Abstract: Based on precipitation observations from 5, 634 meteorological and hydrological stations in Guangdong Province, this study integrated radar quantitative precipitation estimation products using probability density function matching and optimal interpolation. The ART_1km precipitation fusion product was corrected using the space-time multiscale analysis system. Comparison with the domestic similar product ART_1km showed that both methods can effectively enhance product quality through integrating more site observations. Improvement in reduced false alarm rate for precipitation below 2 mm·h^-1 and errors for precipitation over 20 mm·h^-1, as well as decreased errors during the main flood season in site observations or topographically complex regions such as the western Leizhou Peninsula, coastal areas of western Guangdong, and mountainous areas of northern Guangdong. According to independent verification of precipitation process，the fusion of station observations and radar information was achieved through probability density function matching and optimal interpolation. This enhanced overall quality, especially in situations with poor spatial representation of stations, such as sparse sites, extreme and local precipitation. These show relatively high false alarm rates. The space-time multiscale analysis can well capture the shortwave information of station observations, and has good effects in spatial representation of stations (e. g., dense sites, continuous and stable precipitation structures, precipitation with low extreme intensity), but with a relatively high missing rate. Both methods exhibit measurable accuracy degradation in complex terrain regions.