The study of artificial intelligence for predicting land use changes in an arid ecosystem

During the 21st century, artificial intelligence methods have been broadly applied in geosciences to simulate complex dynamic ecosystems, but the use of artificial intelligence (AI) methods to reproduce land-use/cover change (LUCC) in arid ecosystems remains rare. This paper presents a hybrid modeling approach to understand the complexity in LUCC. Fuzzy logic, equation-based systems, and expert systems are combined to predict LUCC as determined by water resources and other factors. The driving factors of LUCC in this study include climate change, ecological flooding, groundwater conditions, and human activities. The increase of natural flooding was found to be effective in preventing vegetation degradation. LUCCs are sensitive under different climate projections of RCP2.6, RCP4.5, and RCP8.5. Simulation results indicate that the increase of precipitation is not able to compensate for the additional evaporation losses resulting from temperature increases. The results indicate that grassland, shrub, and riparian forest regions will shrink in this study area. The change in grasslands has a strong negative correlation with the change in groundwater salinity, whereas forest change had a strong positive correlation with ecological flooding. The application of artificial intelligence to study LUCC can guide land management policies and make predictions regarding land degradation.