Evaluating the accuracy of Climate Hazard Group (CHG) satellite rainfall estimates for precipitation based drought monitoring in Koshi basin, Nepal

摘要

Highlights ? Accuracy of Climate Hazard Group’s (CHG) satellite products is assessed in the Koshi basin, Nepal. ? The satellite only (CHIRP) and station blended (CHIRPS) products have different accuracies. ? Raw satellite products are subjected to the Mean Field Bias (MFB) correction. ? Corrected products are used to estimate the Standardized Precipitation Index (SPI). ? The spatial and temporal trends of the SPI values have reflected historically known drought events. Abstract Study region Koshi basin, Nepal. Study focus While rainfall estimates based on satellite measurements are becoming a very attractive option, they are characterized by non-negligible biases. As such, we assessed the accuracy of two satellite products of the Climate Hazard Group (CHG) – (a) a satellite-only Climate Hazards Group InfraRed Precipitation (CHIRP) product, and (b) a CHIRP blended with ground-based station data (CHIRPS) – at a monthly time scale from 1981 to 2010 in the Koshi basin of Nepal using ground-based measurements. A separate analysis was also made for the data set after 1992, as the number of stations used in the blending has significantly reduced since 1992. Next, both CHG data sets were used to calculate one of the most popularly-used precipitation-based drought indicators – the Standardized Precipitation Index (SPI). New hydrological insights for the study region The accuracy of the CHG data set was found to be better in low-lying regions, while it was worse in higher-elevation regions. While the CHIRPS data set was better for the whole period, the CHIRP data set was found to be better for the period after 1992. Physiographic region-wise bias correction has improved the accuracy of the CHG products significantly, especially in higher-elevation regions. In terms of SPI values, the two CHG data sets indicated different drought severity when considering the whole period. However, the SPI values, and hence the drought severity were comparable when using the data from after 1992.