How do the landslide and non-landslide sampling strategies impact landslide susceptibility assessment? d A catchment-scale case study from China

The aim of this study is to investigate the impacts of the sampling strategy of landslide and non-landslide on the performance of landslide susceptibility assessment(LSA).The study area is the Feiyun catchment in Wenzhou City,Southeast China.Two types of landslides samples,combined with seven non-landslide sampling strategies,resulted in a total of 14 scenarios.The corresponding landslide susceptibility map(LSM)for each scenario was generated using the random forest model.The receiver operating characteristic(ROC)curve and statistical indicators were calculated and used to assess the impact of the dataset sampling strategy.The results showed that higher accuracies were achieved when using the landslide core as positive samples,combined with non-landslide sampling from the very low zone or buffer zone.The results reveal the influence of landslide and non-landslide sampling strategies on the accuracy of LSA,which provides a reference for subsequent researchers aiming to obtain a more reasonable LSM.

Landslide susceptibility zonation method based on C5.0 decision tree and K-means cluster algorithms to improve the efficiency of risk management

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 subsequently 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 cluster 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 frequency density of the high susceptibility zones were 6.76/km^(2) and 0.88/km^(2), 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 containing more " stable" pixels. Therefore, the obtained susceptibility map is suitable for application to landslide risk management practices.

Shallow landslide susceptibility assessment under future climate and land cover changes: A case study from southwest China

There is no doubt that land cover and climate changes have consequences on landslide activity,but it is still an open issue to assess and quantify their impacts.Wanzhou County in southwest China was selected as the test area to study rainfall-induced shallow landslide susceptibility under the future changes of land use and land cover(LULC)and climate.We used a high-resolution meteorological precipitation dataset and frequency distribution model to analyse the present extreme and antecedent rainfall conditions related to landslide activity.The future climate change factors were obtained from a 4-member multimodel ensemble that was derived from statistically downscaled regional climate simulations.The future LULC maps were simulated by the land change modeller(LCM)integrated into IDRISI Selva software.A total of six scenarios were defined by considering the rainfall(antecedent conditions and extreme events)and LULC changes towards two time periods(mid and late XXI century).A physically-based model was used to assess landslide susceptibility under these different scenarios.The results showed that the magnitude of both antecedent effective recharge and event rainfall in the region will evidently increase in the future.Under the scenario with a return period of 100 years,the antecedent rainfall in summer will increase by up to 63%whereas the event rainfall will increase by up to 54%for the late 21st century.The most considerable changes of LULC will be the increase of forest cover and the decrease of farming land.The magnitude of this change can reach+22.1%(forest)and–9.2%(farmland)from 2010 until 2100,respectively.We found that the negative impact of climate change on landslide susceptibility is greater than the stabilizing effect of LULC change,leading to an over decrease in stability over the study area.This is one of the first studies across Asia to assess and quantify changes of regional landslide susceptibility under scenarios driven by LULC and climate change.Our results aim to guide land use planning and climate change mitigation considerations to reduce landslide risk.

Uncertainties of landslide susceptibility prediction:Influences of different spatial resolutions,machine learning models and proportions of training and testing dataset

This study aims to reveal the impacts of three important uncertainty issues in landslide susceptibility prediction(LSP),namely the spatial resolution,proportion of model training and testing datasets and selection of machine learning models.Taking Yanchang County of China as example,the landslide inventory and 12 important conditioning factors were acquired.The frequency ratios of each conditioning factor were calculated under five spatial resolutions(15,30,60,90 and 120 m).Landslide and non-landslide samples obtained under each spatial resolution were further divided into five proportions of training and testing datasets(9:1,8:2,7:3,6:4 and 5:5),and four typical machine learning models were applied for LSP modelling.The results demonstrated that different spatial resolution and training and testing dataset proportions induce basically similar influences on the modeling uncertainty.With a decrease in the spatial resolution from 15 m to 120 m and a change in the proportions of the training and testing datasets from 9:1 to 5:5,the modelling accuracy gradually decreased,while the mean values of predicted landslide susceptibility indexes increased and their standard deviations decreased.The sensitivities of the three uncertainty issues to LSP modeling were,in order,the spatial resolution,the choice of machine learning model and the proportions of training/testing datasets.