Climate Change Will Aggravate South Asian Cropland Exposure to Drought by the Middle of 21st Century

Drought has a paramount impact on global agriculture and food security. However, the study on future cropland areas that can incur drought is inadequate. This paper uses input parameters from 7 CMIP6 models for 7 future scenarios (SSP1‐1.9, SSP1‐2.6, SSP4‐3.4, SSP2‐4.5, SSP4‐6.0, SSP3‐7.0, and SSP5‐8.5) to measure South Asian cropland exposure to drought and its underlying factors. Some defined epochs such as 2021–2040 (near‐term), 2041–2060 (mid‐term), 2081–2100 (long‐term), and 1995–2014 (reference period) are designed to explore diverse outlooks of the change. The Standardized Precipitation Evapotranspiration Index and the Run theory methods are applied to detect drought. Results indicate an intensified cropland (under SSP4‐3.4, SSP3‐7.0, and SSP5‐8.5) in the Indo‐Gangetic Plain region of South Asia, where mostly the variation occurs among scenarios and periods. Notably, the future cropland exposed to drought will increase in the 2021–2040, and 2041–2060 periods, but it intends to decline during the 2081–2100. Relatively, the exposed cropland will upturn highest by 49.2% (SSP3‐7.0) in the mid‐term period and decrease by −8.2% (SSP5‐8.5) in the end future. Spatially, distributed cropland in the central, south‐west, and portion of the northeast of South Asia are subjective to be exposed largely, but it can drop greatly across the eastern part by the end future. Importantly, the climate change effect plays a grounding role in future exposure change over the region during the near to mid‐term periods, while the cropland change effect is predominant in the long‐term perspectives. However, these findings signify the urgency of policymaking focusing on drought mitigation to ensure food security.


Introduction
Agricultural production is the heart of the global food security.It has been reported that about 95% of global food directly or indirectly comes from agro-farms (FAO, 2015).Where ∼58% of the world's food production is derived from rain-fed crop areas (Seck et al., 2012).Alarmingly, the global agricultural sector has been facing increasing production challenges to ensure food security for a growing population (FAO, 2017).The present rates of increase in crop cultivation areas are well below those required to meet future global food demands (Hall & Richards, 2013).To feed the growing population in the future, global crop production needs to increase by 60% by 2050 (FAO, 2017).So it has been a growing concern to address risks that threaten global agriculture sectors.Some previous studies reported that the fluctuation in global cropland can be influenced by the indirect effect of climate change, especially at the regional scale (Paudel et al., 2019;Smaliychuk et al., 2016;Tsegaye et al., 2010).Typically, regional climate behavior plays a pivotal role in crop production, where temperature and precipitation are the two key factors in controlling annual crop yield status, crop growth, crop health, and yield-based year-toyear cropping patterns (Howden et al., 2007;Kang et al., 2009;Liang et al., 2017).A series of studies already identified that climate stress significantly harms global agricultural production as well as trade and prices, which differs in the change of magnitude across regions (Nelson et al., 2013).Particularly, increasing climate and weather-induced extreme events (i.e., droughts) inflict adverse effects on agricultural production (Troy et al., 2015).In the future, such extremes could accelerate a substantial decrease in major crop (rice, maize, and wheat) yields (Lobell et al., 2011).Thus, climate risk mitigation for the agriculture sector has become the key priority to scientific groups, governments, as well as the affected community in recent decades.
Scientists reported drought as one of the drastic climatic hazards that poses a major threat to agriculture production (AghaKouchak et al., 2015;Guo et al., 2021;Lesk et al., 2016).In general, drought exacerbates water shortage that significantly affects global crop production.It poses a substantial threat to water and crop yields, which can trigger food insecurity and global economic challenges (AghaKouchak et al., 2015;Hao & Singh, 2015).Some previous studies already recounted the impact of drought on the global agriculture system at global, regional, and local scales.According to historical records, from 1964 to 2007, global crop production was reduced by 10% due to drought (Lesk et al., 2016).Global rice yield has declined by 3% due to an increase in extreme drought events each day (Guo et al., 2021).The American southern Great Plains experienced an acute drought in the 1930s, resulting in yield losses of 50%.The drought from 2011 to 2012 in South and Central America had overwhelming impacts on agricultural production (Kogan et al., 2013).During the last 1,000 years, at least 15 adverse drought events that have lasted longer than three years have occurred across multiple Chinese provinces (Zhang, 2005).Annually in 1949-2013, drought-affected 209,000 km 2 of cropland in China (Zhai et al., 2014).Addressing such tremendous effects, global communities are worried about how our future crop production could be affected by droughts under the changing climate.It is expected that the increase in drought frequency (DF) and intensity will affect crop production more acutely (IPCC, 2013(IPCC, , 2021)).Global climate change influences the frequency, larger area coverage, and extensive impacts of drought (Field et al., 2012).In the coming 30-90 years, more wide-ranging and severe drought is predicted to occur across all the climate belts (Dai, 2013).Under the warming climate, it could lead to distress in the water supply, human healthiness, natural environment, and farming (Karl et al., 2009;Zhai et al., 2016).Eventually, emergent attention should be given to understanding the spatiotemporal variabilities in drought and the associated impact on global farming (Zipper et al., 2016).So, studies focusing on how drought will affect future cropland areas has been the key concern to ensure global food security.
South Asian countries are solely dependent on agriculture by means of livelihood (Gumma et al., 2022).Nearly 60% of the people here are engaged in crop production, and 57% of their land is used for agriculture (FAO, 2019).Unfortunately, over the past few decades, surface water shortage and climate variability created huge challenges to the agriculture practice in this region.It is reported that around 23 million hectares of rain-fed crops in this region are the most susceptible to climate change (IRRI, 2009).Climate extremes such as frequent droughts over the region triggered immense difficulties for water availability and agriculture (Aadhar & Mishra, 2017;Mishra et al., 2016Mishra et al., , 2019;;Zhang et al., 2017).In the recent decade, South Asian farmers have incurred large agricultural losses due to persistent dryness.Drought stress may trigger yield losses of up to 40% in the region, which causes income losses of 58% (IRRI, 2009).In Pakistan, drought reduced agricultural production by 2.6% in 2002 (Ahmad et al., 2004).In 1987In and 2002In /2003, more than 50% of the cropland area in India was affected by drought; while 40% of crop production declined due to drought once every 2.5 years in Bangladesh (Aadhar & Mishra, 2017).In the future, it is further expected that increased droughts under changing climate will aggravate agriculture production adversely.The complex monsoon system, atmospheric patterns, and climatic processes will exaggerate drought over the domain (IPCC, 2014), which could lead to substantial challenges for food production (Mishra et al., 2019(Mishra et al., , 2020)).Therefore, studies emphasizing drought impacts on the future cropland would be a fundamental basis for water resources management and maintaining food security in South Asia.
Uncertainties related to input parameters, especially in global climate models (GCMs), greatly interrupt a more realistic outlook of climate-induced impact on cropland (Garcia-Carreras et al., 2015).Previous stages of the Climate Model Intercomparison Project (i.e., CMIP5) are recounted to have several shortcomings in assessing climate features globally (Chen & Zhou, 2015;Saha et al., 2014).However, climate model outputs under the latest CMIP6 phase have improved a lot in some aspects.In this phase, upgraded model structures can provide more reasonable findings (Di Luca et al., 2020).Notably, wide-ranging upgraded scenarios grounded on the Shared Socioeconomic Pathways (SSP-based) are one of the most vital aspects of this phase.In this comprehensive scenario pack (total of eight scenarios), four scenarios such as SSP1-2.6,SSP2-4.5, SSP4-6.0, and SSP5-8.5 are the restructured over RCPs from previous CMIP5, and the remaining four are known as "gap scenarios" (SSP1-1.9,SSP4-3.4,SSP3-7.0, and SSP5-3.4-overshoot)that's are newly developed (Eyring et al., 2016;O'Neill et al., 2016).Such "gap scenarios" are likely to heal the existing dispute in CMIP5.Studies considering these firsthand "gap scenarios" are probable to give more convincing conclusions (Hausfather & Peters, 2020).Therefore, it is imperative to carry out studies considering all the scenario combinations to unfold future changes in more narrative ways.
Though a series of studies have already been conducted to inspect drought's impact on future agriculture production, most of the studies conveyed some shortcomings in reliable and reasonable reflection of future drought and associated impacts due to uncertainties related to GCMs, provision and design of scenarios, potential evapotranspiration (PET) calculation methods, and choice of drought index and so on.Importantly, the majority of the existing studies considered fixed cropland areas (only historical cropland areas) to explore drought impacts.Drought impacts based on future standpoints of cropland change received less attention.A recent study already estimated drought-driven yield loss using an ensemble machine learning approach and crop models (Prodhan et al., 2022).This study considers only precipitation and temperature data under one high emission scenario (SSP5-8.5)scenario to estimate drought which may lack reasonable reflection of future drought.Further, this study mainly considered drought intensity to estimate yield loss and does not report underlying factors that can influence future yield loss.However, our present study includes some new aspects of this issue.In light of these issues, our research emphasizes methodically estimating cropland exposure to droughts across South Asia concerning both climate and cropland change.Particularly, this work designed is to quantify how cropland areas (considering future standpoints) in South Asia could experience drought under different SSP-based scenarios.Moreover, to which degree cropland change effect, climate change effect, and joint cropland-climate effect can influence exposure evolution is further discovered.A quantitative assessment has been conducted using seven GCM outputs, and cropland areas from five Integrated Assessment Models (IAMs) under all seven SSP-based scenarios for the following epochs such as 2021-2040 (near-term), 2041-2060 (mid-term), 2081-2100 (longterm) and 1995-2014 (reference period).Besides, to detect a reasonable reflection of future drought, we used the widely accepted Penman-Monteith method-based PET that considers both dynamic and thermodynamic effects and multiple climate factors.This will provide a more comprehensive overview of drought-driven impact on future cropland.To the authors' best knowledge, South Asian cropland exposure to drought in considering the future standpoints of land use change and climate inputs for all the updated scenarios from the newest CMIP6 is a new addition to scientific literature.This study's outcomes would be critically supportive of inducting dynamic policy approaches to cropland drought risk management focusing on climate change as well as future land use measurements.

