Implications of future climate change on crop and irrigation water requirements in a semi-arid river basin using CMIP6 GCMs

Kunal KARAN, Dharmaveer SINGH, Pushpendra K SINGH, Birendra BHARATI,Tarun P SINGH, Ronny BERNDTSSON

1 Department of Water Engineering and Management, Central University of Jharkhand, Brambe, Ranchi 835205, India;

2 Symbiosis Institute of Geo-informatics, Symbiosis International (Deemed University), Pune 411016, India;

3 Water Resources Systems Division, National Institute of Hydrology, Roorkee 247667, India;

4 Division of Water Resources Engineering & Centre for Advanced Middle Eastern Studies, Lund University, Lund Box 117, 22100, Sweden

Abstract: Agriculture faces risks due to increasing stress from climate change, particularly in semi-arid regions. Lack of understanding of crop water requirement (CWR) and irrigation water requirement (IWR)in a changing climate may result in crop failure and socioeconomic problems that can become detrimental to agriculture-based economies in emerging nations worldwide. Previous research in CWR and IWR has largely focused on large river basins and scenarios from the Coupled Model Intercomparison Project Phase 3 (CMIP3) and Coupled Model Intercomparison Project Phase 5 (CMIP5) to account for the impacts of climate change on crops. Smaller basins, however, are more susceptible to regional climate change, with more significant impacts on crops. This study estimates CWRs and IWRs for five crops(sugarcane, wheat, cotton, sorghum, and soybean) in the Pravara River Basin (area of 6537 km2) of India using outputs from the most recent Coupled Model Intercomparison Project Phase 6 (CMIP6) General Circulation Models (GCMs) under Shared Socio-economic Pathway (SSP)245 and SSP585 scenarios. An increase in mean annual rainfall is projected under both scenarios in the 2050s and 2080s using ten selected CMIP6 GCMs. CWRs for all crops may decline in almost all of the CMIP6 GCMs in the 2050s and 2080s (with the exceptions of ACCESS-CM-2 and ACCESS-ESM-1.5) under SSP245 and SSP585 scenarios. The availability of increasing soil moisture in the root zone due to increasing rainfall and a decrease in the projected maximum temperature may be responsible for this decline in CWR. Similarly,except for soybean and cotton, the projected IWRs for all other three crops under SSP245 and SSP585 scenarios show a decrease or a small increase in the 2050s and 2080s in most CMIP6 GCMs. These findings are important for agricultural researchers and water resource managers to implement long-term crop planning techniques and to reduce the negative impacts of climate change and associated rainfall variability to avert crop failure and agricultural losses.

Keywords: climate change; crop water requirement; irrigation water requirement; CMIP6 GCMs; emission scenario;Pravara River Basin

1 Introduction

India's agricultural sector contributes 17.90% to the gross value added and provides employment to 56.40% of the total workforce of the country (Economic Survey, 2021). The Indian agriculture is heavily climate dependent with notable regional and climatic variability, and rain-fed area accounts for approximately 54.00% of the gross cropped area, despite notable efforts to increase the area under irrigation (Gulati et al., 2018). The agricultural sector is facing multi-dimensional pressure from many stressors, such as climate change and socio-economic factors (e.g.,population growth, unbounded urbanization, and industrialization) (Dai et al., 2013; Yu et al.,2019). Additionally, potential implications of climate change vary from a regional to local population scale (Farooq et al., 2022). Changes in climate primarily in the forms of rainfall,temperature, and radiation will impact the availability of water resources and water requirements for both irrigated and rain-fed crops (Jain and Singh, 2020). Masia et al. (2021) found that evaporative demand will increase in tandem with rising global temperature, thus increasing crop evapotranspiration. Crop water requirement (CWR) varies depending on the cropping system, and climate change may have a substantial impact on CWR (Sun et al., 2013; Ye et al., 2015). Elgali et al. (2007) found that CWR and irrigation water requirement (IWR) are very sensitive to changes in climatic variables and will vary due to climate change. Thus, climate change is increasingly threatening major crop production systems (Farooq et al., 2022). Ahmad et al. (2021)revealed that the risk of crop failure will be higher due to projected climate change under Coupled Model Intercomparison Project Phase 5 (CMIP5) General Circulation Models (GCMs). However,the actual impacts of climate change on CWR are complex and difficult to predict (Jain and Singh, 2020).

According to the Fifth Assessment Report (AR5) of IPCC (2014), the surface temperature will increase during the 21stcentury under all emission scenarios and this will adversely impact the agriculture, land and water resources, environment, ecosystems, biodiversity, and even the society(Singh et al., 2015a; Jain and Singh, 2020). Therefore, it is vital to explore the links between agricultural water management and climate change looking into the regional and local variability of food and water security. D?ll (2002) applied global irrigation model to assess the impacts of climate change on net IWR and found that there may be an increase in the global net IWR of 5.00%-8.00% by the end of the 2070s and it might be the largest in the South Asia region. Wada et al. (2013) used CMIP5 GCMs and applied seven global hydrological models to evaluate the impacts of climate change on CWR and pointed out that there would be an increase in CWR and a decrease in water availability by the 2080s with pronounced regional patterns. Rehana and Mujumdar (2013) found that there will probably be an increase in CWR due to climate change.Elliot et al. (2014) revealed that there might be an inversion of 2.0×104-6.0×104km2cropland from irrigated system to rain-fed system due to the limitations of freshwater availability mostly in the irrigated regions of western United States, China, and West, South, and Central Asia.

