Spatiotemporal Analysis of Forest Cover Changes and Climate Dynamics in Castellón Province

Abstract

Understanding forest cover change dynamics and its association with climatic conditions is crucial for sustainable environmental management. This research examined forest cover change in Castellón Province by using satellite-based remote sensing and machine learning classification during 2010-2024. The supervised classification methods—Random Forest (RF), Support Vector Machine (SVM), and Classification and Regression Tree (CART)—were applied on Landsat images using Google Earth Engine environment. In this study, spatiotemporal trends integrate Google Earth Engine (GEE), Python-based statistical analysis, and GIS mapping to identify the best classifier as well as reliable model. RF outperformed other classifiers. It proved maximum accuracy using different performance and evaluation metrics. Classification results indicated a significant forest cover loss, especially between 2015 and 2020 due to urbanization, land use change, and potential climatic disturbances. The dynamics of forest conversion have been explored using transition matrices. It helped in detecting overall loss of areas of forest along with appreciable changes to non-forest categories. The areas of significant deforestation along with afforestation trends were also recognized using spatial maps. This puts focus on finding out the factors behind the degradation like climate change, with anthropogenic effects on land use changes. The time series analysis of NDVI trends and Climate variables (Temperature-2m, Land Surface Temperature (LST) and Precipitation) shows alarming trends in the study area. The climatic variables showed possible strong correlation between vegetation health and climate fluctuations. Pearson correlation analysis confirmed the positive relationship between NDVI and precipitation while demonstrating the negative relationship between temperature and land surface temperatures (LST). The integration of the use of machine learning, remote sensing, along with statistical analysis identifies key insight into data-driven reasoning and detailed analysis of the forest cover changes within a Mediterranean ecosystem. The findings can contribute to sustainable forest management approaches and climate adaptation strategies by providing valuable insights into forest loss patterns including climatic drivers. It can also contribute to regional-level programs about forest conservation, climate adaptive measures, along with sustainable land-use planning and land management using scientific evidence and data-driven knowledge of the province.