Abandoned field determination using MODIS data

Abstract

After the collapse of the USSR, a significant area of unused land was formed on the territory of the countries that were part of the union. Nowadays, abandoned fields have become a valuable resource for investment in agricultural production. It makes the task of identifying unused arable land very urgent. Remote sensing of the earth makes it possible to search for unused lands in an automatic mode and greatly simplifies the monitoring of unused arable land in comparison with traditional ground surveys. This article is about identifying unused arable land from MODIS images captured using Terra and Aqua spacecrafts. The article describes a method for detecting unused arable land based on an object-oriented approach using features based on the PVI vegetation index and a Bayesian classifier. The classification was tested within the fields of the Samara region and showed a classification accuracy of 90%. The results of the classification can be used as input data to identify lands suitable for agricultural activities.