Theory,Modelling And Applications For Risk In Situations With Small Sample

The natural risks pose a serious threat to economic development and the safety of people's life and property.Carrying out an effective risk assessment and management is an important basic work to prevent risk and reduce risk losses.Considering insufficient data and multidimensional risk factors are key confounding issues in natural risk assessment,some researches are made on the basic theory of risk analysis,risk probability prediction and assessment modelling techniques in situations where insufficient data is available.In addition,applications in risks of water resources,extreme precipitation and marine environment are performed.(1)The basic theory of risk analysis is studied.The definition of risk are refined and summarized from the viewpoint of uncertainty.The forming factors of risk are analyzed and summarized,and the extension tread of connotation change of vulnerability is concluded in detail.The cause and effect between risk factors and its consequences is studied,and some theory models for risk are proposed,including risk input-output model,and risk multiple integral model,and risk loss assessment model.(2)Maximum-entropy-logistic risk probability predication model,extreme risk prediction models of Gumbel and Gumbel copula based on maximum entropy estimation are proposed.Moreover,Coupled copula model for extreme risk predication.is introduced(3)The traditional projection index function is improved.The maximum entropy and information entropy projection index functions and two nonlinear risk assessment models are put forward,including S type function model and differential equation model.(4)Risks in Beijing,used as a case study are assessed and analyzed,such as water resources vulnerability,water shortage probability,and water supply and demand risk loss,and monthly risk between water supply and demand.Utilization rate of water resources and treatment rate of domestic sewage are the sensitive impact factors of water resources vulnerability.Water resources vulnerability in 2020 are in the state of extreme vulnerability.The water shortage risk probability in 2020 exceeded 0.95 under 33 different inflow conditions.The expected water supply and demand risk loss in 2020 is about 1159.7 billion CNY.The monthly risk in 2020 is characterized by evident seasonal variation features.After using the transferred and reclaimed water,water resources vulnerability,risk probability and risk losses decline with different degrees for all cases.(5)Water resources vulnerability and water resources security risk probability of Quanzhou are assessed and analyzed.In 2020,water resources of Licheng District and Shishi City are in the state of extreme vulnerability;those of Fengze District,Jinan City and Huian County are in the state of strong vulnerability;those of Quangan belong to moderate vulnerability;those of Luojiang District and Nanan City are in the state of mild vulnerability;and those of other districts are not vulnerable.In 2020,water resources security risk probabilities of Fengze District,Licheng District,Shishi City and Jinjiang City are close to 1,that of Huian County is only 0.034 and those of other districts are close to 0.(6)The extreme precipitation risks at some stations of Yellow River,Jinhe and Heihe River are predicted and analyzed.The hydrologic frequency analysis of the maximum daily precipitation at four weather stations is performed and shows that the Gumbel extreme risk predication model has a good performance in the case of small sample.The Gumbel copula model based on maximum entropy estimation,which requires only the lower and upper bounds of two hydrological variables for parameter estimation,is proved to be generally reliable and robust by studying the dependence pattern of the maximum daily precipitation at two pairs of stations on the tributaries of Heihe and Jinghe River respectively.The Coupled copula model is proved to have much potential by performing the hydrologic frequency analysis of the monthly precipitation in summer at two pairs of stations on the tributaries of Yellow River.Synchronous-asynchronous encounter risks of rich-poor monthly precipitation in summer in neighbouring meteorological subdivisions are studied.(7)Marine environmental risk for naval activity and natural environment risk about Northeast Passage of Arctic are assessed and analyzed.Experiments of marine environmental risk assessment for naval activity are performed to validate the improved projection index functions and sine function assessment model.The results show that maximum entropy index is more reasonable,and the sine model is more reliable and has bigger potential in situations when data is insufficient.The interannual variation rules of natural environment risk of from July to October about Northeast Passage of Arctic are studied and analyzed.