Climate Ingredients
In this paper, climate model outputs are used to inspect drought propagation over South Asia.Several climate ingredients such as precipitation, maximum and minimum temperatures, wind speed, shortwave and longwave solar radiation, and surface relative humidity are collected from the CMIP6 website (https://esgf-node.llnl.gov/search/cmip6/).Notably, we used seven available models with all these variables under seven selected scenarios.One of the prime concerns of our study is to consider all the newly developed SSP-based scenarios (altogether seven).We believe that such a more detailed emission scenario package would unfold a broader outline of the future change.These scenario pathways are imperative for diverse research fields and associated policymaking (Eyring et al., 2016).So we consider the models that provide all the necessary variables under all these seven scenarios.Notably, at the time of writing this paper, we found only the following seven models (listed in Table 1) that provide all the required variables under our quest scenarios.Most of the models that participated in the CMIP6 phase made available variables mainly for Tier-1 scenarios (SSP1-2.6,SSP2-4.5,SSP3-7.0, and SSP5-8.5).A limited number of models are available with all required variables under all the scenarios.So, as other models lack the necessary variables for this research, we have chosen these seven models for our analysis.Table 1 illustrates more detailed features of the nominated GCMs.Moreover, scenarios in the CMIP6 phase are categorized into two key platforms such as Tier-1 scenarios (SSP1-2.6,SSP2-4.5,SSP3-7.0, and SSP5-8.5)and the Tier-2 (SSP1-1.9,SSP4-3.4,and SSP4-6.0).Tier-1 nominators are the most vital for climate science studies in the aspects of IAM (O'Neill et al., 2016).Importantly, spatial distribution (projected cropland, DF, and population exposure) for Tier-1 scenarios are represented in the main manuscript, and the remaining Tier-2 nominators are to be found in the supplementary documents.
Further, all the climate outputs used in this study are bias-corrected and downscaled.This work ensures a common horizontal resolution (of 0.5 × 0.5) for the climate inputs by applying the Spatial Disaggregation approach (Su et al., 2018).Further, inherited biases in GCMs are amended by adopting the Equidistant Cumulative Distribution Functions (EDCDF) in line with reanalysis data (observed).It involves dividing the cumulative probability range into equidistant intervals, making it intuitive and straightforward.The selected EDCDF bias correction method is relatively simple to understand and implement robust, and effective in reducing biases compared to some other bias correction methods (Yang et al., 2018).The observed climate features are downloaded from the Inter-Sectoral Impact Model Intercomparison Project website.This data sets are produced based on WATCH Forcing Data and ERA-Interim (EI) reanalysis data.These outputs are available under the title of EWEMBI (EartH2Observe, WFDEI, and ERA-Interim data Merged and Bias-corrected for ISI-MIP).However, the GCM's performance in terms of capturing observed climate (precipitation, PET, and temperature) and drought features (duration, frequency, and intensity (Figures S1-S5 in Supporting Information S1) are evaluated further.The evaluation for climate variables is carried out considering the annual scale (Figure S1 in Supporting Information S1) and seasonal cycle (Figures S2-S4 in Supporting Information S1) both before (raw) and after biascorrection.The evaluation results suggest that all the downscaled and bias-corrected CMIP6-GCMs demonstrate more promising performance in reproducing climate patterns, where PET and temperature simulation were much better than precipitation.Though the drought features show some regional differences, overall it was satisfactorily captured by the GCMs (Figure S5 in Supporting Information S1).Moreover, the ensemble mean value reveals even enhanced performance (reduced uncertainty) than individual models.So, all the results presented in this manuscript consider multi-model ensemble mean (MME) (after bias-correction).However, the methodological process of downscaling and bias correction is illustrated by Su et al. (2018).

