AI-assisted scenario-based modeling for sustainable tourism development in the context of climate change: evaluating future land use and environmental impacts

This study applies an AI-assisted geospatial modeling framework to evaluate the future environmental impacts of tourism development and support sustainable land-use planning under climate change through 2050. Using the Siraya National Scenic Area in Taiwan as a case study, the research integrates a trained Multi-Layer Perceptron neural network (MLP-NN) within TerrSet’s Land Change Modeler to estimate land-use transition potentials, followed by Markov Chain Analysis to project Land Use and Land Cover (LULC) change under three scenarios: Business as Usual (BAU), Tourism Trend (TOUR), and Natural Reclamation (NARE). The projected land-use patterns are then assessed using the Tourism Environmental Impact (TEI) framework, which incorporates population density, infrastructure, and tourist distribution to evaluate tourism-related environmental pressure. The results show that tourism-related environmental impacts increased from 2007 to 2021, largely driven by infrastructure expansion and land conversion. For 2050, the TOUR scenario produces the highest total TEI, indicating the greatest environmental pressure, whereas NARE records the lowest TEI and the strongest expansion of forest cover, suggesting the greatest potential for ecological restoration and climate resilience. The modeling results demonstrate how AI-assisted land-use simulation can be integrated with tourism-environmental indicators to help policymakers compare trade-offs among tourism growth, conservation, and long-term landscape sustainability.