The CF Model Enhanced By SMPHSM-HGCR For Evaluating The Hazard Susceptibility Of The Floods And Debris Avalanches In The Mid-west Segment Of Langshan Range,Inner Mongolia

Floods and debris avalanches can cause an enormous loss to human life and public properties.Even though they have several characteristics in common,their triggering mechanism and subsequent consequences are inevitably different.For recent decades,a variety of models were applied to evaluate hazard susceptibility.Refining the original hazard susceptibility evaluation model based on regional characteristic will be significant in theory and practice.The study area was in the mid-west segment of Langshan Range,Bayan Nur,Inner Mongolia.The research had utilized 20 related parameters extracted from geomorphological and sedimentary factors and further applied K-means clustering analysis to distinguish floods and debris avalanches.The hazard classification results integrating with the geomorphic classification result were further applied as support for subdividing the input parameters of hazard susceptibility evaluation model,as the name of a Subdivision Method for Parameters of Hazard Susceptibility Model based on Hazard and Geomorphology Classification Result(SMPHSM-HGCR).The SMPHSM-HGCR was proposed to refine the original Certain Factor(CF)model so as to evaluate the regional hazard susceptibility of the study area in the mid-west segment of Langshan Range.The evaluation result was validated by the Receiver Operation Characteristic(ROC)curve,which can be applied to extract the hazard-prone area by implementing the selection of the optimized point of the ROC curve.The major achievement of this study includes:1 The study proposed a practical parameter subdivision method called SMPHSM-HGCR based on the differentiation of geomorphology and hazard classification.This method achieved a refinement of integrating geoscience discipline with statistic principle to refine the traditional hazard susceptibility evaluation model.Additionally,the hazard susceptibility evaluation result by using this method could be comparable in diverse study areas.2 Through the discrepancy of flood and debris avalanches by K-means clustering analysis on the basis of 20 extracted parameters,the classification result illustrated a result that floods were more inclined to occur in the most valleys of the western study area(LS001-007,LS010),while debris avalanches occurred in the eastern area(LS008,LS011-013).3 Both original and modified parameter subdivision methods,respectively representing equal interval and SMPHSM-HGCR,were implemented into the same CF model.Both models shared a similar result that the hazard susceptibility indexes in the piedmont area of Langshan Range and the downstream of the valley were highest.Meanwhile,the validation result of Receiver Operation Characteristic(ROC)curve indicated that the Area Under the Curve(AUC)value of the original CF model was 0.7634,while the AUC value of the modified CF was 0.7977,slightly higher than the AUC value of the original model.4 The hazard-prone areas were identified and were indeed used to explore the reason of hazard occurrence in the study area: the hazard susceptibility index of 0.976,which was in the range of-1 to 0.997,was selected by pinpointing the optimal point in the ROC curve.The spatial distribution of hazard-prone areas suggested that the fault activities in the study region may control the hazard occurrence because the location of hazard-prone areas and faults are very close.To sum up,the SMPHSM-HGCR subdivision method involving the data internal differentiation in every single hazard was proved to be efficient in refining the original CF model in hazard susceptibility evaluation.This method can not only provide a new perspective for parameter subdivision to overcome the restriction that subdivision established on the statistic basis alone,but also a new perspective for the modification of hazard susceptibility evaluation model.Furthermore,it can also afford more intuitional and more direct references for the identification of hazard-prone area and the relevant decision makers.