Urbanization under constraint: predicting land use futures in the geographically fragile city of Rize

This study explores land use and land cover (LULC) dynamics in the geographically constrained and ecologically fragile city center of Rize, Turkey. Using remote sensing-based classification from 1990 to 2025, future LULC changes for 2060 were simulated through the QGIS MOLUSCE module, integrating Cellular Automata-Markov Chain (CA-Markov) and Artificial Neural Network (ANN) models. Results indicate a substantial increase in urban areas, primarily at the expense of agricultural and natural lands, with urban expansion concentrated in low-lying, flood-prone zones such as riverbeds and coastal plains. The ANN-based framework demonstrated high predictive accuracy (93.53% overall accuracy; spatial Kappa: 1.00), confirming its applicability for complex terrain conditions. The findings underscore the critical need for zoning regulations informed by spatial modelling to mitigate potential implications. The proposed methodology provides a scalable and transferable decision-support tool for sustainable land management and climate-resilient urban planning in mountainous and rapidly urbanizing regions.