Abstract

Changes in precipitation, air temperature, and model-simulated soil moisture were examined for the observed (1950–2008) and projected (2010–99) climate for the sowing period of Kharif and Rabi [KHARIF_SOW (May–July) and RABI_SOW (October–December)] and the entire Kharif and Rabi [KHARIF (May–October) and RABI (October–April)] crop-growing periods in India. During the KHARIF_SOW and KHARIF periods, precipitation declined significantly in the Gangetic Plain, which in turn resulted in declines in soil moisture. Statistically significant warming trends were noticed as all-India-averaged air temperature increased by 0.40°, 0.90°, and 0.70°C in the KHARIF, RABI_SOW, and RABI periods, respectively, during 1950–2008. Frequency and areal extent of soil moisture–based droughts increased substantially during the latter half (1980–2008) of the observed period. Under the projected climate (2010–99), precipitation, air temperature, and soil moisture are projected to increase in all four crop-growing seasons. In the projected climate, all-India ensemble mean precipitation, air temperature, and soil moisture are projected to increase up to 39% (RABI_SOW period), 2.3°C, and 5.3%, respectively, in the crop-growing periods. While projected changes in air temperature are robust across India, robust increases in precipitation and soil moisture are projected to occur in the end-term (2070–99) climate. Frequency and areal extents of soil moisture–based severe, extreme, and exceptional droughts are projected to increase in the near- (2010–39) and midterm (2040–69) climate in the majority of crop-growing seasons in India. However, frequency and areal extent of droughts during the crop-growing period are projected to decline in the end-term climate in the entire crop-growing period because of projected increases in the monsoon season precipitation.

1. Introduction

Among natural disasters, drought is considered as one of the most costly (Wilhite et al. 2000), which influences agricultural production and a variety of aspects related to society. Agriculture and allied sectors accounted for about 18% of the gross domestic product (2009–10) and employed about 52% of the total work force in India. The majority of the Indian population depends on agricultural activities for their livelihood (Krishna Kumar et al. 2004). Agricultural production is often hampered by droughts (due to precipitation or soil moisture deficit), which poses tremendous challenges for surface and groundwater resources. Prolonged precipitation deficit and high temperatures (heat waves) result in increased atmospheric water demands, which in turn deplete soil moisture in the root zone leading to agricultural droughts. In depleted soil moisture conditions, plant growth and agricultural production are severely affected, leading to declined crop yields and food insecurity during drought years (Sivakumar et al. 2005).

Global mean temperature increased significantly during the twentieth century (Hansen et al. 2006); however, there is large uncertainty in regional drought projections (Burke and Brown 2008). Studies based on observations reported mixed trends of drought occurrences during the last few decades (Dai 2011b; Wang et al. 2011; Dai 2012; Sheffield and Wood 2007). Dai (2011a), using the Palmer drought severity index (PDSI), reported that the global percentage dry area has increased by a rate of 1.74% decade−1 during the period of 1950–2008. On the other hand, Sheffield and Wood (2008a), using model-simulated, soil moisture–based drought indices, found decreasing trends associated with drought duration and severity between 1950 and 2000. Moreover, Sheffield et al. (2012) reported that there is little change in global drought during the past 60 years. They showed that drought indices such as PDSI can be highly influenced by the choice of potential evapotranspiration (PET) methods and may even lead to overestimation of drought extents.

Climate warming will cause an increase in evapotranspiration and precipitation, leading to an intensified hydrologic cycle (Huntington 2006). A changing climate may result in altered intensity, frequency, duration, and areal extent (AE) of droughts (Sheffield and Wood 2008b; Dai 2011b), which in turn can lead to unprecedented events (Seneviratne et al. 2012). Dai (2012) showed, using model-simulated soil moisture– and PDSI-based drought indices, that there is an increased risk of drought in the twenty-first century. Burke et al. (2006) reported that the fraction of land surface under extreme droughts many increase from 1% to 30% in the twenty-first century. Moreover, droughts under the projected climate could be even more detrimental because of positive feedbacks such as an increased frequency of heat waves (Seneviratne et al. 2006).

Agricultural drought mostly depends on the variability of precipitation and soil moisture and availability of moisture that can counter losses due to evapotranspiration (Sheffield et al. 2004). Under the warming climate, identification of agricultural droughts can be challenging because of complex interactions between precipitation and temperature. Additionally, a simple representation of soil moisture using evapotranspiration parameterization may lead to false conclusions (Sheffield et al. 2012). Despite the importance of soil moisture in monitoring agricultural droughts, long-term soil moisture observations have been limited to only a few regions in the world (Robock et al. 2000). While remotely sensed soil moisture could be a potential alternative to observations (Han et al. 2012), more efforts are required to assess soil moisture in deeper root zones. Land surface modeling (LSM)-based soil moisture is often used as an alternative to observed and remotely sensed soil moisture, which has been widely used for drought monitoring and characterization (Sheffield et al. 2004; Sheffield and Wood 2007; Mishra et al. 2010; Mishra and Cherkauer 2010; Andreadis et al. 2005; Wang et al. 2011). Moreover, LSM-based soil moisture may be better suited for drought analysis under the projected climate than that of other drought indices as it includes physical parameterization and complex interactions between meteorological inputs (precipitation, temperature, wind, and solar radiation) and vegetation and accounts for actual evapotranspiration rather than that of potential evapotranspiration.

Understanding the changes and variability in precipitation, temperature, and soil moisture under retrospective and projected climate in the key crop-growing seasons is important for the planning and management of water resources and ensuring food security. Moreover, an understanding of drought characteristics (frequency, areal extent, severity, and duration) and their potential for occurrence is essential for evaluating the nature of droughts under the projected climate as well as real-time monitoring and forecasting (Svoboda et al. 2002). Efforts toward understanding soil moisture drought (agricultural droughts) variability during the crop-growing seasons in India under the retrospective and projected climate have been largely limited. During the drought years, irrigation and overexploitation of groundwater resources play important roles (Rodell et al. 2009). Considering the importance of the growing population, economy, and the potential challenge to ensure food security in India, analysis of droughts in crop-growing periods is warranted. Here, we highlight soil moisture drought variability largely due to atmospheric forcing under the retrospective and projected climate in India. Moreover, our present work does not consider the influence of irrigation during the drought years. The key science questions that need to be answered are the following:

  1. To what extent have precipitation, temperature, and soil moisture changed in the key crop-growing seasons during the period of 1950–2008?

  2. How have frequency and areal extents of severe, extreme, and exceptional soil moisture droughts changed in India?

  3. How will droughts under the projected climate change during the key crop-growing seasons in India?

2. Methods

a. Observed data

We used daily precipitation data at 0.25° spatial resolution for the period of 1950–2008, which were developed by the Indian Meteorological Department (IMD; Pai et al. 2014). The gridded daily precipitation data obtained from IMD were developed using 6995 stations located across India (Pai et al. 2014), and in the new gridded precipitation product, climatological features of precipitation are well represented, including orographic precipitation in the Western Ghats and northeastern India. Further details on data can be obtained from Pai et al. (2014). Since daily maximum and minimum air temperature data from IMD were not available for the analysis period (1950–2008), we obtained daily temperatures and wind speed forcings from Sheffield et al. (2006). Precipitation at 0.25° was regridded to 0.50° spatial resolution to make it consistent with daily maximum and minimum temperatures and wind speed datasets. Precipitation from IMD better captures temporal and spatial variability of the Indian monsoon (Pai et al. 2014) rather than data from other sources [e.g., Climatic Research Unit (CRU) in Sheffield et al. (2006)]; drought simulations using these datasets can be considered more realistic for India. The Sheffield et al. (2006) dataset is a hybrid of data from the National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis (Kalnay et al. 1996) and global observation–based products of precipitation, temperature, and radiation. Further details about the observed temperature forcing can be obtained from Sheffield et al. (2006). The datasets have been widely used for drought and other hydrologic analysis in many parts of the world (Sheffield and Wood 2007; Sheffield et al. 2012; Wang et al. 2011; Demaria et al. 2013; Jury and Funk 2012). We carefully compared daily maximum and minimum temperature data from Sheffield et al. (2006) and IMD for the overlapping period (1969–2005) and found that daily temperatures from Sheffield et al. (2006) captured seasonal cycle and temporal and spatial variability reasonably well.