Cropland Data
To see the actual agricultural land status in South Asia, the World Bank (WB) logged cropland area for the 1995-2015 period is taken into consideration in this manuscript (https://data.worldbank.org/indicator/AG.LND.CROP.ZS).Besides, we used the spatial distribution of cropland fraction in 2000 provided by the Socioeconomic Data and Applications Center (SEDAC).Detailed can be found in the study by Ramankutty et al. (2008).Moreover, future cropland data produced under the Land-Use Harmonization 2 (LUH2) is available on the following website: http://luh.umd.edu/.Notably, five IAMs (IAM) as part of CMIP6 are nominated to produce such new gridded land use products under 7 scenarios (see Table S1 in Supporting Information S1).It represents annual gridded fractions of land at 0.25°× 0.25°resolution published under all the SSP-based scenarios.More detailed descriptions of the new land-use data sets are styled by Hurtt et al. (2020).Importantly, as the C 3 crops (especially rice, wheat, and maize) are the main cultivated crops in the South Asia region, this study emphasized only the annual C 3 cropland fraction.However, the original grid size of the LUH2 product was upscaled to 0.5°× 0.5°by using the "nearest-grid" interpolation technique.In order to deduce the gridded total cropland area (km 2 ), the proportion in each grid is multiplied by the total area of the corresponding grids of South Asia.
The temporal distribution of recorded cropland for the period of 1995-2015 and the spatial pattern of cropland fraction both for the years 2000 and 2015 are represented in Figure 1. Figure 1a (based on WB data) shows that the cropland area was increasing from 1995 to 1999, and afterward started to decrease.The highest cropland area was 2,721,696 km 2 in 1999 and the lowest was 2,692,330 km 2 in 2010 across South Asia.Further, a downward trend from 1995 to 2015 was revealed, which is an indication of reducing cropland status in the future over South Asia.
Geographically, the cropland fraction in 2000 (Figure 1b based on SEDAC data) and 2015 (Figure 1c based on LUH2 historical simulation) varied between 0 and 1. Spatially, the highest cropland areas over South Asia, are distributed across the Indo-Gangetic Plain (IGP), covering the northeastern part of India, and portions of Bangladesh, Nepal, and Pakistan; and further -the southern Indian region (Figures 1b and 1c).Whereas, the lowest cropland area is distributed across southwestern Pakistan, Afghanistan, and the mountainous, northern border.It is obvious that the model simulation distribution (Figure 1c) shows a similar pattern to observed (Figure 1b).
Notably, to identify the most vulnerable cropland area that can experience adverse drought, we categorized four cropland zones over the studied region considering the intense cropland distribution in 2000 (Figure S1 in Supporting Information S1).We classified the area as a rare cropland zone for the fraction range from 0 to 0.1; a low cropland zone ranging from 0.1 to 0.3; a moderate cropland zone between 0.3 and 0.7, and a high cropland zone with values from 0.7 to 0.9.The high cropland zone is mostly pronounced over IGP as well as portions of the southern Indian region.The medium cropland zone is mainly concentrated in the central to eastern and the southern Indian region, and scattered over other areas.The low cropland zone primarily covers western Afghanistan, scattered in other regions.Whereas, rare cropland areas are identified across the southwest portion of Pakistan and Afghanistan, and the northern border region of South Asia (mountainous regions).These classified zones are further used to explore zone-based changes in exposed cropland to drought.

Drought Reflection Metric
There are numerous metrics to diagnose drought globally, where, to depict South Asian droughts, the Standardized Precipitation Evapotranspiration Index (SPEI) is one of the best-performed indices (Adnan et al., 2018;Gupta & Jain, 2018).It is further widely used to estimate the socioeconomic impact of droughts.Some recent studies suggested the SPEI index to estimate drought impact on cropland (Mohammed et al., 2022;Wang et al., 2023).Moreover, this index with the 12-month time log performs better in capturing long-term trends as well as inter-annual inequities in drought events (Chen et al., 2018).So the use of SPEI-12 in estimating cropland exposure would be the signpost of the long-term water availability which has great significance for irrigation planning over the drought-affected cropland area.Notably, this study does not represent the production/crop loss or yield loss directly due to drought but rather implies a broad outlook of how future drought (water deficit) could evolve in the area with extensive crop cultivation in terms of exposure.The following is the elementary notation of SPEI computation: Here, water surplus/deficit (D), precipitation (P), and potential evapotranspiration (PET).However, PET is the pivotal input of SPEI.Different PET estimation methods can largely influence drought characterization (Zhou et al., 2020).Food and Agriculture Organization authorized the Penman-Monteith method has been one of the comprehensive and widely accepted approaches, which considers thermo-dynamic effects (Dai, 2011;Sheffield et al., 2012;Trenberth et al., 2014).The following basic notation deliberated by Allen et al. ( 1998) is adopted to estimate PET: Earth's Future   2010) stated the step-by-step SPEI computation procedures.The SPEI ≤ 1 is taken as the signboard of the drought condition.

Drought Characterization
The widely recognized "run theory" approach is adopted here to explore spatial-temporal drought features.This technique clarifies one drought event when continuous SPEI ≤ 1 lasts for a minimum of 3 repeated months.Further, how many drought incidents transpire over a region annually called DF.Likewise, maximum grid cells that incur waterfall for a certain duration are characterized as drought area (DA) over the total land (%).The detailed notation of this characterization is outlined by Guo et al. (2018).

Analysis of Cropland Exposure to Drought
The crop-cultivable land located in physical surroundings that can incur drought events is called drought-induced cropland exposure (IPCC, 2014).However, this phenomenon can be expressed in quantity by multiplying the estimated DF with the total cropland area over a drought-prone region (Jones et al., 2015).In this case, the proposed cropland under seven scenarios is combined with corresponding climate change scenarios.However, cropland exposure is assessed by applying the notation stated by Güneralp et al. (2015).
where, CE i,j denotes affected cropland at grid i to hazard j; C i is the cropland at grid i, and F i,j is the frequency of drought j at grid i.

