Quality Assessment of the Precipitation Analysis Products from RMAPS-RISE in Beijing

Based on precipitation measurements after quality control obtained from 528 automatic stations during 2022 in Beijing, we assessed the performance of the 1-hour accumulated precipitation analysis product of RMAPS-RISE through non-independent and independent validation, and the quality of the precipitation analysis product during the flood period of 2022 was specifically assessed. We also evaluated the error of this precipitation analysis product in different underlying surface regions using European Space Agency (ESA) land cover data, and three precipitation events with different intensities were selected for inspection. The results showed that the precipitation analysis products had good correlation (COR), mean error, Mean Absolute Error (MAE), Root-Mean-Square error (RMSE), and bias with observations from automatic stations, and the applicability in Beijing was good. There were more independent validation errors than non-independent validation errors. The RMSE and MAE during the flood period, May to September 2022, were significantly larger than those during the non-flood period. In the flood period of 2022, the COR of the precipitation analysis product in 16 areas of Beijing was above 0.93, the MAE less than 0.04 mm/h, and the RMSE below 0.4 mm/h, except for the Tongzhou area; while automatic stations with large MAE and RMSE in the independent validation were distributed in the Tongzhou, Yanqing, Changping, and Shunyi areas, and precipitation from these stations was significantly overestimated. Errors of precipitation analysis products in tree cover, grassland, cropland, built-up, and bare areas were not significant, and the correlation coefficient was the highest for the underlying surface area in grassland during the flood period of 2022. Evaluation results for precipitation levels in the flood period of 2022, showed that the RMSE increased with an increase in precipitation: the RMSE was 0.45 mm/h in light rain and 9.02 mm/h in downpour; while the TS score remained above 0.91 at different precipitation levels. However, TS score is not good enough for independent validation, performing better only for light and torrential rain. For the three selected precipitation events occurring in Beijing from May to September in 2022, the COR of the precipitation analysis product was above 0.9 in independent validation; while the tendency of the precipitation analysis product appeared consistent with that of observations by automatic stations, the precipitation peak was reflected, and local heavy rainfall was accurately depicted when accumulated precipitation was greater than 100.0 mm.