Investigating water stress using hydrometeorological and remote sensing data

Abstract

Soil moisture investigation at point scale, hydrological modeling for studying cropping pattern scenario and meteorological and agriculture drought conditions were investigated in Jamwa Ramgarh catchment. Soil moisture measurements were taken at agriculture plots in Kharif and part of the Rabi season for year 2020- 21. Distinct soil moisture peaks were observed during heavy precipitation and flood irrigation. More significant change was observed in upper 20 cm during flood irrigation than for lower horizons. Soil moisture during Kharif season at few sites and for drip irrigation remained nearly constant. Moreover, non-parametric methods i.e., Mann-Kendall (MK) and Sen’s slope estimator were used to quantify the spatial and temporal trends at seasonal and annual scale in rainfall during 1980 to 2017 at 17 rain gauge stations. These monsoon months’ trends are not significant at 95% confidence interval, except for Amber station in month of June, which showed significant increasing trend. Further, the unsaturated zone Mike SHE model was run with FAO fine sand, silt loam and sandy laom with sub soil parameters for the sites at Gopalgarh and Roda Nadi. Unsaturated zone model simulation was done for cropping pattern pearl millet- fallow and pearl millet- wheat with two and six irrigation scenario. In two irrigation scenario, reduction of nearly 50% in evapotranspiration was simulated compared to six irrigation scenario. The Rabi season water demand was nearly equal to groundwater recharge. Thus, pearl millet- wheat cropping patter with six irrigations is not sustainable at point scale. Yearly drought magnitude was estimated for 15 stations in the region. Average drought magnitude for the Pan India severe drought years 1987 and 2002 was more than three. For years 1979, 1982, 1986, 1987, 1991, 2002, 2004, 2006, 2009 and 2017, some of which were Pan India major drought years, the average DM was more than two. Average probability of yearly drought magnitude and maximum dry spell weeks were also estimated. For Pan India drought years 1974, 1979, 1982, 1985- 1987, 2002, non- exceedance probability threshold was 70% in either yearly drought magnitude or maximum dry spell weeks. Vegetation Condition Index (VCI) was estimated using MOD13Q1 16 day NDVI. A subset of NDVI data was processed. The data were processed using R- software. Based on reliability raster, cloud and less reliable pixels were removed, the gaps were filled, trend in NDVI data was removed, seasonal data were extracted, VCI was estimated and average VCI for the area was computed. Relationship between average VCI and average probability of DM and dry spell week had small coefficient of determination. This indicated that variables other the precipitation effects the crop vigour. Similarly the satellite imagery of IRS LISS III were also used to derive the VCI time series for different differet vegetation patches and cooreelated with the DM. The comparision shows the similar results from MODIS and LISS dataseries. Different combinations of clusters were formed and forcefully divided into 2, 3 and 4 classes for regionalization of the catchment . On the basis of these combinations it was revealed that cluster distribution of the variables is almost of similar type and pattern. It is observed that the various combinations of clusters are spatially and statistically correlated with the drought magnitude indicating hydrologically similarity. Moreover, the catchment is under the hydrometerological subzone 1b, hence regionality may not be the cause of drought like situation in the area.The physically distributed model MIKESHE, coupled with the hydraulic MIKE 11 model has been used to simulate the discharge in the Jamwa Ramgarh catchment. The model is simulated for 44 years (1974-2017) but due to lack of observed data its calibrated for 10 years (1974-1983) and validated for 22 years (1984-2005) respectively, considering different efficiency criteria between observed and simulated discharge namely coefficient of determination (R2) and Nash-Sutcliffe efficiency (NSE) to analyze the predictability of the simulated discharge.During the calibration R2 0.8 and NSE -0.68 was achieved wheras during validation, the efficiency criteria (R2 0.65 & NSE -0.74). Nash coefficient is poor being negative and is due to over estimation of the volume. So further modeling and observation were required to satisfying the modeling needs.