Forestry Tourism Resource Carrying Capacity Prediction Model Based on Multi-Source Data Algorithm

A novel prediction model for forestry tourism resource carrying capacity, integrating multi-source data fusion with advanced machine learning techniques, demonstrates exceptional accuracy and stability. The framework combines a modal-level multi-head cross-attention mechanism with a GAT-Transformer model to adaptively weigh and fuse heterogeneous data, suppressing noise and capturing long-term temporal dependencies. Tested on two years of multi-source observation data from Wulingyuan National Forest Park, the model excels in temporal scale adaptation, spatial generalization, and resilience against missing data and noise. Its lightweight structure ensures low inference latency, balancing prediction accuracy with engineering practicality.

Ma, Y., & Geng, Y. (2026). Forestry Tourism Resource Carrying Capacity Prediction Model Based on Multi-Source Data Algorithm. Forests, 17(5), 534. https://doi.org/10.3390/f17050534