b. Bias-corrected and spatially downscaled climate projections

Bias-corrected and spatially disaggregated (BCSD) data were used to evaluate changes in soil moisture droughts under the projected climate. The modified BCSD approach (Thrasher et al. 2012) was used to develop daily meteorological forcings using the daily precipitation, maximum and minimum temperatures, and diurnal temperature range (DTR) outputs from seven general circulation models (GCMs) for the period of 1950–2099. Daily outputs of precipitation and air temperature were obtained from the GCMs that participated in phase 5 of the Coupled Model Intercomparison Project (CMIP5). Daily data from the GCMs were obtained from ensemble member r1i1p1 [the first realization, initialization, and set of perturbed physics; see Taylor et al. (2009) for details] for representative concentration pathway (RCP) 4.5 (RCP4.5), which assumes an increase of 4.5 W m−2 in radiative forcing by the end of the twenty-first century (Taylor et al. 2012). Rather than the most pessimistic (RCP8.5) or the most optimistic (RCP2.6), we selected an intermediate scenario (RCP4.5) as it represents midrange mitigation emission scenario in which the change in radiative forcing stabilizes after the twenty-first century. Because of uncertainty in the climate model projections that could vary regionally, data from the seven GCMs [Beijing Climate Center, Climate System Model, version 1.1 (BCC_CSM1.1); Centre National de Recherches Météorologiques Coupled Global Climate Model, version 5 (CNRM-CM5); Institute of Numerical Mathematics Coupled Model, version 4.0 (INM-CM4.0); L’Institut Pierre-Simon Laplace Coupled Model, version 5, coupled with the Nucleus for European Modelling of the Ocean (NEMO), low resolution (IPSL-CM5A-LR); Max Planck Institute Earth System Model, low resolution (MPI-ESM-LR); Meteorological Research Institute Coupled Atmosphere–Ocean General Circulation Model, version 3 (MRI-CGCM3); and Norwegian Earth System Model, version 1 (intermediate resolution) (NorESM1-M)] were used for the downscaling and bias correction. The modified BCSD approach (Thrasher et al. 2012) is different from the original BCSD method (Wood et al. 2002, 2004; Maurer et al. 2010) as this uses daily projections of precipitation and maximum and minimum temperatures rather than monthly precipitation and average temperature. As the modified BCSD approach uses daily datasets, it essentially avoids daily data disaggregation from bias-corrected monthly data using daily time series from a monthly historic climatology as used in the original BCSD approach. The BCSD approach has been widely used for hydrologic impact assessments (Hayhoe et al. 2004; Mishra et al. 2010; Cayan et al. 2008). Moreover, the BCSD approach has been successfully compared to various statistical and dynamical downscaling techniques for both mean and extremes (Wood et al. 2004; Maurer and Hidalgo 2008; Bürger et al. 2012). Bias-corrected and spatially disaggregated daily datasets were developed for all seven GCMs at 0.5° spatial resolution and daily temporal resolution. Consistent with the historic climatology, gridded future climate projections including daily precipitation and maximum and minimum temperatures were developed for the period of 1950–2099. The observed climatological data for the BCSD were obtained from Sheffield et al. (2006).

c. The VIC model

The Variable Infiltration Capacity (VIC; Liang et al. 1994, 1996; Cherkauer et al. 2003; Cherkauer and Lettenmaier 1999) model is a macroscale hydrology model that simulates energy and water fluxes at each grid cell. In the VIC model, multiple vegetation types can be represented in the same grid cell using the mosaic scheme. Land cover classes are represented using the root fraction, canopy resistance, leaf area index (LAI), and some other parameters. The main distinguishing characteristics of the VIC model include subgrid variability in soil moisture storage capacity, subgrid variability in land cover classes, representation of drainage from the lower soil layer as a nonlinear recession, and inclusion of topography that allows for orographic precipitation and temperature lapse rates resulting in more realistic hydrologic processes in the heterogeneous topographic regions (H. Gao et al. 2010, unpublished manuscript, available online at http://www.hydro.washington.edu/Lettenmaier/Publications/Water_Cycle_MEaSUREs_ATBD_VICmodel_submit.doc).

The VIC model has been widely applied for the various hydrologic applications from regional to global scales (e.g., Nijssen et al. 2001; Maurer et al. 2002; Mishra et al. 2010; Raje et al. 2014; Raje and Krishnan 2012). Soil moisture estimations from the VIC model have been proven to be reliable (Nijssen et al. 2001; Maurer et al. 2002; Mishra et al. 2010). Furthermore, the model has been applied to reconstruct droughts in many regions of the world (Sheffield et al. 2004; Sheffield and Wood 2007; Andreadis et al. 2005; Andreadis and Lettenmaier 2006; Mishra and Cherkauer 2010).

In the present study, the VIC model was applied at 0.5° spatial resolution and daily temporal resolution in the water balance mode across India for the observed (1948–2008) and projected (2010–99) climate. The meteorological forcings for the observed climate were obtained from the IMD and Sheffield et al. (2006). For the projected climate, we used bias-corrected and downscaled forcings as described in the previous section. The vegetation parameters used in this study are described in detail in Sheffield and Wood (2007), in which vegetation types were extracted from the 1-km Advanced Very High Resolution Radiometer (AVHRR) global land cover information (Hansen et al. 2000). We obtained soil data from the Harmonized World Soil Database (HWSD), version 1.2, and soil parameters were developed for the VIC model using the methods available on the VIC model website (www.hydro.washington.edu/Lettenmaier/Models/VIC/).

We evaluated soil moisture from the VIC model against the satellite-based observations from the European Space Agency Climate Change Initiative (ESA CCI) and in situ observations at the Indian Institute of Technology (IIT) Kanpur, which is a part of International Soil Moisture Network (ISMN; Dorigo et al. 2011). Soil moisture from ESA CCI is a merged product from various sensors and is available at daily temporal resolution and 0.25° spatial resolution for the period of 1978–2010 (Hollmann et al. 2013). Dorigo et al. (2012) used the ESA CCI soil moisture data to evaluate trends in soil moisture. Moreover, they also reported that ESA CCI soil moisture compares well with the Noah model–simulated soil moisture available through the Global Land Data Assimilation System (GLDAS). Evaluation of annual average VIC-simulated soil moisture for the top layer against the ESA CCI soil moisture shows a good agreement in spatial variability and correlation (Figs. 1a–c). Moreover, the VIC model–simulated soil moisture successfully captured seasonal variability in soil moisture at IIT Kanpur for 25- and 80-cm depths (Figs. 1d,e). During the nonmonsoon months, the VIC model overestimates soil moisture (Figs. 1d,e), which could be attributed to differences in meteorological forcing (point versus grid scale) and other parameters related to soil and vegetation. We compared seasonal cycle and persistence of total-column soil moisture from our simulations with those obtained from the Noah land surface model from GLDAS. Soil moisture in the Noah model developed through soil moisture data assimilation from Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E; Rodell et al. 2004). Comparisons for seasonal cycle and persistence were found satisfactory (Figs. 1f,g) and provide confidence for the implementation of the VIC model for drought study over India. Comparisons of soil moisture persistence at lag 4 and 6 months for all of India as well as for the five selected regions (see Fig. S1) are presented in Fig. S2 (Figs. S1 and S2 are available as supplemental material). Results show that the VIC model exhibits lower soil moisture persistence than that of the Noah model. Soil moisture from the VIC model has been proven to be reliable for drought studies (Sheffield and Wood 2007; Mishra and Cherkauer 2010; Mishra et al. 2010; Wang et al. 2011).

Fig. 1.

Volumetric soil moisture for the top 5 cm from (a) ESA CCI and (b) the VIC model; (c) correlation coefficient between ESA CCI and VIC soil moisture; (d),(e) evaluation of the VIC model–simulated soil moisture against in situ observations at IIT Kanpur (26.52°N, 80.23°E); (f) comparison of seasonal cycle of all-India-averaged soil moisture from the VIC and Noah models; and (g) comparison of all-India-averaged soil moisture persistence from the VIC and Noah model.

Fig. 1.

Volumetric soil moisture for the top 5 cm from (a) ESA CCI and (b) the VIC model; (c) correlation coefficient between ESA CCI and VIC soil moisture; (d),(e) evaluation of the VIC model–simulated soil moisture against in situ observations at IIT Kanpur (26.52°N, 80.23°E); (f) comparison of seasonal cycle of all-India-averaged soil moisture from the VIC and Noah models; and (g) comparison of all-India-averaged soil moisture persistence from the VIC and Noah model.

d. Analysis approach

The analysis for the observed (1950–2008) and projected (2010–99) climate was conducted to estimate trends/changes in climate variables (precipitation and air temperature) and soil moisture for key crop-growing seasons. In the two crop-growing seasons, the sowing period for the Kharif (KHARIF_SOW; May–July), the entire Kharif period (KHARIF; May–October), the sowing period for the Rabi (RABI_SOW; October–December), and the entire Rabi period (RABI; October–April) were identified as those having high importance for agricultural production in India. An analysis was conducted separately for the sowing periods of the KHARIF and RABI as dry conditions in these may lead to late sowing, which in turn can severely affect crop yields. We have selected relatively long sowing periods as the sowing time varies with latitude, type of crop, and sometimes with availability of moisture. Crops (rice, maize, cotton, sunflower, and soybean) in the KHARIF season are mainly based on the monsoon season precipitation. On the other hand, most of the crops (wheat, gram, barley, peas, and mustard) in the RABI season are either rain fed or depend on irrigation. The period of 1961–90 was considered as the base period for the observed as well as the projected climate. For the projected climate, the data for the base period were obtained through the historic runs of the models. Changes under the projected climate were estimated with respect to the base period (1961–90 from climate models), while anomalies in the observed (1950–2008) period were estimated with respect to the observed base period (1961–90).

