Substantial decline in atmospheric aridity due to irrigation in India

We obtained the static irrigated area map from the Food and Agriculture Organization (FAO; Siebert et al 2013, 2015a). The irrigated area from FAO was updated recently and provided the irrigation extent of 2013. FAO irrigated area maps may differ from the actual area under irrigation due to the unaccounted cropping pattern and irrigated secondary crops (Ajaz et al 2019). The application of FAO irrigated area maps is more common in the climate and land surface model simulations as the dataset provides fractional information of the irrigated area in each grid (Ambika et al 2016). The datasets used in our study primarily cover the Indian landmass (Latitude: 5–40° N; Longitude: 65–110° E). We used the FAO version 5.0 irrigated area map that is available at a resolution of 0.083° (∼8–9 km approximately). We resampled the irrigated area from FAO to make it consistent with the WRF resolution (0.25°). The historic irrigation fraction (Historical Irrigation Datasets–HID; Siebert et al 2015a, 2015b) was used to evaluate the fractional increase in the irrigated area over a selected period. The HID is the newly developed irrigation map which represents the area equipped for irrigation (AEI) spanning for the period 1900–2015. HID maps are available at five arcmin resolution and use sub-national level statistics with different area extent of pasture and cropland (Siebert et al 2015a)

Land surface temperature (LST) at 4 km spatial resolution was obtained from National Oceanic and Atmospheric Administration (NOAA) Star Centre for Satellite Application and Research (NSTAR) from 1982 to 2018 (Yu et al 2017). We obtained solar-induced chlorophyll fluorescence (SIF, Li and Xiao 2019) for the 2000–2016 period, which represents plant photosynthetic activity (Jonard et al 2020). We examined the observed changes in LST and SIF during the crop growing season (November–February). Irrigation water use during the major crop growing season (November–March) is higher in comparison to the other seasons over the Indo-Gangetic Plain (Huang et al 2018). Therefore, we estimated the role of irrigation on atmospheric aridity during the growing season in India.

We estimated changes in leaf area index (LAI), normalized difference vegetation index (NDVI), and fraction of absorbed photosynthetic active radiation (FPAR) obtained from NOAA's advanced very-high-resolution radiometer (AVHRR) satellite dataset (Loveland et al 2000). The net and gross primary productivity (GPP) were obtained from the gross and dry matter products (GDMP and DMP) of Copernicus Global Land Operations Vegetation and Energy (CGLOPS-1; Buchhorn et al 2017) for the 1999–2018 period. The DMP and GDMP were adequately scaled to estimate net primary productivity (NPP) and GPP for the growing period. We calculated the autotrophic respiration (AR), which is the contribution of CO₂ respired by plant roots (Bond‐Lamberty et al 2004). AR was estimated using the difference between GPP and NPP. The changes in the selected variables were estimated after interpolating them to 0.25° using nonparametric Mann–Kendall trend test (Mann 1945) and Sen's (Sen 1968) slope method. The growing season LAI is considered as a proxy of vegetation health (Chen et al 2019, Wang et al 2019). Daily gridded precipitation was obtained from the India Meteorological Department (IMD), which is available at 0.25° (Pai et al 2014). ERA-5 reanalysis data (Dee et al 2011) was obtained from the Copernicus climate data storage. ERA-5 dataset is available at 31 km spatial resolution and hourly temporal resolution. We used hourly ERA-5 data of air temperature, relative humidity (RH), and SM to derive hourly VPD for the 1979–2018 period. Daily VPD was calculated using the methodology based on Ficklin and Novick (2017). We used the difference between saturation and actual vapor pressures for estimating the VPD from the ERA-5 reanalysis data. The saturation and actual vapor pressures were calculated based on the method given in Allen et al (1998). The RH used for the estimation of actual vapor pressure was obtained using the Tetens's formula (Tetens 1930) with the parameters based on saturation over water (Buck 1981).

The WRF 4.0 is a non-hydrostatic compressible model (Chen et al 2011). The WRF is a state-of-the-art weather forecasting tool used for operational forecast, real-time prediction, regional climate, and coupled-model application (Skamarock and Klemp 2019). The WRF multi-scheme mesoscale process was designed for numerical weather prediction; however, the model can be used for weather prediction and to understand regional climate dynamics (Chen et al 2011). The WRF can simulate regional and local scale hydrometeorological processes by dynamically linking land surface process, planetary boundary layer, cumulus development, advection of atmospheric moisture, and feedback from the energy and radiation budget at the surface layer (Sridhar 2013). The model follows a terrain-following hybrid-sigma vertical pressure coordinate system that extends from the surface to 50 hPa with constant pressure at the top (Harding and Snyder 2012).

