Modeling Deforestation Trends Using Remote Sensing, Artificial Neural Network Techniques and its Relationship with Socio-Economic Drivers

Abstract

Monitoring temporal forest changes can highlight management gaps that can help Forest department for policy-making and afforestation schemes. Many previous studies done to monitor LULC changes lacked social data. This study used high resolution remote sensing data and social data to examine trends in district Abbottabad. The study's objectives were (1) To determine the temporal trend and drivers of forest cover change (2) To predict the future Land Use / Land Cover for District Abbottabad. Landsat imagery of 2010 and 2020 was classified using supervised classification, more than 500 samples were collected for forest, agriculture, barren and urban class, and 300 samples were collected for water body class. To monitor change in forest cover, area of each class was calculated and compared. A socio-economic survey was carried out using an online questionnaire to evaluate the socio-economic conditions of the population and validated with GIS techniques to identify the drivers behind Deforestation. The cellular automata artificial neural network (CA-ANN) model integrated into the Molusce plugin of QGIS to predict future land use land cover for year 2030. The overall accuracy for the year 2010 and 2020 was obtained as 89% and 85.5%, respectively. Forest cover decreased from 48.6% to 33.3%, a net change of 15.27% in 10 years. A significant increase of 22.27% in agricultural land was observed. Validations shows that population, urbanization and agriculture expansion were 3 major contributing factors to Deforestation. The forecast for Land Use and Land Cover (LULC) suggests that by 2030, the primary land uses in Abbottabad will be built-up areas and agricultural land 653sq.km and 315sq.km respectively. The study recommends that the Forest Department should enforce stringent measures against individuals or activities causing disturbances to forest land.