1) Soil moisture percentiles

Droughts were identified using the soil moisture percentiles (Sheffield et al. 2004; Andreadis et al. 2005). Only severe, extreme, and exceptional droughts (soil moisture percentile <10) were considered in the analysis as crops are often substantially affected by these (Svoboda et al. 2002). To estimate percentiles for the selected crop-growing seasons (KHARIF_SOW, KHARIF, RABI_SOW, and RABI), mean seasonal soil moisture was estimated and then percentiles were calculated using the Weibull plotting positions. For the observed climate, soil moisture percentiles were estimated for the period of 1950–2008. For the projected climate, soil moisture percentiles were estimated for the periods 2010–39, 2040–69, and 2070–99 using the VIC model simulations that were driven by the downscaled and bias-corrected meteorological forcing. The period of 1961–90 from the historic runs of the climate models was considered as a base period to percentile estimation in the observed and projected climate.

2) Trend analysis

The Mann–Kendall trend test (Mann 1945) with Sen’s slope (Sen 1968) was used to estimate trends in observed precipitation, temperature, and soil moisture during the period of 1950–2008. The Mann–Kendall test was applied at 5% significance level. The test has been widely used to detect trends in hydrologic/climatic datasets (Mishra and Lettenmaier 2011). For the Mann–Kendall analysis, an assumption of independence in hydrologic/climatic dataset is often considered suitable for the annual time series of hydrologic data. However, as soil moisture often has persistence, effects of autocorrelation and spatial correlation (field significance) in soil moisture were therefore considered using the modified Mann–Kendall trend test as described in Yue and Wang (2002). In the results, we present changes in the retrospective period that were obtained by multiplying trend slope with total number of years.

3) Mapping model agreement

Projections for the future climate are most often obtained from simulations from multiple GCMs with one or more emission scenarios. As the associated uncertainty with the climate model projections may be large, some form of stippling or robustness that could indicate model agreement is essential (Tebaldi et al. 2011). Robustness of model agreement was estimated using the method described in Tebaldi et al. (2011). The method can display changes and an agreement among models in a multimodel ensemble that can separate lack of signal from lack of information due to disagreement among models (Tebaldi et al. 2011). To map robustness in the BCSD projections, the following procedure was adopted:

  1. For an individual GCM, it was tested if the mean of the reference (base) and projected climate periods (near-term, 2010–39; midterm, 2040–69; and end-term, 2070–99) are significantly different using the two-sided rank-sum test.

  2. If less than 50% of the GCMs (less than four for the present analysis) show a significant change, then multimodel ensemble mean is shown in color.

  3. If more than 50% of the GCMs show a significant change, then agreement in sign (positive or negative) is tested using the following steps: (i) if less than 80% of the models (that showed a significant difference in mean) agree on sign, then display the grid cell as white, and (ii) if more than 80% of the models (that showed a significant difference in mean) agree on sign, then display grid cell with color and stippling. The grid cells with stippling are considered with robust positive/negative changes.

3. Results

a. Changes in the observed (1950–2008) climate

Figure 2 shows trends and spatial variability in precipitation, air temperature, and soil moisture during KHARIF_SOW (May–July) for the period of 1950–2008. Observed precipitation successfully captures the spatial variability associated with higher values in the Western Ghats and the northeastern regions (Fig. 2a). Between 1950 and 2008, precipitation significantly declined during the KHARIF_SOW period in Gangetic Plain, northeastern region, and in southern peninsular India (Fig. 2b). On the other hand, the regions that showed increases in precipitation were located in the Kashmir region and in northwestern India. During the period of 1950–2008, mean seasonal air temperature in the KHARIF_SOW season varied between 5° and 35°C (Fig. 2d) with high temperatures in the western parts and lower temperatures in northern India (primarily in Jammu and Kashmir). Regions that showed significant increases in temperatures are located in Jammu and Kashmir, the southern peninsula, Gujarat, and northeastern India (Fig. 2e). Areas with significant declines are located in the Gangetic Plain (especially the foothills of Himalaya). Model-simulated total-column soil moisture varied between 100 and 800 mm during the KHARIF_SOW period, with higher soil moisture in the northeastern region, the Western Ghats, and northern Kashmir and lower soil moisture in western India (Fig. 2g). The regions with declining trends in soil moisture are located in the Gangetic Plain and northeastern and peninsular India. During the KHARIF_SOW period, trends were not found statistically significant except for soil moisture. Declines in soil moisture in the Gangetic Plain region are largely driven by reduced precipitation (Mishra et al. 2012; Bollasina et al. 2011), as temperature declined in the same period in the majority of the Gangetic Plain region.

Fig. 2.

Changes (trend slope multiplied by total number of years) in precipitation, temperature, and model-simulated total-column soil moisture during the KHARIF_SOW season for the period of 1950–2008. (a),(d),(g) Mean seasonal precipitation (mm), temperature (°C), and model-simulated total-column soil moisture (mm) for the retrospective base period (1961–90); (b),(e),(h) changes in precipitation (%), air temperature (°C), and total-column soil moisture (%) between 1950 and 2008; and (c),(f),(i) all-India-averaged seasonal anomalies of precipitation (%), air temperature (°C), and total-column soil moisture (%) during the period of 1950–2008. Changes were estimated using the nonparametric Mann–Kendall trend test and Sen’s slope, with stippling indicating statistically significant changes at the 5% level in (b),(e), and (h). Changes were estimated for all-India-averaged anomalies of precipitation, air temperature, and simulated total-column soil moisture in (c),(f), and (i). A p value of less than 0.05 indicates that changes are statistically significant. Anomalies for precipitation, temperature, and soil moisture were estimated with respect to mean of the retrospective base period.

Fig. 2.

Changes (trend slope multiplied by total number of years) in precipitation, temperature, and model-simulated total-column soil moisture during the KHARIF_SOW season for the period of 1950–2008. (a),(d),(g) Mean seasonal precipitation (mm), temperature (°C), and model-simulated total-column soil moisture (mm) for the retrospective base period (1961–90); (b),(e),(h) changes in precipitation (%), air temperature (°C), and total-column soil moisture (%) between 1950 and 2008; and (c),(f),(i) all-India-averaged seasonal anomalies of precipitation (%), air temperature (°C), and total-column soil moisture (%) during the period of 1950–2008. Changes were estimated using the nonparametric Mann–Kendall trend test and Sen’s slope, with stippling indicating statistically significant changes at the 5% level in (b),(e), and (h). Changes were estimated for all-India-averaged anomalies of precipitation, air temperature, and simulated total-column soil moisture in (c),(f), and (i). A p value of less than 0.05 indicates that changes are statistically significant. Anomalies for precipitation, temperature, and soil moisture were estimated with respect to mean of the retrospective base period.

Changes in precipitation, air temperature, and soil moisture in the KHARIF (May–October) period are presented in Fig. 3. Since the KHARIF (May–October) period overlaps with the summer monsoon [June–September (JJAS)] season, mean seasonal precipitation is high (Fig. 3a, Table 1). Observed precipitation successfully reproduced spatial variability in the Western Ghats, east-central, and northeastern regions. Between 1950 and 2008, the Gangetic Plain and parts of western India showed significant declines in precipitation (Fig. 3b), which has been observed in other studies (Bollasina et al. 2011; Mishra et al. 2012; Naidu et al. 2009). On the other hand, a few regions with increases in precipitation are located in peninsular India and coastal Orissa. During the KHARIF season, air temperature varied between 5° and 35°C across India (Fig. 3d, Table 1). Air temperature increased significantly over the majority of India, with more prominent increases in the northern and southern peninsula during the period of 1950–2008 (Fig. 3e). In the KHARIF season, all-India-averaged temperature increased by 0.41°C during the last few decades (Fig. 3f). A significant decline in soil moisture was noticed in the Gangetic Plain that was largely driven by reduced precipitation. However, since our precipitation forcing differs from those used in Sheffield and Wood (2008a), our results show a lower extent of drying trends in the Gangetic Plain region. Precipitation data in Sheffield and Wood (2008a) are based on CRU, which exhibits a stronger decline in precipitation over the Gangetic Plain.

Fig. 3.

As in Fig. 2, but for the KHARIF season.

Fig. 3.

As in Fig. 2, but for the KHARIF season.

Table 1.

Mean precipitation, temperature, and soil moisture in the selected crop-growing periods during 1950–2008.

Mean precipitation, temperature, and soil moisture in the selected crop-growing periods during 1950–2008.
Mean precipitation, temperature, and soil moisture in the selected crop-growing periods during 1950–2008.

Figure 4 shows changes and variability in precipitation, temperature, and soil moisture during the RABI_SOW (October–December) period. During the RABI_SOW period, parts of the southern peninsula and northeastern India receive relatively higher precipitation (Fig. 4a). Precipitation increased in the southern peninsula, while prominent declines were found in the Western Ghats, Gangetic Plain, and central and northeastern regions of India (Fig. 4b). Significant increases in air temperature were noticed in the majority of India during 1950–2008 in the RABI_SOW period (Fig. 4e). Moreover, in some parts of India (Kashmir and central parts), air temperature increased over 1°C during the period of 1950–2008. All-India-averaged temperature significantly increased (0.90°C, p value = 0.0003) with consistent positive anomalies after 1985 (Fig. 4f). Soil moisture declined significantly in the Gangetic Plain and central and western India in the RABI_SOW period (Fig. 4h). Prominent increases in temperature during the RABI_SOW period may have profound implications for groundwater resources because of increased requirements for irrigation.

Fig. 4.

As in Fig. 2, but for the RABI_SOW season.

Fig. 4.

As in Fig. 2, but for the RABI_SOW season.

