Most rainfall-induced landslide forecasting models focus on the relation between landslides and rainfall,which is one of the dynamic factors,and seldom consider the stacitc factors,such as geological and geograpical f...Most rainfall-induced landslide forecasting models focus on the relation between landslides and rainfall,which is one of the dynamic factors,and seldom consider the stacitc factors,such as geological and geograpical factors.Landslide susceptibility,however,is determinded by both static and dynamic factors.This article proposes a static and dynamic factors-coupled forecasting model(SDFCFM)of regional rainfall-induced landslides,which quantitatively considers both the static and dynamic factors that affect landslides.The generalized additive model(GAM)is applied to coupling both factors to get the landslide susceptibility.In the case study,SDFCFM is applied to forecast the landslide occurrences in Shenzhen during a rainfall process in 2008.Compared with the rainfall logistic regression model,the resulting landslide susceptibility map illustrates that SDFCFM can reduce the forecast redundancy and improve the hit ratio.It is both applicable and practical.The application of SDFCFM in landslide warning and prevention system will improve its efficiency and also cut down the cost of human and matreial resources.展开更多
How to determine the weight value and how to determine the numbers of variables are tWo difficult questions for the inequality weight moving average forecasting model.Based n explanations of the concept of the weight ...How to determine the weight value and how to determine the numbers of variables are tWo difficult questions for the inequality weight moving average forecasting model.Based n explanations of the concept of the weight contribution rate and that of the key neural node,a new method by which the weight value and the variable number can be determined has been put forward in this paper,and reality-imitating experiments have been made to prove that by way of the neural network,the difficulties existed in the traditional prediction method can be solved and the predictive precision can be improved at the same time.展开更多
基金Supported by the National High Technology Research and Development Program of China("863"Program)(Grant No.2007AA12Z216,2007AA120502)National Natural Science Foundation of China(Grant No.40701134)
文摘Most rainfall-induced landslide forecasting models focus on the relation between landslides and rainfall,which is one of the dynamic factors,and seldom consider the stacitc factors,such as geological and geograpical factors.Landslide susceptibility,however,is determinded by both static and dynamic factors.This article proposes a static and dynamic factors-coupled forecasting model(SDFCFM)of regional rainfall-induced landslides,which quantitatively considers both the static and dynamic factors that affect landslides.The generalized additive model(GAM)is applied to coupling both factors to get the landslide susceptibility.In the case study,SDFCFM is applied to forecast the landslide occurrences in Shenzhen during a rainfall process in 2008.Compared with the rainfall logistic regression model,the resulting landslide susceptibility map illustrates that SDFCFM can reduce the forecast redundancy and improve the hit ratio.It is both applicable and practical.The application of SDFCFM in landslide warning and prevention system will improve its efficiency and also cut down the cost of human and matreial resources.
文摘How to determine the weight value and how to determine the numbers of variables are tWo difficult questions for the inequality weight moving average forecasting model.Based n explanations of the concept of the weight contribution rate and that of the key neural node,a new method by which the weight value and the variable number can be determined has been put forward in this paper,and reality-imitating experiments have been made to prove that by way of the neural network,the difficulties existed in the traditional prediction method can be solved and the predictive precision can be improved at the same time.
