Application of multi layer perceptron neural network Markov Chain model for LULC change detection in the Sub-Himalayan North Bengal

Abstract

Land use and land cover (LULC) changes may occur due to natural and anthropogenic activities. The present study area is situated in the Himalayan foothill region of north-eastern India and covers around 1507 km2. This area comprises Siliguri Municipal Corporation (SMC) and the surrounding five community development blocks. Geopolitical location of the study area makes it significant as it is exposed to the negative impacts of human and commercial activities across the region. The present study aims to identify the LULC classes in the study area from 1991 to 2021 with their change dynamics and predict the LULC changes in 2050. Apart from these, a perception study is also carried out to identify the major drivers of LULC changes in the study area. After delineating the study area, supervised classification of Landsat satellite images was done applying the maximum likelihood classification (MLC) tool for the year 1991, 2001, 2011 and 2021. Each image was classified into six LULC classes i.e., vegetation, plantation, agricultural land, built up, water bodies and fallow land and the overall accuracy (average) of the classified maps is 88.83%. The result shows that vegetation, agricultural land and fallow land are decreased from 23% to 20%, 35%–28% and 23%–19% respectively and plantation and built up areas are significantly increased from 14% to 24% and 2%–8% respectively. Then the land change modeler (LCM) under multi layer perceptron neural network Markov Chain (MLPNN-MC) model was applied to predict the spatio-temporal LULC changes from 2021 to 2050. The result reveals that in 2050 built up areas will be increased significantly and are expected to cover almost 12% of the total study area and only 24% of agricultural land will be left. The model was validated by Pearson's chi-squared test and found no significant distinguishes between simulated and actual classified maps. For identifying the main drivers of the LULC changes in the study area people's perceptions were taken from 25 focus group discussions (FGD) randomly over the area. The result reveals that physical and economic location, immigration, development of transport network, increasing multi-functional land use, expansion of small trade and enterprises, and cutting of trees for fuel wood and construction purposes are the key drivers of LULC change for this particular region. The change in the land surface in this Sub-Himalayan region is threatened by unplanned urbanization and unorganized land use practices so, crucial measures are needed indeed.