Trends and variability in precipitation, air temperature, and soil moisture for the RABI (October–April) period are presented in Fig. 5. The southern peninsula and northeastern and northern India receive precipitation of 150 mm or above in the RABI season (Fig. 5a). Trends in precipitation indicate increases in the western, southern, and northeastern parts of India and declines in central India (Fig. 5b). Precipitation declined significantly in the RABI season in the majority of central India during the period of 1950–2008 (Fig. 5b). However, all-India-averaged precipitation showed increases of 9.4% (p value = 0.16) in the RABI season, which could be attributed to large increases in Jammu and Kashmir, western India, and parts of peninsular India (Fig. 5c). Similar to the RABI_SOW period, air temperature significantly increased in the majority of India during the period of 1950–2008 (Fig. 5e). The western, southern, and northern parts showed increases of more than 1°C in the RABI period. All-India-averaged air temperature showed a significant warming of 0.71°C (p value = 0.000 14) during the period of 1950–2008 (Fig. 5f). Significant declines in soil moisture were found in western India and Gangetic Plain, while some regions in the southern peninsula showed significant increases between 1950 and 2008 (Fig. 5h). In the RABI period, increased temperatures may have profound implications for crop yields as well as requirements for irrigation, especially in northwestern India. Moreover, warming in the RABI_SOW season led to an early sowing period in northwestern India, as reported by Lobell et al. (2013). Observed mean and changes showed a decline in precipitation and soil moisture in all the selected crop-growing seasons in western India (see Table S1 in the supplemental material).

Fig. 5.

As in Fig. 2, but for RABI season.

Fig. 5.

As in Fig. 2, but for RABI season.

b. Observed droughts

Between 1950 and 2008, three major drought events occurred in 1972, 1987, and 2002 (Drought in India report, www.cccindia.co/corecentre/Database/Docs/DocFiles/drought_india.pdf). The model-simulated soil moisture percentile–based droughts successfully reproduced these droughts during the crop-growing periods in India (Fig. 6). During 1972, severe, extreme, and exceptional (soil moisture percentile <10) category drought was predominantly located in the southwestern, central, and northeastern parts of India (Figs. 6a–d). In 1972, the areal extent of droughts in the KHARIF_SOW, KHARIF, RABI_SOW, and RABI periods were 22%, 28.7%, 23.8%, and 25.9%, respectively (Table 2).

Fig. 6.

The major (1972, 1987, and 2002) growing season soil moisture droughts in India. Soil moisture percentiles below 10 indicate severe, extreme, and exceptional category droughts.

Fig. 6.

The major (1972, 1987, and 2002) growing season soil moisture droughts in India. Soil moisture percentiles below 10 indicate severe, extreme, and exceptional category droughts.

Table 2.

Major drought events and their AE during the crop-growing season in India.

Major drought events and their AE during the crop-growing season in India.
Major drought events and their AE during the crop-growing season in India.

During 1987, severe, extreme, and exceptional category soil moisture drought covered most of India (Figs. 6e,f), with a predominance in western India. The most widespread (areal extent = 35%) drought occurred during the KHARIF period (Fig. 6f). About 56.8 million hectares of cropped area were damaged and over 285 million people were affected by the drought of 1987 (www.dsc.nrsc.gov.in/DSC/Drought). In 1987, drought was persistent and widespread and affected all crop-growing periods (Table 2).

Soil moisture percentile–based drought showed a widespread influence and persistent nature during 2002 (Figs. 6i–l). During the KHARIF_SOW period, severe, extreme, and exceptional category drought was located in the Gangetic Plain and northwestern India (Fig. 6i), which covered a majority of India in the KHARIF period (Fig. 6j). In all the crop-growing periods, the Gangetic Plain and northwestern India were severely affected by the drought of 2002. Since the drought started during the KHARIF_SOW period, planted area was reduced from 124 to 112 million hectares. Moreover, total food grain production declined from 212 to 174 million tons (www.dsc.nrsc.gov.in/DSC/Drought). The areal extents of severe, extreme, and exceptional category droughts were 25.7%, 30.0%, 23.7%, and 23.2%, respectively, in the KHARIF_SOW, KHARIF, RABI_SOW, and RABI periods (Table 2). Out of the top five widespread droughts during the crop-growing periods, at least three occurred after 1979, showing an increased tendency of droughts during the recent decades (Table 2).

Figure 7 shows observed variability in areal extents of soil moisture–based droughts in the key crop-growing periods. During the KHARIF_SOW period, areal extent of severe, extreme, and exceptional droughts increased by 0.22% (p value = 0.91) between 1950 and 2008 (Fig. 7a). Median areal extents for the early (1951–79) and late (1980–2008) observed periods were 9% and 7%, respectively, and no statistically significant difference in median was found using the two-sided rank-sum test at 5% significance level (Fig. 7b). Between 1950 and 2008, areal extent of droughts increased by 3.5% (p value = 0.14) in the KHARIF period (Fig. 7c); however, changes in areal extent in the early and the late periods were not statistically significant (Fig. 7d). Areal extents of droughts increased by 4% (p value = 0.13) in the RABI_SOW period (Fig. 7e). Median areal extents in the early and the late observed periods were 4% and 9%, respectively (Fig. 7f). Changes in median areal extents were statistically significant at the 5% level. A positive trend (3.6%, p value = 0.22) was also noticed in droughts in the RABI period (Fig. 7g). The five most widespread droughts in the selected periods are presented in Table 2.

Fig. 7.

AE of severe, extreme, and exceptional droughts (soil moisture percentile <10) for (a) KHARIF_SOW, (c) KHARIF, (e) RABI_SOW, and (g) RABI seasons during 1950–2008. (b),(d),(f),(h) Median extent of droughts in the growing seasons for the early (1951–79) and late (1980–2008) retrospective periods.

Fig. 7.

AE of severe, extreme, and exceptional droughts (soil moisture percentile <10) for (a) KHARIF_SOW, (c) KHARIF, (e) RABI_SOW, and (g) RABI seasons during 1950–2008. (b),(d),(f),(h) Median extent of droughts in the growing seasons for the early (1951–79) and late (1980–2008) retrospective periods.

The frequency of severe, extreme, and exceptional soil moisture–based droughts during the crop-growing periods was estimated for the early (1951–79) and the late (1980–2008) periods to evaluate changes during the recent decades (Fig. 8). In the early and the late observed periods, the number of droughts of severe, extreme, and exceptional category ranged between 1 and 7. In the early observed period, more frequent droughts occurred in northern, northeastern, eastern coastal, and peninsular India during the key crop-growing periods (Figs. 8a,d,g,j). On the other hand, droughts became more frequent in the Gangetic Plain and western India during the late (1980–2008) observed period (Figs. 8b,e,h,k). The Gangetic Plain and western India experienced an increase of about 3–5 droughts in the KHARIF_SOW period during the recent decades (Figs. 8c,f,i,l). These results show an increasing frequency of severe, extreme, and exceptional droughts in one of the most fertile regions in India.

Fig. 8.

Frequency of severe, extreme, and exceptional droughts (soil moisture percentile <10) in the (a),(d),(g),(j) early (1951–79) and (b),(e),(h),(k) late (1980–2008) retrospective periods. (c),(f),(i),(l) Changes in frequency of droughts in the late retrospective period.

Fig. 8.

Frequency of severe, extreme, and exceptional droughts (soil moisture percentile <10) in the (a),(d),(g),(j) early (1951–79) and (b),(e),(h),(k) late (1980–2008) retrospective periods. (c),(f),(i),(l) Changes in frequency of droughts in the late retrospective period.

c. Observed impacts on crop yields

Total food grain yields in India increased significantly between 1950 and 2008 (Fig. 9a). The green revolution and improved irrigation and fertilizer inputs played important roles. To evaluate the impacts of climate on crop yields, we estimated first the difference (Lobell and Field 2007) of total food grain yield in India. The first difference method removes the influence of management practices (e.g., improved seed, irrigation, and fertilizers) from crop yields, and they are therefore considered suitable to assess impacts of climate on yields (Lobell and Field 2007). The first difference time series of total food grain yield showed frequent reductions in yield during the last few decades, which can be attributed to a decline in precipitation and increase in air temperatures during the KHARIF and RABI periods (Fig. 9b). We estimated correlation coefficient between crop yield (first difference time series) and precipitation and soil moisture anomalies and drought extent during the KHARIF_SOW and KHARIF periods. Precipitation and soil moisture anomalies showed positive correlations with crop yields, while drought extent based on soil moisture showed negative correlations (Figs. 9c–h). These results highlight the importance of the monsoon season precipitation on total food grain production in India.

Fig. 9.

(a) Total food grain yields in India between 1950 and 2008; (b) first difference of total food grain yields; (c)–(e) correlation between total food grain yields and precipitation anomaly (%), temperature anomaly (°C), and AE of soil moisture drought during KHARIF_SOW period; and (f)–(h) as in (c)–(e) but for the KHARIF period.

Fig. 9.

