Monitoring hydrological drought using long-term satellite-based precipitation data

Long-term (over 30a) satellite-based quantitative rainfall estimate (SRE) products provide an ideal data source for hydrological drought monitoring. This study mainly explores the suitability of the two long-term SREs, the Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR) and the Climate Hazards Group (CHG) Infrared Precipitation with Stations (CHIRPS), for hydrological drought monitoring. A hydrological drought index called the standardized streamflow index (SSI) was used as an example and the Grid-based Xinanjiang (GXAJ) hydrological model was used for streamflow generation of the SREs. A middle size basin in the humid region of south China was selected as case study. The obtained results show that both SREs present acceptable performances for hydrological modeling, and CHIRPS outperformed PERSIANN-CDR. SSIs calculated by the SRE simulations generally fit well with the trend of observation-based on SSI but apparent deviations in drought intensity were also found. In contrast to hydrological modeling, performance of the SRE-based SSI showed almost no change after model recalibration. Both SREs generally present acceptable classification accuracy but tended to underestimate the levels of drought types. Both SREs accurately captured the beginning, end, and duration of this drought event; however, several deviations were found in severity and intensity estimation of the drought event. In general, both SREs are suitable for hydrological drought monitoring. Although the CHIRPS generally presented better performance, the PERSIANN-CDR is still adequate for hydrological drought monitoring.