According to Jain and Singh (2020), the correlation between global CWR and global warming is high and the benefits of increasing rainfall for irrigation is small. Haz-Amor et al. (2020) used CMIP5 GCMs with CROPWAT model and found an increase in IWR based on the scenario projections. Shrestha et al. (2013) also used coupled CROPWAT model with HadCM3 GCM (A2 and B2 scenarios) for future projections and found that IWR varied with physiographic regions and growth stages of crops. The IWRs in the middle and high hills of Nepal were found to have a decreasing trend, while IWR in the Terai region showed an increasing trend. De Silva et al. (2007)predicted the impacts of climate change by coupling CROPWAT model with HadCM3 GCM under A2 and B2 scenarios and found that IWR increased by 23.00% and 13.00% for paddy crop,respectively. Das et al. (2020) assessed the impacts of climate change on crop yield in the eastern Himalayas. More recently, Abdoulaye et al. (2021) and Poonia et al. (2021) assessed the impacts of climate change on CWR and IWR using CROPWAT model coupled with CMIP5 and Coordinated Regional Downscaling Experiment (CORDEX) GCMs, respectively, for river basins in Niger and eastern Himalayan region. Some other important work in this domain includes the study of Li et al. (2020), who developed a structure for ''water suitable'' agriculture by analyzing factors affecting IWR. Overall, it can be stated that the climate change-induced increase in water demand will bring further challenges to farmers to irrigate and grow crops with limited water availability. However, the studies done by Fl?rke et al. (2018) and Gondim et al. (2018) opined that the improvements in agricultural water use efficiency through improved technological and scientific interventions will help to compensate the adverse impacts of climate change and may supply enough water to meet water demands of other sectors. For improving water use efficiency,it is critical to understand how much water crops require at different times of the year, as well as to develop rational irrigation schedule for irrigated areas. Tubiello and Fischer (2007) found that an alleviated climate may reduce the impacts of climate change on agricultural water requirements by about 40.00%, or 125×108-160×108m3, compared with an unmitigated climate.Schwaller et al. (2021) underscore the importance of effective agricultural water management through a comprehensive understanding of CWR and IWR, but this information is frequently not readily available. Therefore, it is of paramount importance to understand the variability of the climate-induced CWRs and IWRs of different crops for effectively managing the agricultural water resources and mitigate the adverse impacts of climate change on agriculture.

However, previous studies, as discussed above, have used climate projections mainly from Coupled Model Intercomparison Project Phase 3 (CMIP3) GCMs and CMIP5 GCMs to assess the future CWR and IWR, which inherited limitations, particularly in simulating the extreme events of rainfall (Kim et al., 2020). This bias in the estimation of rainfall results in higher uncertainties in projected CWR and IWR. More recently, a new generation of Coupled Model Intercomparison Project Phase 6 (CMIP6) framework has been introduced in recent years (Eyring et al., 2016;Gupta et al., 2020; Mishra et al., 2020) to overcome the drawbacks of CMIP3 and CMIP5 models and fulfil the need of a growing climate community. Therefore, in this study, downscaled and bias-corrected GCM outputs of different climatic parameters generated within new Shared Socio-economic Pathways (SSPs) developed as a part of CMIP6 GCM framework were applied for the estimation of CWR and IWR using CROPWAT model for five crops (i.e., sugarcane,cotton, soybean, wheat, and sorghum) of varying growth periods in a semi-arid river basin located in the state of Maharashtra in India for future periods (2050s and 2080s). This study will aid in the understanding of the long-term impacts of climate change on agriculture and the development of adaptation plans for local agricultural water management by local water managers, researchers,and policymakers.

2 Materials and methods

2.1 Study area

The selected study area is the Pravara River Basin (PRB; Fig. 1), located in the Ahmednagar District of Maharashtra State, India. Ahmednagar is one of the worst drought-hit districts in India.It is home to 4.5×106people, of which about 80.00% are rural. Agriculture and animal husbandry are the prime economic activities and contribute significantly to the total income of the region.Pravara River is one of the smallest tributaries of the Godavari River (the second largest river system in Peninsular India) that originates near Akola on the eastern slopes of the Sahyadris(19°31′45′′N, 73°45′05′′E; 750 m) in the western Ghats. It is a rain-fed and intermittent river that generally dries out in summer. Mahalungi and Mula are two important left and right bank tributaries of the Pravara River that join at Sangamner and Nevasa, respectively.

