Detection of Land Use Change and Future Prediction with Markov Chain Model in a Part of Narmada River Basin, Madhya Pradesh

Abstract

Landuse and land cover change have significant impact on the environment of a river basin and has gained considerable attention. It has a strong effect on the surroundings where increasing agriculture as well as urban areas has led to the rapid deforestation and changes in the ecology. Present study involves detection of landuse and land cover change in a part of Narmada river of Madhya Pradesh where rapid changes such as irrigation planning is leading to changes in the land cover. Hence, change detection in the present landform and probable changes in the near future is required for planning and management. Landsat images of 1990 (TM), 2000 (ETM⁺) and 2011 (LISS-III) were used for the classification and future landuse prediction. Supervised Fuzzy C-Mean classification was applied to generate major five classes of water body, built-up area, natural vegetation, agricultural land and fallow land. Overall accuracy for all images was above 85 %. The Markov Chain model was used for prediction. The classified Landsat images of 1990 and 2000 were used to predict the 2011 landuse with Markov Chain which was again validated with the 2011 classified image. The prediction of 2020 and 2030 land use were done to see the future change. The spatial accuracy achieved for the prediction was about 92.5 %. The results illustrate an increase in agricultural land and urban area with the decrease in natural vegetation.