Land Use Land Cover Change Prediction of Tansen Municipality Using Multi-Layer Perceptron-Markov Chain (MLP-MC) Model

Abstract

LULC is dynamic across all-time series, and precise modelling of LULC dynamics can contribute to more effective planning for a sustainable future. This study aimed to examine the LULC change from 2003 to 2023 and further predict the LULC dynamics of 2033 for Tansen Municipality. Random Forest algorithm in Google Earth Engine (GEE) was used for the supervised image classification for three different years, i.e., 2003, 2013 and 2023. Land Change Modeler (LCM) of TerrSet was used for model development, prediction & its validation. Based on the Cramer's V values, candidate explanatory variables demonstrating a higher association with LULC transitions occurring between 2003 and 2013 were subsequently incorporated into the predictive model's construction. The developed model was then used to predict the LULC map of 2023. After model validation, LULC map for year 2033 was predicted using MLP-MC method. Overall, from 2003 to 2023, forest area continuously decreased, losing a total of 952.38 ha, while agricultural land, barren land and built-up areas steadily increased, by gaining a total of 412.11 ha, 336.18 ha & 206.64 ha respectively. However, the trend of water was unpredictable with a slight decrease of 3.58 ha. Comparing the LULC of 2003 & prediction for 2033, it is predicted that forest area will gradually decline by a total of 25.16%.  Interestingly, water area is expected to remain constant with slight increase of 1.57%. But, agricultural land, barren land and built-up areas are projected to increase by 12.53%, 42.15% and 2175% respectively with a boost by the end of 2033. This model is based on the business-as-usual scenario and appropriate interventions can be implemented to reverse undesirable LULC changes and move towards a more sustainable future. This study offers valuable insights into both current & future land use dynamics, aiding policy makers & land use planners in developing better land use management plans.