The river flows approximately 208 km from its origin Sahayadri to its mouth Pravara Sangam(19°37′00′′N, 75°01′00′′E; 531 m) and forms a basin with an area of approximately 6537 km2,which is confined between latitude of 19°02′16′′-19°41′40′′N and longitude of 73°6'1′′-75°01′33′′E. The entire basin is made up of Cretaceous-Tertiary extrusive basalt flows known as the Deccan Volcanic Province (Wellman and McElhinny, 1970; Alexander, 1981; Widdowson and Mitchel, 1999; Hooper et al., 2010). The major soil category includes shallow alluvium soil,medium black soil, deep black soil, and reddish soil, which account for nearly 38.00%, 48.00%,13.00%, and 0.80% of the total cultivated area, respectively (http://www.kvk.pravara.com/).Topographically, a moderate relief variation is found in the basin, where altitude ranges from 404 to 1424 m. The western section of the basin has a hilly landscape, whereas the eastern section is a relatively flat plateau. Because of its geographical location, the PRB has a semi-arid climate. The basin's average annual rainfall based on the gridded rainfall data (0.25°×0.25°) of the India Meteorological Department for 30 years (1991-2020) is 593.3 mm. The seasonal monsoonal rains, occurring between June and September, contribute to the majority (about 80.00%) of the total annual rainfall. The annual mean of the maximum temperature (Tmax) and the minimum temperature (Tmin) is 34.70°C and 18.90°C, respectively. During summer months (April-June),the maximum daily temperature soars to as high as 49.00°C in the basin. The basin has been brought under intensive agriculture because of the construction of the Bhandardara Dam (height of 507 m, width of 82 m, and capacity of 3.11×108m3) upstream of the river in 1926. It is one of the major irrigation projects in the Ahmednagar District, accounting for nearly 8.70% (570 km2)of the total basin area. Sugarcane, cotton, soybean, wheat, and sorghum are important crops that are grown in the region (Table 1). Sugarcane has become the dominant commercial crop in this area, and the Pravara River serves as the primary irrigation source for agriculture. The region is vulnerable to climate change and has experienced persistent multi-year droughts in the recent past. Therefore, it is of paramount importance to understand the variability of the climate-induced CWRs and IWRs of different crops for effectively managing the agricultural water and mitigating the adverse impacts of climate change on agriculture in the PRB.

Fig. 1 Overview of the Pravara River Basin (PRB) as well as the important places (cities, towns, and dams) in the basin

Table 1 Planting and harvesting date for the major crops grown in the Pravara River Basin (PRB)

2.2 Agro-meteorological data

Figure 2 shows a schematic on the methodology and data adopted in this study. Data used for the calculations of CWR and IWR including climatic variables, soil parameters, and crop parameters were input in CROPWAT 8.0 model. Meteorological data including temperature (TmaxandTmin),rainfall, wind speed, mean relative humidity, and sunshine hours were obtained on daily time scale for 30 years (1991-2020) from the India Meteorological Department(http://www.imdpune.gov.in/). Rainfall data were available at a grid resolution of 0.250°×0.250°,while other climatic variables were offered at a spatial resolution of 0.500°×0.500°. Soil data(total available moisture content, maximum rain infiltration rate, maximum rooting depth, initial soil moisture depletion, and initial available soil moisture) and crop parameters (planting date,length of crop growth stage, crop coefficient, rooting depth, critical depletion factor, yield response factor, and crop height) were obtained from the Food and Agriculture Organization(FAO) Manual 56 available at http://www.fao.org/land-water/database.

Fig. 2 Flowchart showing the adopted methodology for estimating and investigating the implications of future climate change on CWRs and IWRs for major crops in the PRB. CWR, crop water requirement; IWR, irrigation water requirements; Kc, crop coefficient; CMIP6, Coupled Model Intercomparison Project Phase 6; GCMs,General Circulation Models; SSP, Shared Socio-economic Pathway; MK, Mann-Kendall; ET0, reference crop evapotranspiration; FAO, Food and Agriculture Organization; ETc, crop evapotranspiration; Reff, effective rainfall.

2.3 Scenario data of CMIP6 GCMs

In this study, we used the bias corrected downscaled CMIP6 GCM datasets developed by Mishra et al. (2020) for South Asia to investigate the impacts of future climate change on CWR and IWR.They developed these bias corrected datasets based on Empirical Quantile Mapping approach for the historical (1951-2014) and future (2015-2100) periods under four scenarios: Shared Socio-economic Pathway (SSP)126, SSP245, SSP370, and SSP585. In this study, downscaled high spatial resolution (approximately 0.250°×0.250°) data of ten CMIP6 GCMs (Table 2) were used to reveal the middle (2041-2070) and far (2071-2100) future trends in climatic variables and their implications on CWR and IWR under two scenarios (SSP245 and SSP585) in the PRB. SSPs were used to explain the possible future greenhouse gas emissions under different global socio-economic changes that will take place by 2100 (Riahi et al., 2017). They present socio-economic and technological trajectories with a baseline in which no climate policies are enacted after 2010, resulting in 3.00°C-5.00°C of warming above pre-industrial levels by 2100.In addition, the four SSPs can be linked to climate policies to generate different outcomes at the end of the century (analogous to representative concentration pathways (RCPs)), with radiative forcing of 2.6, 3.4, 4.5, 6.0, 7.0, and 8.5 W/m2in 2100.