(a) Total food grain yields in India between 1950 and 2008; (b) first difference of total food grain yields; (c)–(e) correlation between total food grain yields and precipitation anomaly (%), temperature anomaly (°C), and AE of soil moisture drought during KHARIF_SOW period; and (f)–(h) as in (c)–(e) but for the KHARIF period.

d. Changes under the projected climate

We evaluated observed and bias-corrected and downscaled forcing from CMIP5 for the period of 1950–2005 (Fig. 10). Bias-corrected and downscaled precipitation and air temperature were obtained for the historic runs of CMIP5. We notice that bias-corrected and downscaled precipitation and temperature reproduced the observed seasonal cycle in a reasonable manner. However, during the monsoon season, the bias-corrected and downscaled forcings show an underestimation in precipitation and an overestimation in temperature. Bias in temperature and precipitation forcing can be attributed to a large bias present in precipitation and temperature during the monsoon season in CMIP5 (Levine et al. 2013).

Fig. 10.

Evaluation of downscaled and bias-corrected all-India-averaged (a) precipitation and (b) air temperature against observed data for the period of 1950–2005. Red line indicates multimodel ensemble mean while error bars represent intermodel variability (std dev).

Fig. 10.

Evaluation of downscaled and bias-corrected all-India-averaged (a) precipitation and (b) air temperature against observed data for the period of 1950–2005. Red line indicates multimodel ensemble mean while error bars represent intermodel variability (std dev).

Figure 11 shows projected ensemble mean changes in precipitation, air temperature, and soil moisture in the KHARIF_SOW period in the near- (2010–39), mid- (2040–69), and end-term (2070–99) climate. Ensemble mean change for each GCM was estimated with respect to mean of the historic base period (1961–90). During the KHARIF_SOW period, all-India ensemble mean precipitation is projected to increase by 6.0%, 11.1%, and 21.4% in the near-, mid-, and end-term projected climate, respectively (Table 3). Higher increases in precipitation are projected in northeastern, central, and peninsular India (Figs. 11a–c). All-India ensemble mean air temperature is projected to increase by 1.0°, 1.6°, and 2.0°C in the near-, mid-, and end-term climate, respectively (Figs. 11d–f). Moreover, projected increases in temperature are higher in the northern parts of India than that of the southern parts. Projected warming was found robust across India, showing that there is a good agreement among the selected GCMs. Under the future climate, soil moisture is projected to decline in parts of Kashmir while projected to increase in the Gangetic Plain, the east-central region, and in peninsular India (Figs. 11g–i). All-India ensemble mean soil moisture is projected to increase by 1.2%, 2.3%, and 4.5%, respectively, in the near-, mid-, and end-term projected climate (Table 3).

Fig. 11.

Multimodel ensemble mean changes in precipitation, temperature, and soil moisture in the near- (2010–39), mid- (2040–69), and end-term (2070–99) climate during the KHARIF_SOW season: (a)–(c) precipitation (%), (d)–(f) temperature (°C), and (g)–(i) soil moisture (%). Stippling indicates that changes are robust. Changes were estimated for each individual GCM with respect to the historic base period (1961–90) from the climate models and then mean was taken to estimate multimodel ensemble mean change.

Fig. 11.

Multimodel ensemble mean changes in precipitation, temperature, and soil moisture in the near- (2010–39), mid- (2040–69), and end-term (2070–99) climate during the KHARIF_SOW season: (a)–(c) precipitation (%), (d)–(f) temperature (°C), and (g)–(i) soil moisture (%). Stippling indicates that changes are robust. Changes were estimated for each individual GCM with respect to the historic base period (1961–90) from the climate models and then mean was taken to estimate multimodel ensemble mean change.

Table 3.

All-India-averaged multimodel ensemble mean changes in precipitation, air temperature, and soil moisture in the selected crop-growing periods.

All-India-averaged multimodel ensemble mean changes in precipitation, air temperature, and soil moisture in the selected crop-growing periods.
All-India-averaged multimodel ensemble mean changes in precipitation, air temperature, and soil moisture in the selected crop-growing periods.

Statistically downscaled and bias-corrected projections for the KHARIF period showed increases in precipitation, air temperature, and soil moisture under the projected climate (Fig. 12). All-India ensemble mean precipitation is projected to increase by 10.0%, 17.8%, and 27.1%, respectively, in the near-, mid-, and end-term future climate (Figs. 12a–c, Table 3). Air temperature projections suggest a consistent warming in the near-, mid-, and end-term climate across India (Figs. 12d–f). All-India ensemble mean temperature is projected to increase by 0.9°, 1.5°, and 1.9°C, respectively, in the near-, mid-, and end-term climate (Table 3). In the KHARIF period, ensemble mean soil moisture is projected to increase across India, except parts in Kashmir (Figs. 12g–i). All-India ensemble mean soil moisture is projected to increase by 1.6%, 3.1%, and 5.3%, respectively, in the near-, mid-, and end-term climate.

Fig. 12.

As in Fig. 11, but for KHARIF season.

Fig. 12.

As in Fig. 11, but for KHARIF season.

Figure 13 shows ensemble mean projections in precipitation, air temperature, and soil moisture in the RABI_SOW period under the projected climate. Consistent with the KHARIF_SOW and KHARIF periods, precipitation is projected to increase in the RABI_SOW period under the projected climate across India (Figs. 13a–c). All-India ensemble mean precipitation is projected to increase by 17.1%, 25.5%, and 39.3% (RABI_SOW period) in the near-, mid-, and end-term projected climate. During the RABI_SOW period, consistent warming is projected across India in the near-, mid-, and end-term climate (Figs. 13d–f). Moreover, the northern parts in India are projected to experience relatively higher warming than the rest of India (Figs. 13d–f). All-India ensemble mean projected increases in air temperature were 1.0°, 1.8°, and 2.2°C in the near-, mid-, and end-term climate. Because of projected increases in precipitation, soil moisture is projected to increase across India in the near-, mid-, and end-term climate (Figs. 13g–i). All-India ensemble mean projected increases were 1.8%, 3.2%, and 4.7% in the near-, mid-, and end-term climate (Table 3).

Fig. 13.

As in Fig. 11, but for RABI_SOW season.

Fig. 13.

As in Fig. 11, but for RABI_SOW season.

Under the projected climate, precipitation, air temperature, and soil moisture are projected to increase in the RABI period (Fig. 14). Prominent increases in precipitation are projected in western and southern India (Figs. 14a–c). All-India ensemble mean precipitation is projected to increase by 11.3%, 15.0%, and 28.1%, respectively, in the near-, mid-, and end-term climate. A robust warming of 1°–1.5°C was shown in temperature projections in the near-term climate, which increased up to 2°–3°C in the mid- and end-term climate (Figs. 14d–f). All-India ensemble mean temperature is projected to increase by 1.1°, 2.0°, and 2.3°C, respectively, in the near-, mid-, and end-term climate (Table 3). Soil moisture in the RABI period is projected to increase in the majority of India, except in the northern and northeastern regions (Figs. 14g–i). However, soil moisture changes were largely not robust in the near- and midterm climate. The most robust increase was noticed in air temperature, which is projected to increase significantly across India in all the crop-growing periods under the projected climate (Table 4). Moreover, about 62% of India is projected to experience significant increases in precipitation during the KHARIF period (Table 4). Projections for the selected regions (Fig. S1 in the supplemental material) show a prominent warming and increases in precipitation in most of the selected regions in India (Tables S2–S4 in the supplemental material).

Fig. 14.

As in Fig. 11, but for RABI season.

Fig. 14.

As in Fig. 11, but for RABI season.

Table 4.

Percentage area of India under the robust positive (negative) changes under the projected climate. Values in parentheses show percentage with robust negative changes.

Percentage area of India under the robust positive (negative) changes under the projected climate. Values in parentheses show percentage with robust negative changes.
Percentage area of India under the robust positive (negative) changes under the projected climate. Values in parentheses show percentage with robust negative changes.

Uncertainties associated with precipitation projections under the future climate are large (Knutti and Sedláček 2013; Turner and Annamalai 2012); therefore, these results may be considered to develop plausible scenarios for planning and water management. Menon et al. (2013) reported consistent increases in the monsoon season precipitation in India under the projected climate. Moreover, they found that increases in mean and variability in the short and long term in the Indian summer monsoon are robust and present in a wide range of models. Our results, therefore, are in agreement, as we find a consistent increase in precipitation in the key crop-growing seasons under the projected climate.

e. Droughts under the projected climate

Figure 15 shows frequency of severe, extreme, and exceptional droughts during the selected crop-growing periods under the projected climate. In the historic base period (1961–90), ensemble mean drought frequency varied between one and four droughts in the key crop-growing periods (Figs. 15a,e,i,m). The regions that experienced a high frequency of droughts in the historic base period are the Gangetic Plain, the northeastern region, and southern peninsula. Under the projected climate, ensemble mean drought frequency is projected to increase in the near- and midterm climate in all the selected crop-growing periods (Figs. 15b,c,f,g,j,k,n,o). In the near- and midterm, droughts are projected to be more frequent in Kashmir, the western and central regions, and parts of the southern peninsula in the KHARIF_SOW and KHARIF periods. During the RABI_SOW and RABI periods, drought frequency is projected to decline in the majority of India, except the northern and western regions (Figs. 15j,k,n,o). In the end-term climate, frequency of severe, extreme, and exceptional droughts is projected to decline across India, except for parts of Kashmir (Figs. 15d,h,l,p). Regionwise analysis shows that western and northern India are projected to experience frequent droughts under the projected climate (Table S5 in the supplemental material).