Contribution Rates Estimation of Underlying Factors in Exposure Changes
Three key influential aspects are expected to be involved in exposure evolution such as cropland change feature, climate change feature, and cropland-climate joint feature.The following notation represents the three key decomposing aspects: where ΔD = total change in exposure; X i = cropland status (km 2 ) in the period i; Y i = DF in the period i; X j = cropland status in the period j; Y j = DF in the period j; ΔX = cropland change from i to j time period; ΔY = DF change from period i to j; Y i + ΔX = cropland change effect; X i × ΔY = climate change effect; ΔX × ΔY = joint cropland-climate change effect.Finally, the relative contribution of each factor is computed, as follows: where CFR crop signifies the cropland change effect; CFR cli embodies climate change effect; and CFR crop-cli denotes joint cropland-climate effect.

Projected Changes in Key Climate Drivers
Typically, drought events over a region can be better interpreted by the changes in precipitation and PET.The decline/lowering in precipitation leads to drought onset and it's further exaggerated with boosted PET.Anticipated changes in precipitation (Figure 2) and PET (Figure 3) are illustrated in this section to explore the wide-   Earth's Future 10.1029/2023EF003872 ranging outlook of future dryness over South Asia.The two-sample t-test is applied to identify the significant change at the 95% confidence level.
The largest increase is estimated under the high emission scenario (SSP5-8.5)during all three periods, whereas it is lowest under SSP2-4.5 during near-, and mid-term periods and SSP1-1.9 in the long-term period.Relative to the historical period, the spatial distribution of precipitation shows slight regional variations over time and scenarios (Figures 2b-2m).The decrease in precipitation is mostly prominent in the western part of South Asia (particularly Afghanistan and Pakistan regions), whereas the tremendous increase is intended to be dominant over the eastern part.Though some parts of the eastern region is expected to experience reduced precipitation during the near-term period (i.e., SSP2-4.5), it will experience increased precipitation toward the long-term period.Remarkably, a continuous decrease is observed in Bangladesh in almost all the periods and scenarios.Further, the area with increase (significantly) is more dominant than decrease (significantly) for all the periods and scenarios.The highest area (∼38%) with overall decreased precipitation is exhibited under SSP2-4.5 during the near-term period (Figure 2e), while the significant decrease is found over ∼4% area (highest) of the total lands under SSP1-1.9during the long-term period (Figure S7c in Supporting Information S1).Contrary, the area with an overall increase in precipitation is dominant in the long-term period, particularly high emission scenarios.The highest ∼88% (where 79% area significantly) of the total area is expected to experience increased precipitation under the SSP5-8.5 during the long-term period (Figure 2m).So it can be highlighted that future precipitation over South Asia will increase extensively during the long-term period, particularly high emission scenarios (i.e., SSP5-8.5),while it is slothful in the near-term period.The larger area in the eastern part will experience increased precipitation, while it will decline over the western area.
Figure 3 demonstrates changes in PET status over the South Asia domain under the nominated scenarios and periods.In total, projected PET is inclined to increase greatly toward the long-term period.Relative to the historical period, PET is likely to decrease in the near-term period under most of the scenarios (except SSP1-1.9 and SSP1-2.6),whereas it increases under all the scenarios in the mid-term (except SSP4-6.0) and long-term periods (Figure 3a).The highest increase by 5.4% in PET is estimated under the SSP5-8.5 in the long-term period.Spatially, the changes in PET shows slight regional contrast during near-term and mid-term periods among scenarios, but it shows almost homogenous changes (increase) over the entire domain (Figures 3b-3m; Figure S8 in Supporting Information S1).During the near-term and mid-term periods, PET mostly shows increasing trend over the western region of the South Asia and decreasing trend in the eastern part (Figures 3b,3e,3f,3h,3i,and 3k and Figures S8d,S8g, and S8h in Supporting Information S1).During the long-term period, more than >95% of total area in South Asia will experience significantly increased PET under all the scenarios, except SSP4-6.0 (∼73%).So, it is evident that PET will increase significantly over the entire South Asia by the end of the century but is likely to decrease in the early century, particularly eastern part.
Obvious differences among scenarios are perceived in the projections.The highest increase in projected cropland is found under SSP5-8.5 in the 2015-2055 period.Further cropland projection reaches a peak under SSP4-3.4during the period of 2055-2085.The increase in cropland continued under SSP3-7.0toward 2100.The estimated cropland change is slothful for SSP1-1.9 and SSP1-2.6,while SSP2-P4.5 and SSP4-6.0exhibit almost consistent inclination with the historical trend.However, the highest cropland is estimated under SSP5-8.5, which is 1.39 × 10 6 km 2 in 2021-2040 and 1.44 × 10 6 km 2 in 2041-2060.Whereas the greatest increase in 2081-2100 is predicted to be 1.51 × 10 6 km 2 under SSP3-7.0.
In the historical period, concentrated cropland greater than 1,500 km 2 was distributed in the IGP region, extending mostly over Bangladesh, and west Bengal, Bihar, and Punjab of India (Figure 5).Further, cropland areas above Earth's Future 10.1029/2023EF003872 MONDAL ET AL.
1,000 km 2 were pronounced in the southern Indian region: Madhya Pradesh, Maharashtra, Telangana, and Karnataka of India, and the Western part of Afghanistan (Badghis and Maimanah provinces).By contrast, the lowest cropland area, below 100 km 2 can be found across southwestern Pakistan, Afghanistan, and the northern border region of South Asia.For future periods, the spatial pattern of cropland across South Asia is similar to the historical period.Although the spatial distribution is similar, there are some differences in magnitude among scenarios and periods.Importantly, the spatial pattern of cropland is more concentrated and pronounced in the IGP region, where mostly the variation occurs among scenarios and periods.In terms of periods, the cropland pattern is predicted to increase from the 2021-2040 to 2041-2060 period over Bangladesh, and west-Bengal, Bihar, and Punjab of India for most of the scenarios.