Table 2 Detailed description of Coupled Model Intercomparison Project Phase 6 (GMIP6) General Circulation Models (GCMs) used in this study

2.4 Screening of data and trend analysis

The inhomogeneity in time series introduces statistical errors that may lead to false analysis and interpretation of climatic events (Peterson et al., 1998; WMO, 2011). Therefore, initial data quality check (e.g., screening data for outliers, trends, and discontinuities) is recommended on time series (monthly, seasonal, and annual) of climatic variables (WMO, 2011). Outliers and inhomogeneity in the historical data were detected using generalized Extreme Studentized Deviate (ESD) test (Rosner, 1983) and Buishand's Range test (Buishand, 1982) methods,respectively. ESD method was used to test the null hypothesis (Ho) that there are no outliers versus the alternative hypothesis (H1) that there are 'r' outliers in the dataset. The test statistic 'r'was defined by the following equation:

It is a measure of cumulative deviation from the mean for thekthobservation in time series (x1,x2, …,xi, …,xn). The series is homogenous without any change point ifSkfluctuates around zero.When a break point is present in the series,Skreaches a maximum value (negative shift) or minimum value (positive shift) for the yeari=k. The significance of shift was tested using rescaled adjusted range (R) defined by the following equation:

The ratio ofRtosquare root (n) was compared with the tabulated critical values (Buishand,1982) at given significance level (Table 3). It should be noted that 'n' represents the total number of observations in the time series. The null hypothesis of no change point is rejected if the ratio is less than the critical value (1.43) at 5% significance level. These time series of observations were,further, homogenised using a multistep process based on non-parametric statistics as described in Peterson and Easterling (1994).

Table 3 Statistical information of annual and seasonal climatic variables averaged over 1991-2020 in the PRB

After the screening of data, the non-parametric (Mann-Kendall (MK)) test and Sen's slope estimator methods were employed for detecting trends in climatic variables. A detailed description of the methods can be found in Singh et al. (2015b).

2.5 Estimations of reference crop evapotranspiration (ET0), effective rainfall (Reff), CWR,and IWR

CROPWAT 8.0 model was used to estimateET0,Reff, and water requirements (CWR and IWR) for the different investigated crops. CROPWAT 8.0 model calculatesET0,Reff, CWR, and IWR based on climate, crop, and soil data. It has been widely used as a decision-support tool in international settings to calculate regional irrigation needs (Smith et al., 2002; Poonia et al., 2021). The model uses Penman-Monteith (FAO 56 PM) (Savva et al., 2002; Schwallar et al., 2021) equation to calculateET0. This equation usesTmax,Tmin, humidity, wind speed, and sunshine hours as input data, and CROPWAT model uses these data to calculateET0by the following expression (Allen et al., 1998):

where,ET0is the reference crop evapotranspiration (mm/d);Rnis the mean daily net radiation(MJ/(m2·d));Gis the soil heat flux density (MJ/(m2·d));γis the psychometric constant (0.067 kPa/°C);Tis the mean daily air temperature (°C), and the value of (Tmax+Tmin)/2 was measured between 1.5 and 2.0 m height above the ground;U2is the wind speed at 2.0 m height above the ground (m/s);esis the saturation vapour pressure (kPa);eais the actual vapour pressure (kPa);(es-ea) is the vapour pressure deficit (kPa); and Δ is the slope of vapour pressure curve (kPa/°C).

Similarly,Reffwas estimated using the following relationship for the two conditions (whenRmonth＜250.00 mm andRmonth＞250.00 mm, whereRmonthis the monthly average rainfall (mm)) as described in Moseki et al. (2019):

The CWR was calculated using FAO 56 PM equation coupled with the single crop coefficient method as follows (Allen et al., 1998; Luo et al., 2022):

where,ETcandKcare the crop evapotranspiration (crop water requirement; mm/d) and crop coefficient, respectively. TheKcwas determined from the variation in climatic variables, crop types, and growing stages of crops. Notably, in this work, CWR was computed during the whole growth period for all five crops, i.e., sugarcane, wheat, cotton, soybean, and sorghum. Further,IWR was estimated by subtractingRefffromETc(Moseki et al., 2019; Poonia et al., 2021):

3 Results and discussion

3.1 Trends in climatic variables for historical period (1991–2020)

Temporal variability in temperature (TmaxandTmin) and rainfall directly affects the sowing and growing stages of crops. The best way to explain this variability is to look at anomalies inTmax,Tmin, and rainfall and analyze their patterns during the study period (Singh et al., 2016). In this study, anomalies in temperature and rainfall were calculated by subtracting the annual and seasonal time series from the yearly mean averaged for the years 1991-2020 (historical period).Using MK and Sen's slope estimator methods, we determined the magnitudes and directions of changes inTmax,Tmin, and rainfall. The results of MK test (Zs) and Sen's slope test (Q) are given in Table 4. At the annual scale, statistically insignificant decreasing trends inTmaxand rainfall are observed in the PRB for the study period. However, no definite trend in annualTminis detected.Similar patterns are observed for these variables in the post-monsoon but with varying magnitudes. During monsoon, statistically insignificant increasing trends inTmaxandTminare detected, while rainfall reveals a decreasing trend. However, during winter, no trend in rainfall but a decreasing trend in temperature are observed.

