Forest Cover Classification
Forest cover maps use satellite imagery to classify and quantify different Forest cover types, including dense forest, moderate forests, open forest and shrubs. By comparing LULC maps from two different time frames, we can identify areas where forest cover has been lost, gained, or remained unchanged (Buřivalová et al., 2021). The analysis helps monitor deforestation, urban expansion, agricultural encroachment, and other land-use changes affecting forests (Dash et al., 2016). It provides valuable insights into the drivers and patterns of deforestation, aiding policymakers and conservationists in making informed decisions to protect and manage forest resources sustainably (Dayal et al., 2020; Naing Tun et al., 2021; Jamaludin et al., 2022). Continuous monitoring using forest cover maps ensures timely responses to potential threats to forests, contributing to global efforts to combat deforestation, conserve biodiversity, and mitigate climate change impacts. As Similipal is a large area of natural vegetation is very difficult to find changes from land use land cover classification (Vineetha et al., 2018). We find changes from the above satellite imageries when processed with google earth engine shown in (Fig.4). We find that there are rapid positive changes for Dense and Moderate forests and negative changes for Open Forest and Non-Forest. The open forest is changed to moderate forest in forest cover data shown in figure (Fig. 3).
Change Detection
We have used google earth engine which is a web based geo-spatial tool for monitoring and calculating the change detection. The tool is accessed with geo spatial codes written in java script for random forest classification, accuracy assessment using confusion matrix & kapa coefficient. As shown in (Fig. 4), the supervised classification in the study area shows greater variations (Brovelli et al., 2020; Prasai, 2022; Fedotova & Gosteva, 2021).
The forest area calculated in the 2015 image was 2,70,584 hectares, but in the 2022 image, the forest area calculated 2,77,184 hectares. There is a positive change across the study area, which is 2.438% of the forest cover in the 2015 forest cover classification, which is 6597.01 hectares in area.
Forest Cover Prediction
We have used the Google Earth engine to predict the future forest cover of the study area in 2030. First, we added training data sets to the Google Earth Engine, which are supervised classification data of the study area through support vector machine classification in GIS software (S.Panchal & Sharma, 2016; Nartišs & Melniks, 2023). Then, using geo-spatial coding in a Java script (Zhu, 2022), which is a programming language of Google Earth Engine, we have predicted the future forest cover pattern of the study area, which is calculated from previous class and training datasets of 2015 and 2022 forest cover classification. Our web-based geo-spatial tool provides future forest cover data with a 99.9% accuracy value with proper visualisation (Fig. 5).
After the collection of all tabular data from Google Earth Engine (Gautam & Rai, 2022; Patil & Panhalkar, 2023; OLIVEIRA et al., 2019) we are visualising the forest land use land cover map in the graph given below (Fig. 6). The time versus area graph of the study area shows that dense forest was calculated at 1,20,256.95 hectares in 2015 and dense forest was calculated at 1,70,656.19 hectares in 2022, which is a positive change for the evergreen woodlands. Also, the moderate forest and non-forest are 6,791.88 and 15,316.7 hectares in 2022, respectively; the same will be 84,952.31 and 1,420.04 hectares in 2030, respectively, which represents a positive prediction for the evergreen forests of Similipal.