Changes in Future Drought Features
Here we reveal the discrepancy in drought trends in the South Asia region using the SPEI-12 for the defined future periods and scenarios (Figure 6).Compared to the historical period, overall dryness in the South Asia region is expected to concentrate in the western part of the South Asia (particularly Afghanistan and Pakistan regions), whereas the wetness is intended to be prevailing over the eastern part.Notably, continuous dryness is witnessed in Bangladesh in almost all periods and scenarios.The spatial outline of dryness is more concentrated and pronounced in the western region, where mostly the variation occurs among scenarios and periods.Though the overall pattern is mostly similar among scenarios and periods, there is an obvious difference in area coverage and degree of change.The area with the significant increase in dryness is likely to expand toward the long-term period, especially under the low emission scenarios.The highest area coverage with significant dryness among scenarios varies by the highest ∼29% (under SSP1-1.9; Figure S10i in Supporting Information S1), ∼31% (under SSP1-1.9; Figure S10i), and ∼38% (under SSP1-2.6; Figure 6c) during the near-, mid-, and long-term period, respectively.Whereas the lowest area with significant dryness intends to decline toward higher emission scenarios.However, the region with increased dryness (western part) is expressively homogeneous with the region where a continuous decrease in precipitation and increased PET (Figures 2 and 3).So, it is obvious that the decline in precipitation and increased PET could lead to such a drying trend.Further, dryness significantly decreased (increased wetness) in the eastern part where the precipitation is intended to increase largely (Figure 2).So, it is apparent that the degree of the changes in precipitation would be the key driver in modulating future drought in the South Asia region, while the increased PET will worsen the drought condition further.) period is anticipated to rise strongly under the nominated scenarios, such as ∼30% (to 6.84 times/year) under SSP4-6.0 and the lowest 19.3% (to 6.2 times/year) under the SSP5-8.5.Further, DF during 2041-2060 tends to escalate from a lower forcing scenario to a higher one.In this epoch, the DF upsurge can be 22.3% higher (6.4 times/year) under SSP3-7.0.Where the least increase is projected under the sustainability pathway scenario combination (SSP1-1.9)by 7.5% (5.7 times/year).Dramatically, DF will drop in the long-term period (2081-2100).It is surprising that the DF is largely reduced for higher emissions such as 14.9% (4.4 times/ year) under the SSP5-8.5 and followed by 2.0% (SSP3-7.0)and 0.9% (SSP4-6.0).However, in this epoch, DF can increase by 17.4% (6.1 times/year) for SSP1-2.6.
For the affected area, the area coverage was 14.0% of the total grid cells in the reference period (Figure 7).In the coming future (2021-2040), affected area coverage is found to rise for all the SSP-based scenarios.In relative terms, the highest increase in area coverage for an individual event in this period is 14.2% (SSP4-6.0)and followed by 12.2% (SSP1-2.6)and 10.1% (SSP1-1.9).In the middle period (2041-2060), the largest area coverage (by 18.3%) is found under SSP5-8.5,whereas the strongest reduction (by 13.3%) was under the sustainable pathway scenario combination SSP1-1.9.In 2081-2100, the DA will be highly aggravated under SSP2-4.5 (13.0%), and a surprising reduction in the affected area found under SSP5-8.5 by 6.2%.To sum up, the temporal storyline of the DF and affected area status are largely consistent under all the SSP-based for all three future epochs.The utmost growth in DF is expected in 2021-2040 for SSP4-6.0, and the largest area coverage was found during 2041-2060.Whereas, the biggest reduction in DF and DA are found under SSP5-8.5 and SSP1-1.9 in the 2081-2100 and 2041-2060 periods respectively.and Afghanistan) is pronounced mostly under lower emission scenarios SSP1-1.9(Figure S8n in Supporting Information S1), SSP1-2.6 (Figure 8b), and SSP4-3.4(Figure S11q in Supporting Information S1), whereas more prominent pattern (with ≥80% relative increase) is exhibited in north-western and western India under higher emission scenarios (mostly SSP4-6.0 and SSP3-7.0).Besides, DF under SSP4-6.0(Figure S11t in Supporting Information S1) is extended toward the IGP region.In 2041-2060, DF will escalate mostly in central South Asia and IGP region for all the scenarios.In this epoch, the increase of 80% is more prominent under the combination of the SSP5-8.5 (Figure 8l).For 2081-2100, increased DF shifted toward the western part of South Asia, while it decreases largely in the eastern and central parts.Such a decrease over a large area is prominent under a higher emission scenario.The greatest decrease (< 20%) in DF is found under SSP5-8.5.Boldly, DF in the future would be overriding mostly across the central part during the 2021-2040 and 2041-2060 periods, whereas the  Earth's Future 10.1029/2023EF003872 western region (Afghanistan and Pakistan) will experience a dominant increase in 2081-2100.Importantly, DF will drop largely over the eastern-central region.

