Abstract: Precipitation is a key driving factor of urban flooding and other flood disasters that can result in huge losses in human lives, property and social economy. Therefore, a science-based comprehensive understanding of its distribution pattern and development trend is essential for urban flood control and disaster mitigation. This paper presents an analysis of the spatial variability, time trend, and periodicity of the precipitation over Guangdong Province in flood season, using the methods of spatial interpolation, Mann-Kendall test, moving t test, and wavelet analysis, based on the monthly precipitation data of 1960 to 2020 at 26 stations. Results indicate this province can be partitioned into four zones with different peak characteristics of the monthly precipitation at the stations. And changes in precipitation were not synchronous between different months or between coastal and inland regions. We demonstrate the spatial paths traced by the high-value center and low-value zones of precipitation that moved along in flood season. By grouping precipitation sequences into coastal vs inland and pre-flood vs post-flood types, the trends and significant mutation points of each group are detected; the dominant cycles of each group on the significant cycle scale are identified.

Key words: precipitation, flood season, spatiotemporal variability, Mann-Kendall test, moving t test, wavelet analysis