The Monte Carlo Simulation Based On G-h Distribution For Sichuan Loss Assessment Of Provincial Mudslide Disaster

After forty years of rapid development,some aspects of our society have improved significantly.With the improvement of people's living standard,people's understanding of the danger of natural disasters is becoming more and more comprehensive,and the research on natural disasters by disaster management agencies and academic circles is getting deeper and deeper.Some researches have paid more attention to post-disaster relief,but there are obvious deficiencies in post-disaster management and response to natural disasters.With the increasing emphasis on disaster prevention,natural disaster risk management research has become an indispensable part of disaster management.Our country land area ranked third in the world,the vast territory of complex and diverse topography makes our country become a country with frequent natural disasters.In many natural disasters,the debris flow disaster is a high hazard,wide distribution of common natural disasters,Sichuan province is relatively dense distribution of debris flow disasters in provinces of our country.In order to better understand the risk of debris flow disaster in Sichuan province,on the basis of DBSCAN clustering,the cluster analysis of debris flow disaster areas in Sichuan province was carried out,and AcrGIS software was used to do the kernel density analysis.In this paper,it is found that the distribution pattern of debris flow disasters in Sichuan province in the past 30 years is mainly one main,one primary and two pairs.Four regions are defined as the first region of density,the second region of density,the third region of density and the fourth region of density.The first density area is mainly distributed in the west of ya 'an,the north of liangshan yi autonomous prefecture and the northwest of leshan.The second density area is mainly distributed in the southwest of leshan city,the west and southwest of yibin city,the west of zigong city and the southwest of neijiang city.The third area of density is mainly distributed in suining city,guang 'an city and the western area of dazhou city.The fourth area is mainly distributed in the south of liangshan yi autonomous prefectureand the west of panzhihua.Based on clustering,this paper analyzes the risk of debris flow loss in the whole region,the first density region and the second density region of Sichuan province.Based on Sichuan province's debris flow disaster losses is fat-tailed distribution characteristics,this paper applied the g-h-VaR model to measure risk.VaR model is a kind of modle based on historical data distribution on the risk measurement model,on the historical data distribution fitting is necessary for risk measurement based on VaR model,considering the logarithmic distribution of debris flow disaster losses in Sichuan province is relatively close to the normal distribution,so the g-h distribution of debris flow disaster loss distribution fitting in Sichuan province will be more advantageous.By using python software,this paper applies monte carlo simulation(MC)to simulate the historical data of 29 years,and then applies the method of gradient descent to estimate the parameters of the fitted g-h distribution.Finally,VaR model is used to estimate the VaR value of four confidence levels in three different regions.