A SWAT-Copula based approach for monitoring and assessment of drought propagation in an irrigation command

Since drought is a natural disaster that causes significant loss to society and the agrarian economy, detection of drought period is very important for different drought management strategies involving irrigation scheduling, canal scheduling, and reservoir operation. For estimation of drought, although there are a number of drought indices, viz., univariate Standardized Precipitation Index (SPI), Standardized Precipitation Evapotranspiration Index (SPEI), Palmer Drought Severity Index (PDSI), and Soil Moisture Stress Index (SSI) existing in the literature, they are not fully effective for early warning and onset detection of drought events due to the scarce-knowledge of interdependency among the drought-causing factors. As an indirect method, the hydrological Soil Water Assessment Tool (SWAT) alone cannot be used for drought assessment. Hence, in this study, the SWAT is coupled with the multivariate copulas to effectively predict the drought years. The developed approach is tested in the Kangsabati River basin (12,014 km²) having about 48% of the area under paddy land use. The SWAT model is set up for the study area at daily time-scale with the Nash-Sutcliffe Criterion (ENS) of 0.64 and R² of 0.64 during the calibration period. The SPI and Standardized Runoff Index (SRI) are calculated for all the selected areas at four time-scales of 3-months, 6-months, 9-months and 12-months using the time series data from 1982–2010. The correlation analysis between SPI and SRI showed that the onset time for hydrological drought after the occurence of meteorological drought is three months. For identification of drought years, the multivariate distribution of SPI and SSI was constructed using the members of Archimedean copula family. Out of all the members of the copula family, Frank copula gave the best joint probability distribution function. The newly developed Multivariate Standardized Drought Index (MSDI) is capable of determining drought years with very good accuracy. Hence, the results reveal that the developed SWAT-Copula based approach has the potential to be implemented in data-scarce regions for effective drought monitoring with the minimum observed inputs.