Skip to main content
Passa alla visualizzazione normale.

CHRISTIAN CONOSCENTI

Data Mining Technique (Maximum Entropy Model) for Mapping Gully Erosion Susceptibility in the Gorganrood Watershed, Iran

  • Authors: Narges Javidan; Ataollah Kavian; Hamid Reza Pourghasemi; Christian Conoscenti; Zeinab Jafarian
  • Publication year: 2020
  • Type: Capitolo o Saggio
  • OA Link: http://hdl.handle.net/10447/456776

Abstract

Soil erosion is a serious problem affecting most of the countries. This study was carried out in Gorganrood Watershed (Iran), which extends for 10,197 km2 and is severely affected by gully erosion. A gully headcut inven- tory map consisting of 307 gully headcut points was provided by Google Earth images, field surveys, and national reports. Gully conditioning factors including sig- nificant geo-environmental and morphometric variables were selected as predictors. Maximum entropy (ME) model was exploited to model gully susceptibility, whereas the area under the ROC curve (AUC) and draw- ing receiver operating characteristic (ROC) curves were employed to evaluate the performance of the model. The highly acceptable predictive skill of the ME model confirms the reliability of the procedure adopted to using this model in other gully erosion studies, as they are qualified to rapidly producing accurate and robust GESMs (gully erosion susceptibility maps) for making decisions and management of soil and water. The result is useful for local administrators to recognize the areas that are most susceptible to gully erosion and to best allocate resources for soil conservation approaches. Three different sample datasets including 70% for training and 30% for validation were randomly prepared to evaluate the robustness of the model for gully erosion. The accuracy of the predictive model was evaluated by drawing ROC curves and by calculating the area under the ROC curve (AUC). The ME model performed excellently both in the degree of fitting and in predictive performance (AUC values well above 0.8), which resulted in accurate predictions.