We conducted simulations using WRF version 4.0 with ERA-5 reanalysis as the boundary condition for 1979–2018 period. The WRF simulations were conducted with the irrigation on and off scenarios to examine the influence of irrigation on SM and atmospheric aridity. The irrigation-on scenario represents the addition of required water to the soil in agricultural regions. Under the irrigation-off scenario, the model does not provide additional water to the soil. Our WRF simulations used the NOAH land surface model to simulate latent heat, sensible heat, and land surface temperature (Mitchell 2005). Terrestrial radiation processes in WRF are estimated using the Rapid Radiative transfer Model (RRTMG) scheme (Iacono et al 2008). We adopted the Kain–Fritsch (KF) scheme (Kain 2004) for convective parameterization to achieve a realistic pattern and extent of precipitation (Qian et al 2013). Also, we used Mellor–Yamada–Janjic (MYJ) scheme for the planetary boundary physical mechanism (Janjić 1994). Surface variables including 2 m temperature, 10 meter wind, mean sea level pressure, sea surface temperature, land surface temperature, snow density, snow depth, soil temperature at different levels, surface pressure, and volumetric soil water from ERA-5 were used to conduct the control (irrigation-off) simulations at 0.25° spatial resolution. Moreover, we used pressure level variables, which include specific humidity (SH), temperature, U and V wind direction components (U-positive for the west to east flow, and V-positive for the south to north flow), and geopotential height from ERA-5. The irrigation scheme implemented in the WRF model starts applying water when the root-zone SM falls below the field capacity (Kain 2004, Qian et al 2013, Yang et al 2019). More details on the irrigation scheme can be obtained from the previous studies (Ozdogan et al 2010, Yang et al 2019).

The WRF simulations in control (no-irrigation) scenarios were compared with ERA-5 datasets. We estimated the standardized SM index (SSI, Xu et al 2018) using the WRF simulated SM under the irrigation on and off scenarios to evaluate drought condition for the 1979–2018 period. We compared SH, air temperature, and VPD from the WRF simulations against the ERA-5 reanalysis under the control scenarios. Moreover, we examined the role of irrigation on atmospheric aridity (VPD) and drought using WRF simulations.

The net irrigated area has significantly increased in India after the 1950s (figures S1(a) and (b) are available online at stacks.iop.org/ERL/15/124060/mmedia). The Indo-Gangetic Plain is one of the most extensively irrigated regions in India (figure S1(a)). The Indo-Gangetic Plain witnessed changes in LST and vegetation growth during the past few decades (figure 1(a)), which might be partly contributed by the irrigation expansion (figures 1 and S1). We found that LST has considerably (0.4–1.7 °C) declined during the growing season over the Indo-Gangetic Plain and in other parts of India (figure 1(b)). This significant cooling of 0.8 °C (P-value = 0.03) over the Indo-Gangetic Plain in satellite-based LST is associated with the irrigation expansion (Cook et al 2011, 2015, Thiery et al 2017, Ambika and Mishra 2019). On the other hand, a substantial increase in SIF was found over the Indo-Gangetic Plain and in different regions of India during 2000–2018 (Asoka and Mishra 2015, Chen et al 2019). For instance, SIF averaged over the Indo-Gangetic Plain has increased (change = 2.81 mW m⁻² nm⁻¹ sr⁻¹; here SIF unit is the radiance of a surface per unit wavelength) significantly (P-value = 0.00001) during 2000–2016 period (figure 1(e)). Greening in vegetation (Asoka and Mishra 2015, Chen et al 2019) and cooling in LST (Ambika and Mishra 2019) show the influence of irrigation over the intensively irrigated Indo-Gangetic Plain (figures 1(b) and (c)).

