Sahoo, DP; Sahoo, B; Tiwari, MK (2020). Copula-based probabilistic spectral algorithms for high-frequent streamflow estimation. REMOTE SENSING OF ENVIRONMENT, 251, 112092.

The recent increase in population growth and urbanisation demands for real-time riverine water management to fulfil the daily-scale domestic and ecological water needs, necessitating for high-frequent streamflow estimation at any river section. Although there are several conventional methods available in the literature, using these methods to obtain streamflow information at finer spatiotemporal resolutions is not feasible. Moreover, streamflow estimation using single or multi-satellite remote sensing approaches is still in the experimental stage. As advancement in the existing approaches, this study advocates two novel models, namely CMOD and CFUS, which use the Frank copula-based single MODIS satellite data and copula-based multi-satellite MODIS-Landsat fusion data, respectively. These developed models fit a box-centre matrix to the pixel ratios of water (within the river at the gauging station) and land (in the riparian zone) in the near-infrared spectrum. Two other approaches using the stand-alone MODIS data (MOD model) and enhanced spatiotemporal fusion of MODIS-Landsat datasets (FUS model) are also developed to inter-compare with the CMOD and CFUS models for reproducing the daily time series of streamflow at three gauging stations on the Brahmani River in eastern India. The calibration and validation results reveal that the 30 m resolution CFUS model is the best approach for daily-scale streamflow estimation with sufficient accuracy with an average Nash-Sutcliffe Coefficient (NSC) of 0.92 followed by the CMOD (NSC = 0.80), FUS (NSC = 0.78), and MOD (NSC = 0.66) models. Hence, the CFUS model can be used as a potential tool for the next-generation hydrometry in semi-gauged river basins for riverine water resources assessment and inter-state water-sharing conflict resolution.