IDENTIFICATION OF GHOST VILLAGES IN UTTARAKHAND USING MACHINE LEARNING OF NDVI TIME SERIES

Abstract

Rural depopulation and the phenomenon of abandoned villages known as ‘Ghost Villages’ are the major challenges in the Uttarakhand state of India. Over four million people, or roughly 40% of Uttarakhand's population, left the hilly state, according to the 2011 India census. As a result, more and more villages in Uttarakhand are becoming abandoned, and some districts, like Pauri, Garhwal, and Almora, are experiencing negative population growth. A 2018 study conducted by the Uttarakhand Migration Commission found that 1048 of the state's villages—often referred to as ghost villages have been abandoned since 2011. This research project aims to identify ghost villages within Uttarakhand. There are some potential issues that can be solved after identification of Ghost villages like Urbanization and migration studies, Land use and planning, Policy development, Prediction and early warning system. Remote sensing technology is the important tool for monitoring the abandoned and active croplands because it is very fast and can be used for long time series with large scale observation. Using the help of satellite imagery and geospatial data, the study employs a systematic approach to identify, characterize, and map abandoned settlements within Uttarakhand. This study explores the potential of using time series analysis of the Normalized Difference Vegetation Index (NDVI) to determine the status of a specific village as either an active or abandoned settlement, commonly referred to as a ghost village. This research attempts to identify unique patterns and trends associated with ghost villages, which are defined by the absence of agricultural and human activities, through the systematic analysis using Machine Learning model and multi-temporal NDVI data as an input. For this identification, 400 villages are selected in which 200 are active and 200 are abandoned acquired from 2011 census of India. These data worked as an input to train the machine learning models like Random Forest, Decision Tree and Logistic Regression. For clear prediction of ghost villages, the multi temporal NDVI data is collected in various durations (2 years, 5 years and 10 years). This input assigned in machine learning model, now with the help of testing data the villages are classified as either active or ghost.