Table 4 Annual and seasonal trends in temperature and rainfall anomalies in the PRB for historical period(1991-2020)

These results show dissimilarity with previous studies of Guhathakurta et al. (2013) and Singh et al. (2021) where a statistically insignificant decreasing trend of rainfall in winter and an increasing trend of rainfall in pre-monsoon, monsoon, and post-monsoon and annual rainfall are reported in the Ahmednagar District. However, a recent report ''Observed Rainfall Variability and Changes over Maharashtra State'' published in 2020 by the India Meteorological Department shows a decreasing trend in mean annual and monsoonal rainfall in the Ahmednagar District(Guhathakurta et al., 2020). The difference in these results might be attributed to the length of the data period used for the trend analysis. Specifically, Guhathakurta et al. (2013) and Singh et al.(2021) used long-term gridded data obtained from the India Meteorological Department for 1901-2006 and 1901-2018, respectively, while the data period in the report was 1989-2018.Furthermore, the decreasing trend observed in rainfall can be attributed to the decreased number of rainy days recorded in the Ahmednagar District in the recent decades (1990-2020s).

3.2 Changes in climatic variables under SSP245 and SSP585 scenarios in the 2050s and 2080s

Figures 3 and 4 show projected changes in mean annualTmax,Tmin, and rainfall under SSP245 and SSP585 scenarios for future periods 2041-2070 (2050s) and 2071-2100 (2080s) with respect to the baseline period 1991-2020. All models under both scenarios predict rises in mean annualTminand rainfall in the 2050s and 2080s. The increasing ranges are 0.59°C-3.47°C in the 2050s and 1.21°C-6.47°C in the 2080s forTmin, and 11.70%-73.05% in the 2050s and 2.80%-163.80% in the 2080s for rainfall. Within scenarios, relatively high increases in mean annualTminand rainfall are observed under SSP585 scenario compared to SSP245 scenario. However, in general, eight out of ten models predict a decrease in mean annualTmaxunder all scenarios in all future periods except for a higher emission scenario (i.e., SSP585) in the 2080s.

Projected changes in seasonal rainfall under both SSP245 and SSP585 scenarios are investigated. In general, nine out of ten CMIP6 GCMs predict a significant increase in monsoonal rainfall under both scenarios in the 2050s and 2080s. Specifically, the increasing ranges are 21.00%-73.00% in the 2050s and 4.00%-99.00% in the 2080s under SSP245 scenario, and 17.00%-94.00% in the 2050s and 17.00%-176.00% in the 2080s under SSP585 scenario.However, a decrease in post-monsoonal rainfall ranging from -0.02% to -73.00% is projected under SSP245 scenario in the 2050s. Similarly, the analysis of future scenarios in general reveals a decrease inTmaxfor all seasons under a lower emission scenario of SSP245 in the 2050s;however, an increase inTmaxis predicted under a higher emission scenario of SSP585 for all seasons in the 2080s. Opposite to this, the increase inTminis predicted for all seasons under both scenarios for most of the models in the 2050s and 2080s. The predicted increase is relatively high forTminas compared toTmaxunder both scenarios for all seasons. The results discussed here are in accordance to previous work of Todmal et al. (2021) who reported increase in mean annual and monsoonal rainfall and minimum temperature over the state of Maharashtra in India for future period (2015-2100) using REgional MOdel (REMO).

Fig. 3 Projected changes in annual Tmin (a-d) and Tmax (e-h) in the PRB using different CMIP6 GCMs under SSP245 and SSP585 scenarios in the 2050s and 2080s. Tmin, minimum temperature; Tmax, maximum temperature.

3.3 Estimations of CWR and IWR for historical period (1991–2020)

3.3.1ET0andReff

Table 5 presentsET0andReffestimated for different months using data of 1991-2020.ET0exhibits variations in different months. Due to high temperature in summer, it has the maximum value in April (7.92 mm/d); however, the minimum value occurs in August (4.12 mm/d) as temperature is likewise low in this month. From 1991 to 2020, the long-term yearly average ofET0is 5.27 mm/d. The volatility inET0is attributed to changes in other climatic variables. Low humidity, high wind speed, and high temperature causeET0to be at its highest level during the summer season (i.e., dry season).Reffexperiences similar variations throughout the year owing to variations in rainfall, with September having the highest value (113.34 mm/month) and January having the lowest value (1.02 mm/month). The long-term (1991-2020) yearly average ofReffis 492.21 mm/month.

Fig. 4 Projected changes in mean annual rainfall in the PRB using different CMIP6 GCMs under SSP245 and SSP585 scenarios in the 2050s (a and b) and 2080s (c and d).

Table 5 Estimations of monthly reference crop evapotranspiration (ET0) and effective rainfall (Reff) in the PRB from meteorological data averaged over 1991-2020

3.3.2 Estimations of CWRs and IWRs for different crops

CWRof any crop is the total amount (depth) of water lost owing to evapotranspiration and is determined fromETc. Every crop has various water requirements depending on the location,climatic conditions, soil type, cultivation technique, andReff(Ewaid et al., 2019). Tables 6-10 show thatReff, CWR, and IWR fluctuate across the developmental stages of all crops. Moreover,Kcvalues are not constant at any development stage that indicates seasonal crop water needs(Allen et al., 1998; Azevedo et al., 2007; Irmark et al., 2013).ETcincreases during the growth stage and lowers significantly during the later phases based onKcvalues. Tables 6-10 indicate thatETcvalues are lower in the early and late stages when crops are in the productive stage and higher in the middle stage. IWRs for all five crops in 10 d are descended in the order of 1707.00 mm/10 d (sugarcane)＞500.80 mm/10 d (wheat)＞401.60 mm/10 d (cotton)＞381.50 mm/10 d(sorghum)＞212.80 mm/10 d (soybean), as shown in Tables 6-10.

