Long series remote sensing of precipitation products in drought monitoring of the Yellow River basin: applicability assessment

Grid precipitation (CMA) obtained by interpolation from 298 meteorological stations from 1983 to 2019 were used as standard． Correlation coefficient, relative deviation, root mean square error, and other statistical indexes were applied to evaluate three precipitation products for characterization of drought on four different time scales (SPI1, SPI3, SPI6, SPI12)． PERSIANN-CDR was found better than the other two products in monitoring interannual drought fluctuation (SPI12) and in quantifying drought areas． MSWEP was found better than the other two products in precipitation estimation and precipitation spatial distribution pattern capture． But in the upstream region with sparse weather stations, complex terrain, and changeable climate, consistency of MSWEP showed large regional differences． PERSIANN-CDR and CHIRPS were found suitable for medium drought characterization, while MSWEP was suitable for short drought characterization． MSWEP could capture the SPI trend of CMA and spatial distribution of typical drought events． These three products need to be further improved in the identification of typical drought events (precipitation and spatial capture)． These data can be used to select remote sensing data sources for drought monitoring in the Yellow River basin, and to improve remote sensing of precipitation inversion algorithm．