Fig. 15.

Frequency of severe, extreme, and exceptional soil moisture droughts during the (a),(e),(i),(m) historic base period (1961–90) and in the (b),(f),(j),(n) near- (2010–39); (c),(g),(k),(o) mid- (2040–69); and (d),(h),(l),(p) end-term (2070–99) projected climate. Drought frequency was estimated for each grid cell as the number of years when soil moisture percentiles fell below 10.

Fig. 15.

Frequency of severe, extreme, and exceptional soil moisture droughts during the (a),(e),(i),(m) historic base period (1961–90) and in the (b),(f),(j),(n) near- (2010–39); (c),(g),(k),(o) mid- (2040–69); and (d),(h),(l),(p) end-term (2070–99) projected climate. Drought frequency was estimated for each grid cell as the number of years when soil moisture percentiles fell below 10.

Despite projected increases in mean precipitation in all the selected crop-growing periods, droughts may become more frequent in the near- and midterm future climate, especially in the KHARIF_SOW period (Figs. 16a–d). In the KHARIF_SOW period, ensemble mean areal extents of droughts are projected to increase in the near- and midterm climate but to decrease in the end-term climate (Fig. 16e). In the KHARIF period, areal extent of drought is projected to decline in the end-term climate (Fig. 16f). In the RABI_SOW period, ensemble mean areal extents are projected to decline in the mid- and end terms (Fig. 16g). Areal extent of droughts were also projected to decline in the RABI period in the near-, mid-, and end-term climate (Fig. 16h). Projections of areal extents of droughts under the future climate exhibited large uncertainties, especially during the near and midterms (Figs. 16e–h); however, projected declines in areal extents during the end-term climate are robust.

Fig. 16.

Median frequency of severe, extreme, and exceptional droughts in the historic base period (1961–90), near- (2010–39), mid- (2040–69), and end-term (2070–99) projected climate for the (a) KHARIF_SOW, (b) KHARIF, (c) RABI_SOW, and (d) RABI periods. (e)–(h) Multimodel ensemble mean change (%) in the AE of severe, extreme, and exceptional droughts in the projected climate. Changes in the AE of droughts were estimated for the individual GCM and then mean of changes from all the GCMs was estimated. Error bars show intermodel variation (%) estimated using the std dev.

Fig. 16.

Median frequency of severe, extreme, and exceptional droughts in the historic base period (1961–90), near- (2010–39), mid- (2040–69), and end-term (2070–99) projected climate for the (a) KHARIF_SOW, (b) KHARIF, (c) RABI_SOW, and (d) RABI periods. (e)–(h) Multimodel ensemble mean change (%) in the AE of severe, extreme, and exceptional droughts in the projected climate. Changes in the AE of droughts were estimated for the individual GCM and then mean of changes from all the GCMs was estimated. Error bars show intermodel variation (%) estimated using the std dev.

4. Discussion and conclusions

Observed drying in the Gangetic Plain is consistent with previous studies (Bollasina et al. 2011; Mishra et al. 2012; Naidu et al. 2009). However, declining trends in precipitation and soil moisture over the Gangetic Plain are milder than those reported in Sheffield and Wood (2008a). The decline in precipitation in the Gangetic Plain could be attributed to several factors. For instance, Bollasina et al. (2011) attributed declines in precipitation to the human-induced increases in aerosol emissions. On the other hand, Mishra et al. (2012) reported a prominent mode of year-to-year variability associated with the sea surface temperature (SST) in the Indian Ocean that could lead to droughts in the Gangetic Plain. Frequency and areal extent of severe, extreme, and exceptional droughts increased during the period of 1950–2008. Moreover, in most of the crop-growing periods, frequency and areal extent of droughts are projected to increase in the near- (2010–39) and midterm (2040–69) climate. Droughts, therefore, may become more persistent and widespread in the presence of robust warming that is projected to occur across India. Projected increases in frequency and areal extent of droughts over India indicate that, under the projected climate, irrigation demands may increase, posing tremendous pressure on the groundwater resources that have already been depleting at a rapid pace (Rodell et al. 2009).

Despite the significant warming projected by the bias-corrected and downscaled data, projected increases in drought during the crop-growing seasons are not substantial. This may be associated with a significant increase in precipitation under the projected climate, which offsets the influence of increased warming on drought occurrence. However, in most of the crop-growing seasons, the frequency of severe, extreme, and exceptional droughts is projected to increase until the mid-twenty-first century. Trenberth et al. (2014) reported that increased heating from global warming may not cause droughts; however, it may contribute to the intensity and persistence of droughts under the projected climate. Therefore, an interaction of precipitation and temperature and its response on soil moisture is complex and needs to be better understood under the projected climate. Sheffield and Wood (2008b), using soil moisture data from CMIP3, reported that in global warming models show decreases in soil moisture globally in all scenarios, which is somewhat consistent with the findings of Dai (2012). On the other hand, our results, which are consistent with the findings of Raje et al. (2014), demonstrate increases in soil moisture under the projected climate in the majority of India in response to increased precipitation. Differences between our results and the findings of Sheffield and Wood (2008b) and Dai (2012) can be attributed to representation of land surface hydrological parameters in climate models. Soil moisture sensitivity to atmospheric forcing may be associated with the vegetation and soil parameterization in climate models, which may reflect higher sensitivity to global warming.

Precipitation and soil moisture demonstrate a strong seasonal cycle and coupling (Koster et al. 2004) in India as most rainfall occurs during the monsoon (JJAS) season. All-India-averaged soil moisture anomalies show a high persistence up to 4–6 months (Fig. 1). Moreover, the monsoon season precipitation affects soil moisture during the KHARIF season, and because of persistence, the high/low soil moisture remains in the RABI_SOW period (Table 1). These results further indicate the importance of precipitation in the monsoon season on agricultural droughts in India. For instance, a weak monsoon season may hamper agricultural productivity in the KHARIF season and may also lead to more irrigation demands in subsequent seasons. Moreover, persistent droughts in one crop-growing season may hamper crop production and water management in other seasons as well.

Agricultural production is sensitive to short- and long-term changes in climate (Lobell and Field 2007; Battisti and Naylor 2009). Precipitation, air temperature, and soil moisture are projected to increase up to 39.3%, 2.3°C, and 3.5%, respectively, in India by the end of the twenty-first century. The projected increase in air temperature was found to be robust in the near-, mid-, and end-term climate. On the other hand, increases in precipitation and soil moisture are not projected to become robust until the end-term climate. The influence of projected changes on agricultural production is somewhat uncertain (Lobell and Burke 2008); however, robust projected warming could be detrimental for crop production in India, especially in the RABI_SOW and RABI periods. Future work is needed to evaluate direct and indirect implications of changes in precipitation, soil moisture, and air temperature on water resources and agricultural production. However, observational trends and projected changes during the crop-growing periods will likely impose tremendous pressure on the water resources and agricultural production in India in the coming decades. The study also highlights the need for a continuous and dense soil moisture measurement network in India, which can be used for model evaluation and development and to understand the linkage between climate variability and climate change and soil moisture.

Based on the analysis, the following conclusions can be made:

  • Between 1950 and 2008, precipitation declined significantly in the Gangetic Plain (one of the most fertile regions in India), which in turn resulted in declines in soil moisture during the KHARIF_SOW and KHARIF periods. The Gangetic Plain showed significant declines in soil moisture during the RABI_SOW period. While changes in precipitation and soil moisture were largely insignificant at the all-India level, a significant warming was noticed in three out of four key crop-growing periods in India. In the KHARIF period, all-India-averaged temperature increased significantly (0.41°C, p = 0.0005) during the period 1950–2008. All-India-averaged temperature increased by 0.90° and 0.71°C in the RABI_SOW and RABI periods, respectively, during 1950–2008.

  • The model-simulated soil moisture successfully reproduced all the major drought events in India. The most detrimental droughts, which were persistent in all the crop-growing periods, occurred in 1972, 1987, and 2002. Out of the five most widespread droughts during 1950–2008, three or more occurred during recent decades. Moreover, in most of the crop-growing periods, increasing trends in the areal extent of droughts were found, suggesting that recent droughts were more widespread. In the crop-growing periods, the frequency of severe, extreme, and exceptional droughts increased substantially during the recent decades. In the Gangetic Plain and western India, an increase of about three to four droughts was noticed during 1980–2008 in comparison to 1951–79.

  • In all the crop-growing periods, precipitation, air temperature, and soil moisture are projected to increase under the future climate in the majority of India. All-India ensemble mean precipitation, air temperature, and soil moisture are projected to increase by 6%–39.3%, 0.9°–2.3°C, and 1.2%–3.5% during the crop-growing periods in the near- (2010–39), mid- (2040–69), and end-term (2070–99) climates, respectively. While the ensemble mean projected changes in air temperature are found to be robust in the near-, mid-, and end-term climate across India, changes in precipitation and soil moisture largely become robust in the end-term climate.

  • In the near- and midterm climate, the ensemble mean frequency of soil moisture drought is projected to increase in the KHARIF_SOW and KHARIF periods. However, during the RABI_SOW and RABI period, drought frequency is projected to decline in the majority of India, except the northern and western parts. In the end-term (2070–99) climate, drought frequency is projected to decline in the majority of India in all the key crop-growing periods. Droughts are projected to be widespread in the near- and midterm climate in the majority of the KHARIF_SOW crop-growing period. However, in the end-term climate, areal extents of projected droughts will decline because of prominent increases in precipitation.

