Change Detection of Chandaka Forest and Analysis of Remote Sensing Data Using Machine Learning and Geographic Information System

Resumo

Change detection is a challenging and significant task in satellite applications due to the unavailability of satellite photos of a given geographic area acquired at various times. This study employs remote sensing data to analyze the changes in the area by utilizing machine learning methods alongside a geographic information system. The satellite images are pre-processed, gathered from multiple sources, and classified according to features. Two of the most significant and efficient factors underlying many ecological systems are land use and land cover maps, and change detection. The integration of remote sensing, Geographic Information System (GIS), and machine learning techniques offers a promising approach for studying and monitoring forest ecosystems. Remote sensing offers a means of collecting large-scale data on forest cover and land use change, while GIS technology enables the analysis and visualization of this data. Machine learning algorithms can be used to classify and analyze the data, allowing researchers to analyze the changes occurring in forest ecosystems on the Google Earth Engine framework. Built-up planning, ecosystems, land resources, crop patterns, agriculture, and natural resources depend on land use and land cover assessment and analysis. Change detection mapping is done by Geographic Information System software. For accuracy assessment, identify the classes of land cover and land use that have changed in the area. This study area, Chandaka Forest, merely uses the Landsat 8-9 OLI/TIRS imageries for land use and land cover mapping. Using the proposed algorithm, the historical images of Chandaka Forest were obtained, and seven different class values were stored in the cloud. Using the data, the change detection has been obtained with an accuracy of 95.1%. A system for cloud computing named GEE implements Support Vector Machine and random forest as its primary classification algorithm models to make two-year interval land use land cover maps for 2014, 2016, 2018, 2020, and 2022. This research is based on machine learning classification plus geographic information system techniques for change detection of satellite images. Analyses of both satellite images captured at two distinct times in a specific geographical region demonstrate the effectiveness of the method. The effectiveness is verified using change detection accuracy, and the results are superior to those of the currently used approaches. This paper contributes to environmental monitoring, such as deforestation of Chandaka Forest, Odisha, using GIS and machine learning classification applied to remote sensing data.