从江县汛期暴雨频率时空特征及地形关系分析

Spatiotemporal Characteristics of Rainstorm Frequency in Flood Season and Its Relationship with Topography in Congjiang County

摘要: 为研究从江县汛期暴雨，基于1985~2024年从江县长序列气象观测数据与高分辨率地形数据，采用小波分析、最小二乘法(OLS)、地理加权回归(GWR)等方法，系统分析了区域汛期暴雨的时空分布特征，定量解析了海拔、坡度等地形因子对汛期暴雨频率的局地影响。从江县暴雨事件时空分布非均匀性显著，空间上呈“西多东少”的分布格局，时间上，从江国家站年际变化存在显著的15~20年准周期震荡，且近20年该周期信号的震荡强度显著增强。地形对暴雨的影响存在明显的空间异质性，全局尺度下海拔对暴雨频率存在边缘显著的正影响，而地理加权回归模型有效捕捉到了局地差异：县域西部山区海拔与坡度对暴雨频率存在显著的正影响，东部丘陵区域该影响则较弱。相较于传统的回归模型，地理加权回归模型的拟合优度提升近2倍，能够更好地刻画复杂山地地形对汛期暴雨的局地影响。

Abstract: To investigate rainstorms during the flood season in Congjiang County, based on the long-term meteorological observation data and high-resolution topographic data of Congjiang County from 1985 to 2024, methods including wavelet analysis, Ordinary Least Squares (OLS), and Geographically Weighted Regression (GWR) were adopted to systematically analyze the spatiotemporal distribution characteristics of regional flood-season rainstorms, and quantitatively explore the local impacts of topographic factors (e.g., elevation and slope) on rainstorm frequency in the flood season. The spatiotemporal distribution of rainstorm events in Congjiang County presents remarkable heterogeneity. Spatially, rainstorms follow a distribution pattern of “higher frequency in the west and lower frequency in the east”. Temporally, the interannual variation of rainstorms at Congjiang National Meteorological Station shows a significant quasi-periodic oscillation of 15~20 years, and the oscillation intensity of this periodic signal has increased significantly in the past 20 years. The impact of topography on rainstorms has obvious spatial heterogeneity. At the global scale, elevation exerts a marginally significant positive effect on rainstorm frequency, while the GWR model effectively captures local differences: elevation and slope show significant positive correlations with rainstorm frequency in the mountainous areas of the western county, whereas this effect is weak in the eastern hilly regions. Compared with traditional regression models, the goodness of fit of the GWR model is nearly doubled, which can better characterize the local impacts of complex mountainous topography on flood-season rainstorms.