Table 6 Estimations of CWR and IWR for wheat in the PRB for historical period (1991-2020)

Table 7 Estimations of CWR and IWR for sorghum in the PRB for historical period (1991-2020)

Table 8 Estimations of CWR and IWR for soybean in the PRB for historical period (1991-2020)

Table 9 Estimations of CWR and IWR for sugarcane in the PRB for historical period (1991-2020)

Continued

Table 10 Estimations of CWR and IWRs for cotton in the PRB for historical period (1991-2020)

3.4 Estimations of CWR and IWR for crops under future scenarios

3.4.1ET0andReff

Using CROPWAT 8.0 model, we calculatedET0for different months in the 2050s and 2080s under SSP245 and SSP585 scenarios, as shown in Tables 11 and 12. In the 2050s and 2080s,ET0is highest in April and lowest in August as compared to the other months for both the scenarios.The volatility inET0is attributed to changes in climatic variables. Low humidity, high wind speed, and high temperature causeET0to be at its highest level during the summer season (i.e.,dry season). Here, it can be deduced that ACCESS-CM-2 GCM consistently projects the highest averageET0as compared to the rest of the CMIP6 GCMs under both scenarios, i.e., SSP245 and SSP585, in the 2050s and 2080s.

Table 11 Projected estimations of monthly ET0 using different CMIP6 GCMs in the PRB under SSP245 and SSP585 scenarios in the 2050s

Table 12 Projected estimations of monthly ET0 using different CMIP6 GCMs in the PRB under SSP245 and SSP585 scenarios in the 2080s

Further, Figure 5 shows the percentage change inET0under SSP245 and SSP585 scenarios in the 2050s and 2080s. For all of the CMIP6 GCMs, the percentage change inET0is negative (i.e.,ET0is decreasing) under SSP245 scenario in the 2050s and 2080s, except for the three CMIP6 GCMs, i.e., ACCESS-CM-2, MRI-ESM2-0, and ACCESS-ESM-1.5, in which the percentage change inET0is positive, i.e., 2.70% and 4.60%, respectively, in the 2050s and 2080s with respect to averageET0inthe baseline period (1991-2020). Specifically, in the 2080s, only two CMIP6 GCMs (i.e., ACCESS-CM-2 and ACCESS-ESM-1.5) predict a positive increase inET0under SSP245 scenario. Mondal et al. (2021) reported an increase of 4.00% inET0over the Indus River Basin under the medium emission scenario (SSP245). A similar pattern is observed under SSP585 scenario in the 2050s in this study, where for almost all of the CMIP6-GCMs, the percentage change inET0is negative (i.e.,ET0is decreasing), except for the two CMIP6 GCMs,i.e., ACCESS-CM-2 and ACCESS-ESM-1.5, in which the percentage change inET0is positive,i.e., 3.80% and 1.70%, respectively, with respect to averageET0inthe baseline period(1991-2020). However, under SSP585 scenario in the 2080s, all of CMIP6 GCMs reveal consistently a positive change inET0(i.e.,ET0is increasing), up to 20.30%, which is much higher than the percentage increase inET0(4.60%) under SSP245 scenario in the 2080s. Overall,ET0is predicted to increase for all ten CMIP6 GCMs under SSP585 scenario (high emission scenario) in the 2080s, while both increasing and declining trends are observed with decreasing pronounced pattern for the remaining periods and scenarios.

Fig. 5 Projected percentage change in ET0 using different CMIP6 GCMs in n the PRB under SSP245 (a and b)and SSP585 (c and d) scenarios in the 2050s and 2080s

