Study Of High-Risk Storm Regionalization Based On Regional L-Moments And Its Application In Guangxi

Storm disaster and secondary disasters has become the severest threat to people and property safety among natural disasters. The existing storm design standard in China, could not properly response in today’s storm objective laws in the background of frequently extreme hydrometeorological eventsL-Moments, based on probability weighted moments, the most advanced parameter estimation method in the world, shows good robustness and unbiased. Combining L-Moments method with regional analysis method becomes regional L-Moments method which could be applied in regional storm frequency design and by which could improve the accuracy of parameter estimating and precision of frequency estimations significantly. Estimates of regional L-Moments has been the national flood design standard in United States of America and regional L-Moments also showed good effect when applied in many countries and regionsThis paper introduced the theory of regional L-Moments and its theoretical and computational methods in detail. For annual maximum1h,3h,6h,12h and24h rainfall data of Guangxi646data-quality-controlled rain gauges, storm frequency analysis was conducted. Based on the basic climate regions, the paper divided Guangxi hydrometeorological homogenous regions in different durations by discordancy measure and heterogeneity measure. In addition, via independent sample test, the correlation of each sites in homogenous regions was reduced.The best distribution of each homogenous region was chosen from the following5three-parameter distributions:Generalized Pareto distribution (GPA), Generalized Logistic Distribution (GLO), Generalized Extreme Value Distribution (GEV) Generalized Normal Distribution (GNO) Pearson Type Ⅲ Distribution (P-Ⅲ). Via3goodness-of-fit measures, Monte Carlo simulation, sample L-Moments mean square error test and measured data test, to determine the best distribution in each region. According to the test results, best distributions were selected in every homogenous region of5durations and results showed GEV and GNO better than other distributions in Guangxi. In the next, estimates of different return periods and durations were obtained by regional analysis. When scanning the results, the internal consistency adjustment was conducted and in this way, it made the estimates more reasonable and reliable.Finally, the concept of storm high risk zoning was introduced. By kriging spatial interpolation, the storm frequency estimates calculated by regional L-Moments could be divided into storm high risk zoning directly. The results show that Guangxi has3theoretical storm high risk zonings:southern coastal areas (highest-risk areas), northwestern mountain areas and northern mountain areas (short-duration high-risk areas).