Prediction of Land use and Land cover using Google Earth Engine (GEE) and Machine Learning Algorithms

Spatiotemporal distribution of Land Use and Land Cover changes, in addition to a variety of socio-ecological concerns, including the links between human activity, physical surroundings, and decadal shifts in LULC caused by anthropogenic activities, are all important topics that can be studied using satellite imagery. This study explores the spatial-temporal assessment of land features and sensitive ecosystem for 2013-2023, with a focus on predicting future land characteristics. Using the CA-Markov chain model, the research forecasts potential changes to the landscape and evaluates their impact on changes of Land features. Images were classified into four classes using Support Vector Machine (SVM), Random Forest (RF), Naïve Baye (NB), and Classification and Regression Tree (CART). After that, kappa statistics as well as overall accuracy techniques were used to gauge their accuracy. This study used the northern region of India, Dehradun, as a case study. Three Landsat-8 Images were used in this study which is obtained in 2013, 2018 and 2023. The satellite images were obtained from United States Geological Survey (USGS). Finally, this paper reveals that the land features has been identified from 2013-2018 and 2018-2023 and predicted the next 5 years using different social, environmental and physical parameters. From 2013-2018 Vegetation are decreased from 19.53% to 14.33% and the buildup areas are decreased from 11.1% to 5.53% within 5 years interval while from 2018-2023 Vegetation are decreased again from 14.33% to 9.76% while buildup areas increased from 5.53% to 7.37% within 5 years interval. If this trend continues, it looks that vegetation areas will permanently lose their unique traits and ecosystems, having a severe influence on Dehradun meanwhile the buildup areas are predicted to increase within the next 5 years.