The establishment of the multi-dimensional meteorological early warning criterion of landslide and the division of the"grid"early warning unit can provide a scientific basis for the landslide early warning,for the purpose of which 205 rainfall-induced landslides in Panan County,Zhejiang Province were studied in this paper.Firstly,based on the average effective rainfall intensity-diachronic(I-D)threshold model,the critical threshold curves were divided by ordinary least squares regression(OLSQ)and quantile regression(QR).Secondly,the I-D-R threshold model was established by the I-D threshold model optimized by considering the daily rainfall(R),and different parameter estimation methods were used to compare the accuracy of different threshold models.The optimal threshold model was considered as the meteorological early warning criterion for landslide disasters in Pan'an County.Finally,considering the difference of rainfall distribution,the township level grid early warning unit was established by the terrain zoning and Voronoi diagram(VD)of Pan'an.The results show that:(1)The I-D-R threshold model has better early warning accuracy than the I-D model.The I-D-R threshold model based on QR has a better warning ability,and the accuracy of the threshold degree of warning and above is increased to 50%,and the accuracy of the threshold level of special attention and above is increased to 88.9%;(2)the rainfall conditions with I-D-R based on QR rainfall threshold are proposed as the early warning criteria(red,orange,yellow and blue)of 51 early warning units in Pan'an County,and the emergency response measures are put forward.A new threshold model is established on the basis of the research results,which can provide reference for regional meteorological early warning in Pan'an County.
landslidemeteorological early warningrainfall thresholdordinary least square regressionquantile regressionI-D-R threshold modelengineering geology
NETL
NSTL
万方数据