Acknowledgments

This work was supported by the Varahamihir Ministry of Earth Science’s Fellowship to the first author. The work is undertaken as part of the ITRA Media Lab Asia project entitled “Measurement to Management (M2M): Improved Water Use Efficiency and Agricultural Productivity through Experimental Sensor Network.” The authors thank three anonymous reviewers for their valuable comments and suggestions.

REFERENCES

REFERENCES
Andreadis
,
K. M.
, and
D. P.
Lettenmaier
,
2006
:
Trends in 20th century drought over the continental United States
.
Geophys. Res. Lett.
,
33
, L10403, doi:.
Andreadis
,
K. M.
,
E. A.
Clark
,
A. W.
Wood
,
A. F.
Hamlet
, and
D. P.
Lettenmaier
,
2005
:
Twentieth-century drought in the conterminous United States
.
J. Hydrometeor.
,
6
,
985
1001
, doi:.
Battisti
,
D. S.
, and
R. L.
Naylor
,
2009
:
Historical warnings of future food insecurity with unprecedented seasonal heat
.
Science
,
323
,
240
244
, doi:.
Bollasina
,
M. A.
,
Y.
Ming
, and
V.
Ramaswamy
,
2011
:
Anthropogenic aerosols and the weakening of the South Asian summer monsoon
.
Science
,
334
,
502
505
, doi:.
Bürger
,
G.
,
T. Q.
Murdock
,
A. T.
Werner
,
S. R.
Sobie
, and
A. J.
Cannon
,
2012
:
Downscaling extremes—An intercomparison of multiple statistical methods for present climate
.
J. Climate
,
25
,
4366
4388
, doi:.
Burke
,
E. J.
, and
S. J.
Brown
,
2008
:
Evaluating uncertainties in the projection of future drought
.
J. Hydrometeor.
,
9
,
292
299
, doi:.
Burke
,
E. J.
,
S. J.
Brown
, and
N.
Christidis
,
2006
:
Modeling the recent evolution of global drought and projections for the twenty-first century with the Hadley Centre climate model
.
J. Hydrometeor.
,
7
,
1113
1125
, doi:.
Cayan
,
D. R.
,
E. P.
Maurer
,
M. D.
Dettinger
,
M.
Tyree
, and
K.
Hayhoe
,
2008
:
Climate change scenarios for the California region
.
Climatic Change
,
87
,
21
42
, doi:.
Cherkauer
,
K. A.
, and
D. P.
Lettenmaier
,
1999
:
Hydrologic effects of frozen soils in the upper Mississippi River basin
.
J. Geophys. Res.
,
104
,
19 599
19 610
, doi:.
Cherkauer
,
K. A.
,
L. C.
Bowling
, and
D. P.
Lettenmaier
,
2003
:
Variable infiltration capacity cold land process model updates
.
Global Planet. Change
,
38
,
151
159
, doi:.
Dai
,
A.
,
2011a
: Characteristics and trends in various forms of the Palmer drought severity index during 1900–2008. J. Geophys. Res.,116, D12115, doi:.
Dai
,
A.
,
2011b
:
Drought under global warming: A review
.
Wiley Interdiscip. Rev.: Climate Change
,
2
,
45
65
, doi:.
Dai
,
A.
,
2012
:
Increasing drought under global warming in observations and models
.
Nat. Climate Change
,
3
,
52
58
, doi:.
Demaria
,
E. M. C.
,
E. P.
Maurer
,
J.
Sheffield
,
E.
Bustos
,
D.
Poblete
,
S.
Vicuña
, and
F.
Meza
,
2013
:
Using a gridded global dataset to characterize regional hydroclimate in central Chile
.
J. Hydrometeor.
,
14
,
251
265
, doi:.
Dorigo
,
W. A.
, and Coauthors
,
2011
:
The International Soil Moisture Network: A data hosting facility for global in situ soil moisture measurements
.
Hydrol. Earth Syst. Sci.
,
15
, 1675–1698, doi:.
Dorigo
,
W. A.
,
R.
Jeu
,
D.
Chung
,
R.
Parinussa
,
Y.
Liu
,
W.
Wagner
, and
D.
Fernández-Prieto
,
2012
:
Evaluating global trends (1988–2010) in harmonized multi-satellite surface soil moisture
.
Geophys. Res. Lett.
,
39
, L18405, doi:.
Han
,
E.
,
V.
Merwade
, and
G. C.
Heathman
,
2012
:
Implementation of surface soil moisture data assimilation with watershed scale distributed hydrological model
.
J. Hydrol.
,
416–417
,
98
117
, doi:.
Hansen
,
J.
,
M.
Sato
,
R.
Ruedy
,
K.
Lo
,
D. W.
Lea
, and
M.
Medina-Elizade
,
2006
:
Global temperature change
.
Proc. Natl. Acad. Sci. USA
,
103
,
14 288
14 293
, doi:.
Hansen
,
M. C.
,
R. S.
DeFries
,
J. R.
Townshend
, and
R.
Sohlberg
,
2000
:
Global land cover classification at 1 km spatial resolution using a classification tree approach
.
Int. J. Remote Sens.
,
21
,
1331
1364
, doi:.
Hayhoe
,
K.
, and Coauthors
,
2004
. Emissions pathways, climate change, and impacts on California. Proc. Natl. Acad. Sci. USA,101,
12 422
12 427
, doi:.
Hollmann
,
R.
, and Coauthors
,
2013
:
The ESA climate change initiative satellite data records for essential climate variables
.
Bull. Amer. Meteor. Soc.
,
94
,
1541
1552
, doi:.
Huntington
,
T. G.
,
2006
:
Evidence for intensification of the global water cycle: Review and synthesis
.
J. Hydrol.
,
319
,
83
95
, doi:.
Jury
,
M. R.
, and
C.
Funk
,
2012
:
Climatic trends over Ethiopia: Regional signals and drivers
.
Int. J. Climatol.
, 33, 1924–1935, doi: .
Kalnay
,
E.
, and Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
, doi:.
Knutti
,
R.
, and
J.
Sedláček
,
2013
:
Robustness and uncertainties in the new CMIP5 climate model projections
.
Nat. Climate Change
,
3
,
369
373
, doi:.
Koster
,
R. D.
, and Coauthors
,
2004
:
Regions of strong coupling between soil moisture and precipitation
.
Science
,
305
,
1138
1140
, doi:.
Krishna Kumar
,
K.
,
R. K.
Kumar
,
R. G.
Ashrit
,
N. R.
Deshpande
, and
J. W.
Hansen
,
2004
:
Climate impacts on Indian agriculture
.
Int. J. Climatol.
,
24
,
1375
1393
, doi:.
Levine
,
R. C.
,
A. G.
Turner
,
D.
Marathayil
, and
G. M.
Martin
,
2013
:
The role of northern Arabian Sea surface temperature biases in CMIP5 model simulations and future projections of Indian summer monsoon rainfall
.
Climate Dyn.
,
41
,
155
172
, doi:.
Liang
,
X.
,
D. P.
Lettenmaier
,
E. F.
Wood
, and
S. J.
Burges
,
1994
: A simple hydrologically based model of land surface water and energy fluxes for general circulation models. J. Geophys. Res.,99, 14 415–14 428, doi:.
Liang
,
X.
,
E. F.
Wood
, and
D. P.
Lettenmaier
,
1996
:
Surface soil moisture parameterization of the VIC-2L model: Evaluation and modification
.
Global Planet. Change
,
13
,
195
206
, doi:.
Lobell
,
D. B.
, and
C. B.
Field
,
2007
:
Global scale climate–crop yield relationships and the impacts of recent warming
.
Environ. Res. Lett.
,
2
,
014002
, doi:.
Lobell
,
D. B.
, and
M. B.
Burke
,
2008
:
Why are agricultural impacts of climate change so uncertain? The importance of temperature relative to precipitation
.
Environ. Res. Lett.
,
3
,
034007
, doi:.
Lobell
,
D. B.
,
J. I.
Ortiz-Monasterio
,
A. M.
Sibley
, and
V. S.
Sohu
,
2013
:
Satellite detection of earlier wheat sowing in India and implications for yield trends
.
Agric. Syst.
,
115
,
137
143
, doi:.
Mann
,
H. B.
,
1945
:
Nonparametric tests against trend
.
Econometrica
,
13
,
245
259
, doi:.
Maurer
,
E. P.
, and
H. G.
Hidalgo
,
2008
:
Utility of daily vs. monthly large-scale climate data: An intercomparison of two statistical downscaling methods
.
Hydrol. Earth Syst. Sci.
,
12
,
551
563
, doi:.
Maurer
,
E. P.
,
A. W.
Wood
,
J. C.
Adam
,
D. P.
Lettenmaier
, and
B.
Nijssen
,
2002
:
A long-term hydrologically based dataset of land surface fluxes and states for the conterminous United States
.
J. Climate
,
15
,
3237
3251
, doi:.
Maurer
,
E. P.
,
H. G.
Hidalgo
,
T.
Das
,
M. D.
Dettinger
, and
D. R.
Cayan
,
2010
:
