暴雨灾害 (Dec 2023)
Assessment of multi-source observation merged 1 km-grid precipitation product during the disastrous rainstorms in Guangdong
Abstract
This paper aims to assess the latest 1 km-grid Analysis Real Time (ART_1 km) precipitation product developed by the National Meteorological Information Center of China Meteorological Administration (CMA), which can provide great support for disaster weather monitoring and warning, intelligent grid forecasting and weather services. Observed precipitation data from the independent stations (including non-uploaded regional meteorological stations and hydrometric stations) that were not integrated into the ART_1 km precipitation product as well as precipitation classification inspection are used to assess the quality of this product during twenty disastrous rainstorm cases from May to August during 2019-2022 in Guangdong. The results show that the ART_1 km precipitation product successfully reproduces the precipitation location, strength, and trends in these cases, with the best performance in the Pearl River Delta, the east of eastern Guangdong, and the north of northern Guangdong. The stronger the precipitation, the greater the correlation as well as the root mean square error (RMSE) and mean error (ME) between the ART_1 km precipitation and the observed precipitation. When the hourly precipitation is not classified, about 60% of these independent stations present a correlation efficient ≥ 0.8, more than 90% of the stations present an RMSE within the range of [1.0, 5.0) mm, and more than 60% of the stations present a ME within ±0.1 mm. When the hourly precipitation is < 5 mm, most of the stations have a correlation efficient < 0.5, an RMSE within the range of [1.0, 5.0) mm, and a ME within [0.0, 0.5] mm. When the hourly precipitation is ≥ 20 mm, 42%~56% of the stations have a correlation efficient ≥ 0.5, and most of the stations have an RMSE ≥ 10 mm and a ME < 0 mm, even when the hourly precipitation is ≥ 50 mm, most of the stations have a ME < -10 mm. Overall, ART_1 km precipitation is usually underestimated at the independent stations, and integrating observations from more sites into producing ART_1 km precipitation is helpful to improve the quality of the products.
Keywords