3.4.2Estimations of CWRs and IWRsfor different crops

Table 13 shows CWR estimates for different crops using ten CMIP6 GCMs under two scenarios,i.e., SSP245 and SSP585, in the 2050s and 2080s. It can be observed from Table 13 that two CMIP6 GCMs, namely ACCESS-CM-2 and ACCESS-ESM-1.5, project increases in CWRs for cotton and soybean crops under both SSP245 and SSP585 scenarios as compared to the rest of the eight CMIP6 GCMs, which estimate decreases in CWRs for all crops in the 2050s and 2080s.However, the overall averages of projected CWRs for cotton, sorghum, sugarcane, soybean, and wheat for all ten models are 667.80, 417.20, 2102.40, 501.20, and 542.80 mm in the 2050s, and 671.70, 420.50, 2112.40, 500.90, and 547.50 mm in the 2080s, respectively. These values are comparatively lower than those during historical period (1991-2020). Figure 6 also shows the projected percentage change (percentage increase or decrease) of CWRs for all crops using the ten CMIP6 GCMs under SSP245 and SSP585 scenarios in the 2050s and 2080s, with respect to the baseline period 1991-2020. With the exceptions of ACCESS-CM-2 and ACCESS-ESM-1.5,which predict increases of CWRs in soybean, cotton, and wheat under SSP245 scenario in the 2050s and 2080s, decreases in the estimated percentage change of CWRs for all crops are found for nearly all of the CMIP6 GCMs. Similarly, under SSP585 scenario, there are decreases in the projected percentage change of CWRs for almost all of the CMIP6 GCMs in the 2050s, except for ACCESS-CM-2 and ACCESS-ESM-1.5. However, a detailed analysis shows that there exist increases in the projected percentage change of CWRs for all crops under SSP585 scenario in the 2080s as compared to the SSP245 scenario in the 2050s. This could be related to the rise inTmaxduring this period. Additionally, results of ACCESS-CM-2 and ACCESS-ESM-1.5 indicate that there are larger increases in the projected percentage change of CWRs for soybean, cotton, and wheat than for sugarcane and sorghum under both scenarios in the 2050s and 2080s. Overall, the results show that the projected percentage change ofCWR is decreasing for almost all of CMIP6 GCMs (except for ACCESS-CM-2 and ACCESS-ESM-1.5) under SSP245 scenario in the 2050s and 2080s. However, under SSP585 scenario, the projected percentage change of CWRs for all crops may increase for five GCMs (ACCESS-ESM-1.5, ACCESS-CM-2, BCC-CSM2-MR,INMCM4-8, and NorESM2-MM) in the 2080s. As an example, by using MRI-ESM2-0, the projected percentage change of CWR for sugarcane is -2.30%, whereas the values are -4.60%,-3.90%, -3.60%, and -2.30%, respectively, for sorghum, wheat, cotton, and soybean.

Table 13 Projected CWRs of five crops using different CMIP6 GCMs in the PRB under SSP245 and SSP585 scenarios in the 2050s and 2080s

Fig. 6 Projected percentage change in CWRs of five crops using different CMIP6 GCMs in the PRB under SSP245 (a and b) and SSP585 (c and d) scenarios in the 2050s and 2080s

However, the overall averages of the projected percentage decrease of CWRs for all five crops using the ten CMIP6 GCMs reveal that cotton has the largest percentage decrease of CWR(-3.70%), followed by sorghum (-3.60%), sugarcane (-3.30%), soybean (-3.30%), and wheat(-2.9%) under SSP245 scenario in the 2050s and 2080s (Fig. 6), with respect to the baseline period (1991-2020) as given in Tables 6-10. Similarly, under SSP585 scenario, sugarcane shows the largest percentage decrease in projected CWR (-3.10%), followed by cotton (-2.80%),sorghum (-2.80%), soybean (-2.50%), and wheat (-2.00%) (Fig. 6). This indicates that the pattern of the percentage decrease in CWRs for all five crops is similar in the 2050s under both scenarios (i.e., SSP245 and SSP585), except for sugarcane, which shows a larger percentage decrease in CWRs under SSP585 scenario in the 2050s as compared to the other crops. Similarly,in the 2080s, the overall averages of the projected percentage decrease of CWRs for all five crops using the ten CMIP6 GCMs show that soybean exhibits the largest percentage decrease in projected CWR (-3.70%), followed by cotton (-3.30%), sugarcane (-3.00%), sorghum (-2.90%),and wheat (-2.20%) under SSP245 scenario. The availability of increasing soil moisture in the root zone due to increasing rainfall and a decrease in projectedTmaxmay be responsible for this decline in CWR.

Figure 6 indicates that the percentage decreases in CWRfor wheat crop under both SSP245 and SSP585 scenarios in the 2050s and 2080s are lower as compared to the other crops for most of the CMIP6 GCMs, in which CWR is projected to increase. This shows that crops like wheat and sugarcane will have higher CWRs in future than other crops in the PRB. However, at the same time, if we look into the CWRs projected by ACCESS-ESM-1.5 and ACCESS-CM-2 GCMs,CWRs of cotton and soybean will increase under both SSP245 and SSP585 scenarios in the 2050s and 2080s. Acharjee et al. (2017) found that future CWRs are projected to fluctuate and depend on the rainfall pattern. Overall, our results imply that projected CWRs are decreasing for most of the CMIP6 GCMs (except for ACCESS-CM-2 and ACCESS-ESM-1.5) under SSP245 scenario in the 2050s and 2080s. However, under SSP585 scenario, projected CWRs for all crops show increases for half of the CMIP6 GCMs in the 2080s in comparison to the 2050s. This analysis indicates that crops such as wheat and sugarcane have lower decreases in CWRs as compared to other crops in the PRB in comparison to the baseline period (1991-2020).