The utility of daily large-scale climate data in the assessment of climate change impacts on daily streamflow in California
.
Hydrol. Earth Syst. Sci.
,
14
,
1125
1138
, doi:.
Menon
,
A.
,
A.
Levermann
,
J.
Schewe
,
J.
Lehmann
, and
K.
Frieler
,
2013
: Consistent increase in Indian monsoon rainfall and its variability across CMIP-5 models. Earth Syst. Dyn.,4, 287–300, doi:.
Mishra
,
V.
, and
K. A.
Cherkauer
,
2010
:
Retrospective droughts in the crop growing season: Implications to corn and soybean yield in the Midwestern United States
.
Agric. For. Meteor.
,
150
,
1030
1045
, doi:.
Mishra
,
V.
, and
D. P.
Lettenmaier
,
2011
:
Climatic trends in major U.S. urban areas, 1950–2009
.
Geophys. Res. Lett.
,
38
, L16401, doi:.
Mishra
,
V.
,
K. A.
Cherkauer
, and
S.
Shukla
,
2010
:
Assessment of drought due to historic climate variability and projected future climate change in the Midwestern United States
.
J. Hydrometeor.
,
11
,
46
68
, doi:.
Mishra
,
V.
,
B. V.
Somalik
,
D. P.
Lettenmaier
, and
J. M.
Wallace
,
2012
:
A prominent pattern of year-to-year variability in Indian summer monsoon rainfall
.
Proc. Natl. Acad. Sci. USA
, 109, 7213–7217, doi:.
Naidu
,
C. V.
,
K.
Durgalakshmi
,
K.
Muni Krishna
,
S.
Ramalingeswara Rao
,
G. C.
Satyanarayana
,
P.
Lakshminarayana
, and
L.
Malleswara Rao
,
2009
: Is summer monsoon rainfall decreasing over India in the global warming era? J. Geophys. Res.,114, D24108, doi:.
Nijssen
,
B.
,
G. M.
O’Donnell
,
A. F.
Hamlet
, and
D. P.
Lettenmaier
,
2001
:
Hydrologic sensitivity of global rivers to climate change
.
Climatic Change
,
50
,
143
175
, doi:.
Pai
,
D. S.
,
L.
Sridhar
,
M. R.
Badwaik
, and
M.
Rajeevan
2014
:
Analysis of the daily rainfall events over India using a new long period (1901–2010) high resolution (0.25° × 0.25°) gridded rainfall data set
.
Climate Dyn.
, doi:, in press.
Raje
,
D.
, and
R.
Krishnan
,
2012
:
Bayesian parameter uncertainty modeling in a macroscale hydrologic model and its impact on Indian river basin hydrology under climate change
.
Water Resour. Res.
,
48
, W08522, doi:.
Raje
,
D.
,
P.
Priya
, and
R.
Krishnan
,
2014
:
Macroscale hydrological modelling approach for study of large scale hydrologic impacts under climate change in Indian river basins
.
Hydrol. Processes
,
28
,
1874
1889
, doi:.
Robock
,
A.
,
K. Y.
Vinnikov
,
G.
Srinivasan
,
J. K.
Entin
,
S. E.
Hollinger
,
N. A.
Speranskaya
,
S.
Liu
, and
A.
Namkhai
,
2000
:
The global soil moisture data bank
.
Bull. Amer. Meteor. Soc.
,
81
,
1281
1300
, doi:.
Rodell
,
M.
, and Coauthors
,
2004
:
The Global Land Data Assimilation System
.
Bull. Amer. Meteor. Soc.
,
85
,
381
394
, doi:.
Rodell
,
M.
,
I.
Velicogna
, and
J. S.
Famiglietti
,
2009
:
Satellite-based estimates of groundwater depletion in India
.
Nature
,
460
,
999
1002
, doi:.
Sen
,
P. K.
,
1968
:
Estimates of the regression coefficient based on Kendall’s tau
.
J. Amer. Stat. Assoc.
,
63
,
1379
1389
, doi:.
Seneviratne
,
S. I.
,
D.
Lüthi
,
M.
Litschi
, and
C.
Schär
,
2006
:
Land–atmosphere coupling and climate change in Europe
.
Nature
,
443
,
205
209
, doi:.
Seneviratne
,
S. I.
, and Coauthors
,
2012
: Changes in climate extremes and their impacts on the natural physical environment. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation, C. B. Field et al., Eds., Cambridge University Press, 109–230.
Sheffield
,
J.
, and
E. F.
Wood
,
2007
: Characteristics of global and regional drought, 1950–2000: Analysis of soil moisture data from off-line simulation of the terrestrial hydrologic cycle. J. Geophys. Res.,112, D17115, doi:.
Sheffield
,
J.
, and
E. F.
Wood
,
2008a
: Global trends and variability in soil moisture and drought characteristics, 1950–2000, from observation-driven simulations of the terrestrial hydrological cycle.
J. Climate
,
21
,
432
458
, doi:.
Sheffield
,
J.
, and
E. F.
Wood
,
2008b
:
Projected changes in drought occurrence under future global warming from multi-model, multi-scenario, IPCC AR4 simulations
.
Climate Dyn.
,
31
,
79
105
, doi:.
Sheffield
,
J.
,
G.
Goteti
,
F.
Wen
, and
E. F.
Wood
,
2004
: A simulated soil moisture based drought analysis for the United States. J. Geophys. Res.,109, D24108, doi:.
Sheffield
,
J.
,
G.
Goteti
, and
E. F.
Wood
,
2006
:
Development of a 50-year high-resolution global dataset of meteorological forcings for land surface modeling
.
J. Climate
,
19
,
3088
3111
, doi:.
Sheffield
,
J.
,
E. F.
Wood
, and
M. L.
Roderick
,
2012
:
Little change in global drought over the past 60 years
.
Nature
,
491
,
435
438
, doi:.
Sivakumar
,
M. V. K.
,
H. P.
Das
, and
O.
Brunini
,
2005
:
Impacts of present and future climate variability and change on agriculture and forestry in the arid and semi-arid tropics
.
Climatic Change
,
70
,
31
72
, doi:.
Svoboda
,
M.
, and Coauthors
,
2002
: The Drought Monitor. Bull. Amer. Meteor. Soc.,83, 1181–1190.
Taylor
,
K. E.
,
R. J.
Stouffer
, and
G. A.
Meehl
,
2009
: A summary of the CMIP5 experiment design. WCRP CMIP5 Doc., 33 pp. [Available online at http://cmip-pcmdi.llnl.gov/cmip5/docs/Taylor_CMIP5_design.pdf.]
Taylor
,
K. E.
,
R. J.
Stouffer
, and
G. A.
Meehl
,
2012
:
An overview of CMIP5 and the experiment design
.
Bull. Amer. Meteor. Soc.
,
93
, 485–498, doi:.
Tebaldi
,
C.
,
J. M.
Arblaster
, and
R.
Knutti
,
2011
:
Mapping model agreement on future climate projections
.
Geophys. Res. Lett.
,
38
, L23701, doi:.
Thrasher
,
B.
,
E. P.
Maurer
,
C.
McKellar
, and
P. B.
Duffy
,
2012
:
Technical note: Bias correcting climate model simulated daily temperature extremes with quantile mapping
.
Hydrol. Earth Syst. Sci.
,
16
,
3309
3314
, doi:.
Trenberth
,
K. E.
,
A.
Dai
,
G.
van der Schrier
,
P. D.
Jones
,
J.
Barichivich
,
K. R.
Briffa
, and
J.
Sheffield
,
2014
:
Global warming and changes in drought
.
Nat. Climate Change
,
4
,
17
22
, doi:10.1038/NCLIMATE2067.
Turner
,
A. G.
, and
H.
Annamalai
,
2012
:
Climate change and the South Asian summer monsoon
.
Nat. Climate Change
,
2
,
587
595
, doi:.
Wang
,
A.
,
D. P.
Lettenmaier
, and
J.
Sheffield
,
2011
:
Soil moisture drought in China, 1950–2006
.
J. Climate
,
24
,
3257
3271
, doi:.
Wilhite
,
D. A.
,
M. J.
Hayes
,
C.
Knutson
, and
K. H.
Smith
,
2000
:
Planning for drought: Moving from crisis to risk management
.
J. Amer. Water Resour. Assoc.
,
36
,
697
710
, doi:.
Wood
,
A. W.
,
E. P.
Maurer
,
A.
Kumar
, and
D.
Lettenmaier
,
2002
: Long-range experimental hydrologic forecasting for the eastern United States. J. Geophys. Res.,107, 4429, doi:.
Wood
,
A. W.
,
L. R.
Leung
,
V.
Sridhar
, and
D. P.
Lettenmaier
,
2004
:
Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs
.
Climatic Change
,
62
,
189
216
, doi:.
Yue
,
S.
, and
C. Y.
Wang
,
2002
:
Applicability of prewhitening to eliminate the influence of serial correlation on the Mann–Kendall test
.
Water Resour. Res.
,
38
, 1068, doi:.

Footnotes

*

Supplemental information related to this paper is available at the Journals Online website: http://dx.doi.org/10.1175/JHM-D-13-0177.s1.

Supplemental Material