Cropland Exposure to Drought
Figure 9 reveals that the overall cropland area is exposed to drought across South Asia.Relatively, the estimated cropland area that experiences drought is envisioned to amplify during the near-term period (2021-2040) for all the nominated scenarios.During this epoch, the total drought-exposed area (highest) is projected to be 4,991 km 2 under SSP4-6.0,which is 40.8% higher than the historical period.While SSP1-1.9 shows the smallest increase (by 18.0%).Further, in 2041-2060, the larger exposed area is estimated at 5,287 km 2 (49.2% higher than the historical period) for SSP3-7.0.In contrast, exposure is projected to decrease largely toward higher emission scenarios 2081-2100.Surprisingly, the greatest decrease in drought-exposed cropland is estimated by 8.2% under SSP5-8.5.It can be highlighted that drought-affected cropland across South Asia is of great concern for the 2021-2040, and 2041-2061 periods, while it will drop in the 2081-2100.
Figure 10 displays the outline of the drought-exposed cropland for the historical period and relative changes for three future periods under different climate change scenarios.In the historical period, cropland exposure (by 5,000 km 2 ) was pronounced across IGP regions.Exposed cropland in the south-central part of India was found to be 3,000-4,000 km 2 , while the lowest (1,000 km 2 ) exposure was across south-western Pakistan, Afghanistan, and the northern border region.In the near-term period, exposed cropland is anticipated to upturn by 40%-80% in the region relative to the historical period.The exposed area will escalate by >80% mainly in the south-west part of   Pakistan and Afghanistan as well as in the north-east part of India which is dominant under SSP4-3.4(Figure S12q in Supporting Information S1).Whereas, an increased pattern (by 80%) was prominent across central south Asia and IGP under SSP3-7.0 and SSP4-6.0.Again, cropland exposure is projected to increase (>80%) across the south-west part of Pakistan and Afghanistan as well as the north-east part of India, under SSP4-3.4and SSP4-6.0(Figures S12r and S12u in Supporting Information S1).The increased pattern of 80% is largely pronounced under SSP5-8.5 (Figure 10l) in central South Asia and the IGP region.In 2081-2100, exposed cropland will decline across South Asia under all the SSP-based scenarios.The largest decline in the eastern to central areas is more dominant toward higher emission scenarios.Surprisingly, exposed cropland decreased over a larger area under SSP5-8.5 (Figure 10m).But the increased exposure is pronounced mostly across the western region.Further, cropland exposure will increase by >80% across the south-west part of Pakistan and Afghanistan as well as the north-east part of India, under SSP4-3.4and SSP4-6.0(Figures S12s and S12v in Supporting Information S1).However, cropland exposure is pronounced over the central, south-west, and portion of the northeast part.
Overall, exposure is expected to decline over eastern South Asia for all the nominated scenarios, toward 2021-2040 to 2081-2100.
To discern the zone-wise cropland exposure to drought, relative exposure changes for defined time epochs (2021-2040, 2041-2060, and 2081-2100) and the selected scenarios are publicized in Figure 11.For 2021-2014, the exposed area in the high cropland zone will upsurge evidently, especially in the direction of higher emission scenarios.With regard to 1995-2014, the highest 42.39% area of high cropland zone will be exposed to drought under SSP4-6.0,while the largest 46.22% area of high cropland is prone to experience drought in 2041-2060 for SSP5-8.5.Notably, compared to the near and mid-term periods, the lowest area percentage of this zone is likely to incur drought in 2081-2100 under all the scenarios.The exposure is projected to decrease toward higher emission scenarios and the largest decrease ( 12.69%) is estimated in SSP5-8.5, while exposure is intended to upturn by 25.0% for SSP2-4.5 scenarios.Likewise, for the moderate cropland zone, the exposure is anticipated to rise by 36.09% under SSP4-6.0 in 2021-2040, then it reaches 40.37% in 2041-2060 for SSP5-8.5.Further, in the longrun epoch (2081-2100), the amount of affected area across the moderate cropland zone would decrease largely ( 14.15%), predominantly for SSP5-8.5.However exposure in the low cropland zone is estimated to be the highest: 42.83% and 53.23% under SSP5-8.5 for the 2021-2040 and 2041-2060 periods, respectively.Under SSP3-7.0, it was 41.55% and 51.61% in these two periods, accordingly.In the long-run epoch (2081-2100), the highest affected areas (37.46%) of the low cropland zone were projected to be exposed under the SSP3-7.0scenario.To sum up, cropland areas for all three zones are evidently prone to incur drought mostly in the near-to mid-term epoch, where increases are inclined toward higher emission scenarios.Notably, for all the zones, leading exposure is estimated in 2041-2060 under the 46.22% of high cropland;40.37% of moderate cropland,and 53.37% of low cropland).The cropland area exposed to drought will be demolished in 2081-2100 for all the zones across South Asia.

Underlying Issues on Exposure Change
Figure 12 illustrates the contribution rates of underlying factors in South Asian drought-induced exposure changes for each scenario and period.In the 2021-2040 period, the role of climate change effect on total exposure change is projected to increase.The largest contribution rates of climate change effect even reach 97% under SSP4-6.0,while the cropland change effect contributes 34% in SSP5-8.0.Likewise, the climate change effect is a predominant issue in 2041-2060 for the nominated scenarios, where the largest contribution is 90% under the SSP4-6.0.However, for 2081-2100, the cropland change effect contributes as the leading underlying factor, specifically for SSP4-3.4,SSP4-6.0, and SSP3-7.0.In this period, the significant contribution of cropland change effect reaches more than 85% under SSP3-7.0.So, climate change is expected to be the prime issue of South Asian cropland exposure to drought, principally in the near-term and mid-term periods.On the other hand, the cropland change effect plays the leading role in the long-run epoch (2081)(2082)(2083)(2084)(2085)(2086)(2087)(2088)(2089)(2090)(2091)(2092)(2093)(2094)(2095)(2096)(2097)(2098)(2099)(2100).For the climate change effect (in 2021-2040 and 2041-2060), the prominent contribution reaches about 97% and 90% respectively under SSP4-6.0,while SSP3-7.0indicates the highest contribution of cropland change effect by more than 85% for 2081-2100.