Table 14 shows the estimations of future IWRs for the five selected crops using ten CMIP6 GCMs under two different scenarios, i.e., SSP245 and SSP585, in the 2050s and 2080s. The overall averages of projected IWRs for cotton, sorghum, sugarcane, soybean, and wheat in the 2050s for all CMIP6 GCMs are 415.00, 357.30, 1730.70, 249.60, and 471.10 mm under SSP245 scenario, and 367.40, 317.80, 1562.90, 213.10, and 430.00 mm under SSP585 scenario,respectively. Similarly, projected IWRs in the 2080s are 405.30, 350.60, 1720.80, 230.80, and 474.40 mm under SSP245 scenario, and 354.30, 318.30, 1575.80, 196.9, and 427.5 mm under SSP585 scenario, for cotton, sorghum, sugarcane, soybean, and wheat crops. In general, there is a decrease in the overall projected IWRs for both scenarios in the 2050s and 2080s with respect to baseline period (1991-2020). We evaluated the percentage deviation between the overall average of projected IWRs for all five crops under SSP245 and SSP585 scenarios with respect to the baseline period (1991-2020). The results show that under SSP245 scenario, sorghum has the largest percentage decrease in projected IWR (-6.30%), followed by wheat (-5.90%) as compared to the other crops, whereas soybean has the largest percentage increase in projected IWR(17.30%), followed by cotton (3.30%) in the 2050s. Under SSP585 scenario, again, sorghum cropcrop has the largest decrease in projected IWR (-16.70%) with respect to the baseline period,followed by wheat (-14.40%), cotton (-8.50%), and sugarcane (-8.40%) in the 2050s.

Table 14 Projected IWRs of five crops using different CMIP6 GCMs in the PRB under SSS245 and SSP585 scenarios in the 2050s and 2080s

Similarly, Figure 7 shows the projected percentage change in IWRs for all crops under SSP245 and SSP585 scenarios in both periods. In general, the projected percentage change of IWRs for all crops under SSP245 and SSP585 scenarios are decreasing or slightly increasing, except for soybean and cotton, which have the largest increase in IWRs (up to 51.30% and 24.50%,respectively) in the 2050s under SSP245 scenario. Sorghum has the largest percentage decrease in IWR (-25.00%) in the 2050s under SSP245 and SSP585 scenarios (Fig. 7). Similar pattern (with slight increase or decrease) is observed for this crop in the 2080s under both scenarios. The main reason for the decrease in projected IWR may be attributed to the increase in average annual rainfall. The projected percentage change of IWR show high variations under SSP245 and SSP585 scenarios in the 2050s and 2080s mainly due to large variability in projected rainfall.Figure 7 further indicates that in general there are increases in the projected percentage change of IWRs in the 2050s and 2080s under both scenarios for soybean and cotton by using five CMIP6 GCMs (ACCESS-ESM-1.5, ACCESS-CM-2, BCC-CSM2-MR, INMCM4-8, and INMCM5.0).Konzmann et al. (2013) observed that increased rainfall resulted in a modest drop in irrigation needs in several regions worldwide, including southeastern China and India. There is also the argument that the structural crop responses caused by elevated CO2concentrations may offset the negative impacts of climate change on agriculture, which would result in lower IWRs in many parts of the world (Konzmann et al., 2013).

Fig. 7 Projected percentage change in IWRs using different CMIP6 GCMs in the PRB under SSP245 (a and b)and SSP585 (c and d) scenarios in the 2050s and 2080s

4 Conclusions

In this study, multi-scenario projections from CMIP6 GCMs were used to investigate the impacts of climate change on CWR and IWR in a semi-arid river basin of India under two future scenarios(SSP245 and SSP585) in the 2050s and 2080s. We found that projected CWRs decrease for all five crops for almost all of the CMIP6 GCMs (except for ACCESS-CM-2 and ACCESS-ESM-1.5 for soybean and cotton) under SSP245 and SSP585 scenarios in the 2050s and 2080s. This increase in CWRs observed for ACCESS-CM-2 and ACCESS-ESM-1.5 under both scenarios, i.e.,SSP245 and SSP585, is more pronounced in the 2080s. The study also found thatCWRs in crops such as sugarcane, sorghum, and wheat would decrease for all ten CMIP6 GCMs under both scenarios in the 2050s and 2080s. This could be attributed to the projected decrease inTmax.Similarly, projected IWRs for all five crops under SSP245 scenario would decrease or slightly increase in the 2050s and 2080s, except for soybean and cotton. For about four CMIP6 GCMs(EC-Earth3, EC-Earth-veg, MPI-ESM1-2-HR, and NorESM2-MM), a decrease in IWR is found in the 2050s and 2080s under both scenarios. Sorghum crop has the largest decrease in IWR in the 2050s and 2080s under both scenarios, followed by wheat and sugarcane. However, at the same time, soybean exhibits the largest increase in IWR in the 2050s and 2080s under both scenarios,followed by cotton. The main reason for the drop in IWRs may be attributed to the increase in average annual rainfall. The results further show that crops like cotton and soybean are more vulnerable to climate change than other crops in the PRB. This implies that farmers may opt to increase the acreage of sorghum and wheat as compared to soybean and cotton. These results will be of great assistance to agricultural researchers and water resource managers in adopting long-term crop planning techniques to lessen the detrimental impacts of climate change on water resources management in agricultural semi-arid regions like the PRB.

Acknowledgements

This study is supported by the research project Developing Localized Indicators of Climate Change for Impact Risk Assessment in Ahmednagar using CMIP5 Data through University Grant Commission-Basic Science Research (UGC-BSR) Start-Up Grant (No. F. 30-525/2020 (BSR)). We acknowledge University Grant Commission, New Delhi for providing fund. We also extend our gratitude to two anonymous reviewers and editors for their constructive comments and great support during the revision of this paper.