Zoom In Zoom Out Reset image size

Next, we estimated the changes in NDVI, LAI, and FPAR during the 1982–2018 period (figure S2). We found a greening trend across India (figure S2). However, the increase in NDVI, LAI, and FPAR was more prominent and significant (P-value = 0.0001) over the Indo-Gangetic Plain during the 1982–2018 period (figure S2). This significant rise in greening over the Indo-Gangetic Plain can be partly attributed to the increase in an irrigated area in the region (Ambika et al 2016, Ambika and Mishra 2019, Thiery et al 2020) (figure S1(b)). Other human land management practices (e.g. fertilizer application, improved seed varieties) can also contribute to greening over the Indo-Gangetic Plain (Schmidhuber and Tubiello 2007, Brown and Funk 2008). In addition to greening trend, we examined the changes in GPP and NPP, and AR during the recent period (2000–2018) based on the availability of the satellite datasets. GPP, NPP, and AR have significantly (P-value < 0.05) increased over the Indo-Gangetic Plain (figure S3). Greening trends and an increase in the GPP and NPP further confirmed the potential role of irrigation over the Indo-Gangetic Plain. However, the land surface cooling due to irrigation is attributed to an increase in evapotranspiration (Shah et al 2019) and latent heat flux (Seneviratne et al 2010). Besides, greening and increase in NPP indicate the modulation of SM deficit due to irrigation (Ozdogan 2011).

Low SM and high atmospheric vapor pressure deficit (VPD) are considered as the two main stressors on ecosystem productivity during drought (Zhou et al 2019b). VPD is a measure of atmospheric aridity and drives evapotranspiration (Zhou et al 2019a). We estimated changes in VPD, RH, and SM from ERA-5 reanalysis for 1979–2018 period (figure 2). We find that the intensive irrigation over the Indo-Gangetic Plain has resulted in cooling and greening (figure 1). VPD from ERA-5 reanalysis has significantly (P-value = 0.00002) declined (−1.38 kPa) during the growing season over the Indo-Gangetic Plain for 1979–2018 period (figures 2(a) and (b)). The decline in atmospheric aridity and the increase in vegetation health can be attributed to increased soil moisture (figures 2(c) and (d)). In addition, during irrigation, the near-surface temperature is expected to be cooler and moist static energy is higher near the middle to low boundary layer (Qian et al 2013). Moreover, RH in the irrigated regions is higher than the non-irrigated regions (Qian et al 2013). We find an increase in RH from ERA-5 during the growing season across India, which can be attributed to irrigation and the warming climate (Sherwood and Huber 2010, Fischer and Knutti 2013) (figure 2(c)). However, the increase (2.05%) in RH over the Indo-Gangetic Plain was more prominent (P-value = 0.0052). As expected, SM over the Indo-Gangetic Plain has significantly (P-value < 0.05) increased (1.63 m³ m⁻³) during 1979–2018, indicating the potential role of increased irrigation in the region (figures 2(e) and (f)).

Zoom In Zoom Out Reset image size

Our analysis based on ERA-5 reanalysis shows that irrigation over the Indo-Gangetic Plain modulates both SM and atmospheric aridity by increasing RH. The increase in RH is due to increased evapotranspiration and latent heat flux (Seneviratne et al 2010, Zhou et al 2019a). Since ERA-5 reanalysis assimilates observations, it partly captures the influence of irrigation on land and atmospheric processes (Zohaib and Choi 2020). Our results show that VPD–RH (correlation = −0.95) and VPD–SM (correlation = −0.84) are negatively correlated over the Indo-Gangetic Plain (figure S4) indicating that increased soil moisture due to irrigation results in increased RH (Kueppers et al 2007), which in turn, reduces atmospheric aridity. Since precipitation during the growing season is not sufficient to meet the crop water requirements, irrigation plays a vital role in food production over the Indo-Gangetic Plain (figure S5). The estimated anomaly composites of VPD, RH, and SM show a reduction in atmospheric aridity and an increase in RH and SM during drought over the Indo-Gangetic Plain (figure S6). Since LAI is negatively associated with atmospheric aridity while positively related to RH (figure S7), decreased VPD and increased RH have supported greening in India.

A significant increase in VPD stimulates high evaporative demand, reduces SM, and results in enhanced heating and drying in the lower level atmosphere (Lansu et al 2020). However, increased SM due to irrigation in the Indo-Gangetic Plain alleviates exacerbation of atmospheric aridity during droughts (Cook et al 2011, Thiery et al 2017, 2020). Atmospheric circulation plays a substantial role in VPD–SM coupling through blocking (Pfahl et al 2015, Rodrigues and Woollings 2017), tropospheric warming, and subsidence (Helama et al 2009). Our results based on ERA-5 reanalysis show that irrigation influences land–atmospheric interaction and modulates SM drought and atmospheric aridity. Moreover, modulation in SM by irrigation can influence atmospheric circulation (Koster et al 2016, Zhou et al 2019a).

