.
This study demonstrates that projected rainfall change varies from a small decrease to a large increase across the northwest region of Bangladesh. Mean annual rainfall could increase by 11.8% with ±1% across the region, which is positive for the agricultural productions. Increase in rainfall during the Rabi season creates positive impacts, as it reduces the demand for groundwater feed irrigation. Results are consistent with recent studies (e.g., [44
]) on future climate projections where an increase in rainfall, especially for the northwest region, is reported. Based on 17 GCMs, Caesar et al. [44
] projected precipitation changes in the range of −1.25 to 10.28% over the period of 2041–2060 and 11.75 to 23.66% over 2080–2099. Similarly, based on results from five regional climate models, Hasan et al. [46
] reported mean increase of 4 to 35% over Bangladesh and up to 26% in the northwest region over the period of 2050s. However, increase in rainfall may not always produce positive impacts, because there is a possibility of more extreme events, such as floods with harmful impacts on agriculture [17
]. Increase in rainfall during the Kharif season is not positive because there is no shortage of water. An increase in PET is projected by most GCMs. This implies a higher irrigation demand for agricultural production which produces negative impact to threatened groundwater resource.
It is important to note there is a large variability between month and seasons. Although annual rainfall and PET may increase in the northwest region, there is a possibility of a decrease in some months. The changes in rainfall and PET during the Kharif and Rabi seasons may have significant implications for agricultural productions in the region. As reported by Kamruzzaman et al. [47
], increase in future precipitation may reduce the drought occurrence, which is good news for agricultural productions. However, an annual increase in rainfall could be accompanied by a dry season decrease, thus deepening the dry season drought despite the overall increase. There is also a possibility of increasing extreme events such as flood, which has noticeable adverse effects on crop production [46
]. Increased frequency of precipitation ensures regular supplies of water for growing plants and is, therefore, a good outcome from an agricultural perspective. Another important factor that can affect the crop yield is the duration and frequency of wet and dry periods. Therefore, increase in rainfall may not always be advantageous for agricultural production. It is important to note that Kharif season is mostly rainfed and directly impacted by any changes in rainfall and PET. Therefore, increase in rainfall during Kharif season may not be advantageous for agricultural production. As the Rabi season is mostly irrigated agriculture, changes in rainfall and PET could impact production indirectly. For example, increase in PET during the Rabi season indicates that irrigation demand is generally projected to be higher under future climate [18
This study provides an assessment of future changes on rainfall and PET in the northwest region of Bangladesh over the period of 2045–2075 for the RCP4.5 and RCP8.5 emission scenarios. The study was conducted using results from 28 GCMs, based on IPCC’s 5th assessment report. It describes a step by step procedure to regionalize best performing GCMs instead of selecting a different GCM for individual grids. The scenario modelling combining rainfall and PET is an efficient way to capture the uncertainty in future climate projections particularly for agricultural impact assessment. The method captures the uncertainties in projections in a consistent way across the region. The algorithm developed here is applicable to any region. However, if the area is very large and hydroclimate varies largely across the region, it is recommended to divide the area into several sub-zones instead of a single zone.
Increases in rainfall are generally projected in the northwest region of Bangladesh, with some projections showing decreases. Almost all projections show increases in PET. Although increases in annual rainfall and PET are projected, there may be decreases in some months. Therefore, it is recommended to use monthly scaling factors for the studies on extreme events, such as floods and droughts, while annual scaling factor may suffice for long term impact assessment. This study also investigated the impacts on rainfall and PET during two major cropping seasons in Bangladesh (Kharif and Rabi), and found an increase in rainfall and PET for both seasons. At annual scale, the increase in rainfall is greater than the increase in PET for the Kharif season, but less for the Rabi season. While, at monthly scale, a reduction in rainfall was projected for December and January and in PET for July, no reduction was found at a seasonal scale (i.e., Kharif/Rabi). The findings are useful to quantify the impacts on streamflow and its subsequent impact on groundwater, which is the major source of water for irrigated crops. This information is crucial for better adaptation under increased water demand for agricultural and domestic use.