Enhanced Markov-FLUS Model Incorporating Spatial Distribution Characteristics

The traditional Markov-FLUS model for land use prediction primarily emphasizes quantity changes and spatial distribution of land types, neglecting the influence of their inherent spatial characteristics on prediction precision. This paper introduces an innovative approach by designing shape control parameters and developing an enhanced Markov-FLUS model. The model integrates artificial neural networks to capture the relationship between land type occurrence probability and driving factors, incorporating common points, common edges, distance, and aggregation parameters alongside a cellular automata model. Using the Yellow River basin as a case study, the paper compares the model's simulation performance before and after enhancement, focusing on the land types with the most and least improvement. Results indicate that the refined model achieves a superior fitness function value in simulating land use within the Yellow River basin.