Irrigation considerably influenced greening and cooling during the growing season over the Indo-Gangetic Plain, as shown by the satellite-based observations (figures 1, 2 and S2, S3). Moreover, ERA-5 reanalysis showed a substantial increase in SM while a decline in atmospheric aridity in the region during 1979–2018. We used WRF simulations for causal attribution of the land–atmospheric feedback to further diagnose the role of irrigation on soil drought and atmospheric aridity (figure 3). The WRF simulations for the control (irrigation-off) and irrigation-on scenarios were conducted using ERA-5 reanalysis as a boundary condition. The control simulations from the WRF model capture the spatial variability in SH, air temperature, and VPD reasonably well for the 1979–2018 period (figure S8).

Zoom In Zoom Out Reset image size

We estimated SSI (Xu et al 2018) using the root-zone (30 cm) SM to examine the role of irrigation during droughts (figure 3(a)). The three major growing season droughts that occurred in 1980, 1988, and 1995 were identified to examine the role of irrigation using the WRF simulations (figure 3(a)). Positive surface temperature (∼1 K) and VPD (∼0.4 kPa) anomalies persist during the growing season drought across north India under control (no-irrigation) simulations (figures 3(b), (d) and S9). Increased atmospheric aridity depletes soil moisture (∼−0.1 m³ m⁻³), leading to low RH (5–10%) under the growing season droughts (figures 3(c) and (e)). We used WRF simulations with the irrigation-on scenario to examine the modulation of soil drought and atmospheric aridity by irrigation. Consistent with the satellite observations and ERA-5 reanalysis, irrigation cools down the increased land surface temperature caused by droughts (figure 3(f)). Similarly, irrigation during the growing season enhances SM and RH over the Indo-Gangetic Plain (figures 3(g) and (i)). Due to increased SM and RH, the atmospheric aridity declines during the growing season droughts (figure 3(h)). Moreover, irrigation during droughts results in increased sea level pressure and reduced planetary boundary layer (figure S10). These results further show the modulation of soil drought and atmospheric aridity due to irrigation, which plays a crucial role in triggering short- or long-term droughts (Zhou et al 2019a, Pendergrass et al 2020).

Next, we examined the role of irrigation in alleviating the growing season atmospheric aridity and SM drought using WRF simulations for irrigation-on and off scenarios during 1979–2018 (figure 4). We estimated the difference in VPD, SM, SH, and 2 m air temperature under the irrigation-on and irrigation-off (irrigation-on–irrigation-off) scenarios (figure 4). The difference in climatological mean VPD, SM, surface temperature, and SH for the growing season showed a considerable influence of irrigation across India depending on the irrigated area fraction (figure 4). However, we noted a more prominent influence of irrigation over the Indo-Gangetic Plain (figure 4(a)). The difference in the irrigation-on and off scenarios from the WRF showed the cooling of air temperature (0.5–1.0 °C), increase in SM (0.7–1.0 m³ m⁻³) and SH, and a reduction in atmospheric aridity (0.3–0.5 kPa) (figures 4(a)–(d)).

Zoom In Zoom Out Reset image size

We analyzed the surface energy budget from the WRF simulations under irrigation on and off scenarios to understand the physical mechanism behind the modulation of soil moisture drought and atmospheric aridity by irrigation (figures 5 and S11). Irrigation causes an increase in latent heat flux (Marcella and Eltahir 2014), reduction in sensible heat flux (figures 5 and S11). Increase in latent heat is compensated by the decrease in sensible heat, leading to cooling (figure 5(d)). Irrigation alters the partitioning of energy budget over the land. For instance, Bowen's ratio, which is the ratio of sensible and latent heat flux, declines significantly (p < 0.05, figures 5(d) and (h)) due to irrigation. Therefore, the cooling over the Indo-Gangetic Plain can be attributed to changes in the land surface energy budget (figures 5(e) and (f)), which results in reduced sensible heat flux (figure S11). The increased evapotranspiration and humidity cause reduced atmospheric aridity (VPD). Atmospheric aridity is reduced by the increase in moist-enthalpy in the lower atmosphere due to the reduction in planetary boundary layer height (figures S10 and S11) (Gentine et al 2013). Overall, we found that irrigation over the Indo-Gangetic Plain modulates SM drought and atmospheric aridity through the partitioning of the land surface energy budget.

Zoom In Zoom Out Reset image size