An integrated approach for gully erosion susceptibility mapping and factor effect analysis
Jingge Liu,Alireza Arabameri,Chandan Surabhi Das,Pritam Sarkar
DOI: https://doi.org/10.1016/j.asr.2024.12.021
IF: 2.611
2024-12-14
Advances in Space Research
Abstract:Gully erosion is one of the major global environmental threats that frequently affects semi-humid to arid Mediterranean regions and contributes to a wide range of ecological problems. Recognizing vulnerable areas to gully erosion and creating a comprehensive gully erosion susceptibility map (GESM) can assist in the lessening of land degradation and damage to numerous infrastructures. The primary goal of this research is to build a random subspace-based function tree (RSFT), i.e., an ensemble model, and compare it with other standard models such as Fisher's linear discriminant analysis (FLDA), Nave Bayes tree (NBTree), J48 Decision Tree, and random forest (RF) models in order to identify which model generates the most accurate outcomes. Overall, a total number of 489 gully sites were utilised for modelling and validation purpose, with 377 (70%) used for modelling and 112 (30%) used for validation. Fourteen salient gully erosion conditioning factors (GECFs) were implemented for constructing the GESMs. The efficacy and significance of several GECFs were assessed through the random forest, or RF, model for gully erosion modelling. Using the GES maps, we computed the success rate curve (SRC) and prediction rate curve (PRC), as well as their areas under the curves (AUC). The AUC (SRC, PRC) scores for the RSFT model were 0.906 and 0.916, consequently, while the outcomes for the RF, NBTree, FLDA, and J48 models were 0.875 and 0.869, 0.861 and 0.859, 0.792 and 0.816, and 0.779 and 0.811. AUC findings indicated that the RSFT model delivered the most precise predictions, trailed by the RF, NBTree, FLDA, and J48 models. In terms of RMSE, each of the models performed adequately; however, RSFT exhibits the lowest RMSE values of all models, with 0.31 (training dataset) and 0.29 (validation dataset), which shows that RSFT is substantially more accurate than other models in forecasting gully erosionThus, the results of this research can be used by local managers and planners for environmental management. The results from our study suggests that all of the GESM models have high efficiency, and can be employed to formulate adequate measures for safeguarding of soil and water.
geosciences, multidisciplinary,meteorology & atmospheric sciences,astronomy & astrophysics,engineering, aerospace