Machine learning algorithms are an important measure with which to perform landslide susceptibility assessments,but most studies use GIS-based classification methods to conduct susceptibility zonation.This study presents a machine learning approach based on the C5.0 decision tree(DT)model and the K-means cluster algorithm to produce a regional landslide susceptibility map.Yanchang County,a typical landslide-prone area located in northwestern China,was taken as the area of interest to introduce the proposed application procedure.A landslide inventory containing 82 landslides was prepared and subse-quently randomly partitioned into two subsets:training data(70%landslide pixels)and validation data(30%landslide pixels).Fourteen landslide influencing factors were considered in the input dataset and were used to calculate the landslide occurrence probability based on the C5.0 decision tree model.Susceptibility zonation was implemented according to the cut-off values calculated by the K-means clus-ter algorithm.The validation results of the model performance analysis showed that the AUC(area under the receiver operating characteristic(ROC)curve)of the proposed model was the highest,reaching 0.88,compared with traditional models(support vector machine(SVM)=0.85,Bayesian network(BN)=0.81,frequency ratio(FR)=0.75,weight of evidence(WOE)=0.76).The landslide frequency ratio and fre-quency density of the high susceptibility zones were 6.76/km2 and 0.88/km2,respectively,which were much higher than those of the low susceptibility zones.The top 20%interval of landslide occurrence probability contained 89%of the historical landslides but only accounted for 10.3%of the total area.Our results indicate that the distribution of high susceptibility zones was more focused without contain-ing more"stable"pixels.Therefore,the obtained susceptibility map is suitable for application to landslide risk management practices. 相似文献
In this study, three high frequent occurrence regions of tropical cyclones(TCs), i.e., the northern South China Sea(the region S), the south Philippine Sea(the region P) and the region east of Taiwan Island(the region E), are defined with frequency of TC's occurrence at each grid for a 45-year period(1965–2009), where the frequency of occurrence(FO) of TCs is triple the mean value of the whole western North Pacific. Over the region S, there are decreasing trends in the FO of TCs, the number of TCs' tracks going though this region and the number of TCs' genesis in this region. Over the region P, the FO and tracks demonstrate decadal variation with periods of 10–12 year, while over the region E, a significant 4–5 years' oscillation appears in both FO and tracks. It is demonstrated that the differences of TCs' variation in these three different regions are mainly caused by the variation of the Western Pacific Subtropical High(WPSH) at different time scales. The westward shift of WPSH is responsible for the northwesterly anomaly over the region S which inhibits westward TC movement into the region S. On the decadal timescale, the WPSH stretches northwestward because of the anomalous anticyclone over the northwestern part of the region P, and steers more TCs reaching the region P in the greater FO years of the region P. The retreating of the WPSH on the interannual time scale is the main reason for the FO's oscillation over the region E. 相似文献
To understand the impacts of large-scale circulation during the evolution of El Niño cycle on tropical cyclones (TC) is important and useful for TC forecast. Based on best-track data from the Joint Typhoon Warning Center and reanalysis data from National Centers for Environmental Prediction for the period 1975–2014, we investigated the influences of two types of El Niño, the eastern Pacific El Niño (EP-El Niño) and central Pacific El Niño (CP-El Niño), on global TC genesis. We also examined how various environmental factors contribute to these influences using a modified genesis potential index (MGPI). The composites reproduced for two types of El Niño, from their developing to decaying phases, were able to qualitatively replicate observed cyclogenesis in several basins except for the Arabian Sea. Certain factors of MGPI with more influence than others in various regions are identified. Over the western North Pacific, five variables were all important in the two El Niño types during developing summer (July–August–September) and fall (October–November–December), and decaying spring (April–May–June) and summer. In the eastern Pacific, vertical shear and relative vorticity are the crucial factors for the two types of El Niño during developing and decaying summers. In the Atlantic, vertical shear, potential intensity and relative humidity are important for the opposite variation of EP- and CP-El Niños during decaying summers. In the Southern Hemisphere, the five variables have varying contributions to TC genesis variation during peak season (January–February–March) for the two types of El Niño. In the Bay of Bengal, relative vorticity, humidity and omega may be responsible for clearly reduced TC genesis during developing fall for the two types and slightly suppressed TC cyclogenesis during EP-El Niño decaying spring. In the Arabian Sea, the EP-El Niño generates a slightly positive anomaly of TC genesis during developing falls and decaying springs, but the MGPI failed to capture this variation.