Conclusion and Discussion
Socioeconomic exposure assessment has been an inevitable part of climate risk assessment in changing global climates (Wang, Jin, & Liu, 2020;Wang, Wang, et al., 2020).The exposure assessment connotes the rapidity at which the socioeconomic system could be affected by the hazards.In this paper, we integrated the DF and future Earth's Future 10.1029/2023EF003872 MONDAL ET AL.
cropland to explore the associated area quantity that is prone to experience drought in South Asia for different time epochs (1995-2014, 2021-2040, 2041-2060, and 2081-2100) and scenarios.
Dryness over a region is mainly described by the changes in precipitation and PET, where decreased precipitation and increased PET accelerate drought events (Cook et al., 2020;Huang et al., 2022;Miao et al., 2020;Su et al., 2018Su et al., , 2020)).Likewise, our results suggest that the dryness pattern over the South Asia domain is expressively homogeneous with decreased precipitation and intense PET (Figures 2,3,and 6).With regard to historical, future dryness is likely to be prominent over the western part of South Asia, while it will decline largely (increase wetness) in the eastern part.It is evident that the combined effect of decreased precipitation and larger PET in this region will aggravate dryness, contrary, the significant upsurge in precipitation over the eastern part will enhance wetness (decrease dryness), especially in the long-term perspective.Typically, the decline in precipitation would lead to soil moisture deficit.Limited actual evapotranspiration due to drying soil further enhances the atmospheric aridity (vapor pressure deficit; VPD) which decreases the probability of precipitation.Likewise, atmospheric evaporative demand increases due to high VPD, which further worsens soil water scarcity where PET becomes higher, resulting in drought (Teuling et al., 2013).This phenomenon can be explained as the land-atmosphere coupling.However, some previous studies also reported that both rainfall and evapotranspiration are expected to be pronounced in Asia, where rainfall is even greater than PET over India (Ha et al., 2020).Furthermore, for drought characteristics, the DA coverage shows strong consistency with the DF trend.Relatively, both the DF and DA are inclined to increase for near-to mid-century and decrease in the longrun period, where the largest reduction is found under SSP5-8.5.In spatial terms, a strong westward shift of droughts is projected from near-term to long-term periods across South Asia, and a substantial decline is found in the eastern part.Huang et al. (2022) also reported a reduction in drought conditions in South Asian countries for a long-term perspective.However, dwindling precipitation would influence DF and DA in the early century, and tremendous precipitation would diminish drought occurrence in the late part of the century (see Table S2 in Supporting Information S1).Ample precipitation over the region will weaken in future droughts (Cook et al., 2020;Wang, Jin, & Liu, 2020;Wang, Wang, et al., 2020).Mondal et al. ( 2021) also specified similar observations over the region.Lessening precipitation can trigger drought with short duration, whereas events with prolonged duration happen due to escalated PET.Fundamentally, increased evaporation can stimulate the durability of drought largely compared to precipitation, especially creating a shortage of runoff and soil water content (Sun et al., 2019).Further, Our exposure investigation indicates a clear disparity in the exposed cropland among the scenarios and the periods.Overall drought affected cropland area in the studied region is found to be stronger during 2021-2040, and 2041-2060, contrary to weaker in 2081-2100.Boldly, the prominent exposure is estimated in 2041-2060 than the near-term period for most of the circumstances.With regard to 1995-2014, exposure is likely to inflate by 49.2% (highest) during 2041-2060, especially for SSP3-7.0.Unexpectedly, the greatest reduction in exposure by 8.2% is estimated during 2081-2100 for SSP5-8.5.However, such pronounced increases and decreases in exposure are mainly derived by the number of drought events that occur over the region as well as areas with droughts and the degree of cropland change during a particular period.Drought frequency in the mid-term period was higher (∼22%) under SSP3-7.0than others which is the grounding aspect to soar exposure in this period.Whereas a surprising decline in the long-term period is predominantly influenced by the decrease in DF, as well as reduced DA and the downward cropland trend.In this period, precipitation increases faster than PET (wet conditions) which influences DF reduction, especially under SSP5-8.5.When precipitation increases by >25% contributes to the decrease in DF in the long-term period (Table S1 in Supporting Information S1).Spatially, dominant exposure emerges across the central, south-west, and portion of the northeast part of South Asia.Further, exposure could decline largely in the eastern region, especially on the way to 2081-2100 and for all scenario cases.However, during 2041-2060, cropland across South Asia is highly vulnerable to being adversely affected by drought under higher emission scenarios.Our findings can be validated by some previous studies.Based on CMIP5, the global socioeconomic risk of drought is reported to be higher in 2046-2065 than in 2016-2035 under higher emission scenarios (Liu & Chen, 2021).Such adverse exposure to drought could lead to food shortage and hike food prices adversely over the region.FAO (2021) stated that about 70% of global cropland incurred different types of droughts in 2012, resulting in global food prices rising.Likewise, in 2022, a sizable amount of crop yield loss in the studied area is estimated due to the extensive hot and dry conditions (USDA, 2022).Some recent studies also reported dominant drought risk to South Asian cropland (Mahto & Mishra, 2023;Wang et al., 2023).From scenario perspectives, cropland exposure is dominant under SSP4-6.0 in 2021-2040, which is the combination of socioeconomic scenario SSP4 with medium-end forcing 6.0.Under this scenario, the regional climate effect of land use is designed to be stronger (O Neil, 2016).The SSP4 implies inequality (a road divided) where strong regulation on land-use change and a high increase in cropland for highincome countries are predicted.But low-income countries are expected to become unproductive in agricultural activities, while a tropical deforestation tendency may occur.These storylines of SSP4 strongly corroborate our findings that large amounts of cropland in low-income regions are prone to experience drought, as a result, cropland over some regions can decrease due to lack of water for irrigation.Further, SSP3-7.0 demonstrates the overruling exposure during 2041-2060.This "gap scenario" (SSP3-7.0)embraces the greater societal vulnerability case (SSP3) and high-end forcing (7.0).In this combination of scenarios, land-use change regulation is limited and the cropland increase rate is projected to decline over time, especially in developing countries.Furthermore, unequal distribution of future socioeconomic exposure to climate extremes and the unpredictability of anticipated climate change could influence drought damages beyond the average value at the national level (Hsiang et al., 2017).In this study, cropland exposure in each grid cell is estimated to the intense agriculture zone to explore disparity in total exposure and density.For the zone-based analysis, exposed cropland will be prominent in the high cropland zone, mainly in 2021-2040 and 2041-2060.Notably, this exposure is distributed across the IGP area; pronounced mostly over Bangladesh, and west-Bengal, Bihar, and Punjab of India.Im et al. (2017) also informed that population and agriculture activity in the IGP region are adversely affected by climatic hazards.Further, during 2081-2100, exposure can be dominant in the low cropland zone which is located in the western part.However, such exposure is a signpost of reducing agricultural production in the future.Therefore, it is likely that food product prices will rise in countries that import food from the South Asia domain as supplies are reduced, with an especially severe effect in the poorest and most vulnerable countries.According to the Organisation for Economic Cooperation and Development-Food and Agricultural Organization Agricultural Outlook for 2011-2020, yield-based production variations (mainly due to droughts/floods) in prime cropexporting countries have been the main source of worldwide price instability.It is also anticipated that international price volatility will be even more serious due to climate-induced cropland changes.Drought in large crop-exporting countries is likely to continue causing rising pressure on global food prices in the future.However, by altering the trade structure, rich countries can lessen their food uncertainty.For instance, after the cascading effect of drought and wildfires in 2010, Russia stopped all exports (Wegren, 2013).World food insecurity is thus influenced by supply and demand in crises.Hence, control measures for future drought mitigation in high cropland zones must grab the highest priority.
Further, the South Asian cropland exposed to drought is principally influenced by the climate change effect, particularly the contribution is highest at 97% in 2021-2040 and 90% in 2041-2060.During these periods, cropland change slows down but climate change-induced drought escalates faster with time.Whereas it is the cropland change effect for 2081-2100, which reaches >85% under SSP3-7.0.This effect implies that cropland may change faster over time but drought formation will slow down in this period across South Asia.A similar result for socioeconomic exposure (i.e., population exposure) to drought was also stated by Mondal et al. (2021) that climate effect is dominant during 2021-2060 while socioeconomic parameter change effect contributes largely during late future.However, climate change aggravated cropland exposure to drought could be more acute due to reducing soil moisture, surface water supply, and precipitation deficit.Therefore, advanced technology design for climate change mitigation through CO 2 dropping globally is urgent.Whereas, strong land-use change regulation implementation in South Asian countries needs to be a priority in the long-term period (2081-2100).A well-designed policy provision of cropland balance is essential to ensure a safe domestic-and international-level food supply.Therefore, to maintain future cropland growth and reduce exposure, a more coordinating and adaptive governance institution is imperative in the South Asia Domain.Multiple adaption strategies can be taken, such as irrigation infrastructure development, adaptive crop varieties, crop rotation, cropping conditions, and surface water storage to cope with drought and sustain food production.However, policymaking and effective However, this study discourses some uncertainties/limitations which may affect the reliability of our findings.GCM-related gaps such as internal variability, model parameterization, emission pathways, as well as land-use projections modeling, and others could affect the trustworthiness of our estimated exposure.Besides, the global GCMs are recognized as "too hot" and anticipate climate warming in response to CO 2 forcing that might be greater than existing evidence (Eyring et al., 2020).This study does not consider such "hot model" category defined by Hausfather et al. (2022) that may affect our findings' reliability.However, we used the biascorrected MME mean to reduce uncertainty, which performs quite well against observation (Figures S1-S4 in Supporting Information S1).Further, types of drought indices, time scales, and methods of PET calculation may affect the trustworthiness of these findings.As the total area for a grid cell mostly depends on the resolution, the estimated cropland area in this study considering 0.5°× 0.5°may be different from South Asia's real cropland area.
slope of the saturation vapor pressure curve (∆) ; daily air temperature (T ); saturation vapor pressure (e a ); net radiation at the surface (R n ) ; actual atmospheric water vapor pressure (e d ) ; ground heat flux (G); psychometric constant (γ) ; and 2m above wind speed (u 2 ) .

