Multi-scale accuracy assessment of multiple high spatial and temporal resolution precipitation products in the Pearl River Delta region.

Based on hourly precipitation data from 28 rainfall stations in the Pearl River Delta (PRD) region from 2001 to 2012, the accuracy of CMORPH, GPM and ERA5 is evaluated in the PRD on the annual, monthly and daily time scales by calculating 6 statistical metrics and exploring the frequency of different precipitation events. The results show that: (1) Compared with the three time scales of year, month and day, three products showed a highly positive correlation with the measured rainfall station on the monthly scale, but the correlation is weak on the annual and daily scales. (2) At the daily scale, all three precipitation products perform worse in the flood season than in the non-flood season. They are less accurate in detecting precipitation events in dry months than in other months, but the false alarm rate is higher in autumn and winter, and they all have a higher false alarm rate for stations near the coast than for inland stations. (3) In contrast, GPM can more accurately reflect the spatio- temporal trend and spatial distribution characteristics of precipitation in the PRD region. (4) Both CMORPH and GPM show high capacity to estimate various types of precipitation events, which can well reflect the contribution of the measured data to different types of precipitation events. Overall, GPM can best reflect the actual situation in the PRD region among the three products. [ABSTRACT FROM AUTHOR]