Comparing geographic information systems-based fuzzy-analytic hierarchical process approach and artificial neural network to characterize soil erosion risk indexes

The pressure on the lands has increased with the dramatic increase in the world population in the last century. Erosion which is a natural process has become a serious artificial concern with this growing pressure. Especially, most of the farmlands in Turkey are particularly affected by erosion. In the current study, it is aimed to determine erosion risk index classes and generate their maps using F-AHP and ANN approaches applied for the estimate of soil erosion risk index (ERI). In addition, these approaches were associated with GIS and geostatistical techniques based on seven soil erosion indicators in Sinop Province including humid and sub-humid coastal environmental ecosystems in the central Black Sea Region of Turkey. In this research, vegetation cover, land use, soil depth, erosivity (precipitation), erodibility (USLE-K), slope (%), and parent material/geology were used as input data by taking into consideration of several literature reviews. According to study results, index values of ERIF-AHP and ERIANN classes were determined quite close to each other. The soil erosion risk index for Sinop province in Turkey indicates that less than 35% of the study area has a low and very low erosion risk area (34.3%), 32.4% is of moderate soil erosion risk area and about 33.2% of the area has high and very high erosion risk when based on F-AHP method. In addition, as for ERIANN, high and very high erosion risk classes made up 30.9% of the total area, while low- and very-low-risk classes made up 37.3%.