Figure 1 .
Figure 1.(a) Recorded agricultural land area during the 1995-2015 period in South Asia; (b) Spatial pattern of agriculture land area fraction, 2000; (c) Spatial pattern of agriculture land area fraction, 2015.In panel (a), a linear trend is indicated by drawing the black dotted line.

Figure 2 .
Figure 2. Anticipated changes in panel (a) annual mean precipitation as well as it's (b-m) spatial structure under the nominated scenarios for three defined time epochs (2021-2040, 2041-2060, and 2081-2100) and a historical period (1995-2014).Spatial maps are created based on relative change with regard to historical periods.Here, the vertical black line over the (a) bar graph indicates global climate models ranges and the stippled (black dots) in the spatial maps (b-m) indicate the area where more than 50% of models (4-7 models) demonstrate homogeneous increase/decrease pattern.

Figure 3 .
Figure 3. Anticipated changes in panel (a) annual mean potential evapotranspiration as well as it's (b-m) spatial structure under the nominated scenarios for three defined time epochs (2021-2040, 2041-2060, and 2081-2100) and a historical period (1995-2014).Spatial maps are created considering relative change with regard to historical periods.Here, the vertical black line over the (a) bar graph indicates global climate models ranges and the stippled (black dots) in the spatial maps (b-m) indicate the area where more than 50% of models (4-7 models) demonstrate homogeneous increase/decrease pattern.

Figure 4 .
Figure 4. Temporal changes in model-simulated South Asian cropland area under the nominated scenarios for the historical future periods.The vertical dash lines indicate the estimated cropland area for every 20 years of the studied periods (2021-2040, 2041-2060, and 2081-2100).

Figure 7
Figure7signifies the annual DF and percentage of the affected area for1995-2014, 2021-2040, 2041-2060, and  2081-2100.Total drought events in the historical period occurred around 5.3 times per year.Relatively, DF in the near-term (2021-2041) period is anticipated to rise strongly under the nominated scenarios, such as ∼30% (to 6.84 times/year) under SSP4-6.0 and the lowest 19.3% (to 6.2 times/year) under the SSP5-8.5.Further, DF during 2041-2060 tends to escalate from a lower forcing scenario to a higher one.In this epoch, the DF upsurge can be 22.3% higher (6.4 times/year) under SSP3-7.0.Where the least increase is projected under the sustainability pathway scenario combination (SSP1-1.9)by 7.5% (5.7 times/year).Dramatically, DF will drop in the long-term period (2081-2100).It is surprising that the DF is largely reduced for higher emissions such as 14.9% (4.4 times/ year) under the SSP5-8.5 and followed by 2.0% (SSP3-7.0)and 0.9% (SSP4-6.0).However, in this epoch, DF can increase by 17.4% (6.1 times/year) for SSP1-2.6.

Figure 8
Figure 8 explores the spatial outline of DF across South Asia during the historical period, and relative changes for the future epochs.Droughts occurred more frequently (≥6 times annually) in the eastern (eastern ghat of India, Orissa, and West Bengal), and western (a portion of Pakistan and Afghanistan) part of Asia during 1995-2014 (Figure 8a).Drought events occurred five times per year in central South Asia (mostly northwestern and western India).With regard to the historical period, areas with a high frequency for the nominated future time slices and scenarios are geographically concentrated (Figures 8b-8m and Figures S11n-S11v in Supporting Information S1).Drought frequency during 2021-2040 is dominant in the central and western parts of South Asia: northwestern and western India, Pakistan, and Afghanistan.Notably, DF in the western region (Pakistanand Afghanistan) is pronounced mostly under lower emission scenarios SSP1-1.9(FigureS8nin Supporting Information S1), SSP1-2.6 (Figure8b), and SSP4-3.4(FigureS11qin Supporting Information S1), whereas more prominent pattern (with ≥80% relative increase) is exhibited in north-western and western India under higher emission scenarios (mostly SSP4-6.0 and SSP3-7.0).Besides, DF under SSP4-6.0(FigureS11tin Supporting Information S1) is extended toward the IGP region.In 2041-2060, DF will escalate mostly in central South Asia and IGP region for all the scenarios.In this epoch, the increase of 80% is more prominent under the combination of the SSP5-8.5 (Figure8l).For 2081-2100, increased DF shifted toward the western part of South Asia, while it decreases largely in the eastern and central parts.Such a decrease over a large area is prominent under a higher emission scenario.The greatest decrease (< 20%) in DF is found under SSP5-8.5.Boldly, DF in the future would be overriding mostly across the central part during the 2021-2040 and 2041-2060 periods, whereas the

Figure 6 .
Figure 6.Projected changes in dryness and wetness identified by the Standardized Precipitation Evapotranspiration Index (SPEI) drought index over South for (a-l) under the nominated scenarios and three defined periods (2021-2040, 2041-2060, and 2081-2100) and a historical period (1995-2014).The negative value of SPEI implies dryness (red-colored area) and the positive value indicates wetness (green-colored area).Change is estimated compared to the historical period.The stipples (black dots) indicate significant changes (based on two-sample t-test) at the 95% confidence level based on the p-value.

Figure 7 .
Figure 7. Annual drought frequency (DF) and affected area (drought area (DA)) estimated over South Asia during the selected periods and scenarios.The bars denote DF based on multi-model ensemble mean (see left y-axis), and red color-filled circles with dot line represent DA coverage based on multi-model ensemble mean (see right y-axis).The straight black line (top on bars) indicates global climate models spreads.

Figure 9 .
Figure 9. Cropland area exposed to drought estimated over South Asia during the selected periods and scenarios.The blue color-filled circles characterize the exposed area and the deep yellowish straight bar indicates the global climate models ranges.
interventions through upgraded technology designs are pivotal for securing the global food supply under a changing climate.

Table 1
Nominated Climate Model's Outline