A comparison of the Bayesian classifier and the Estimates-based algorithm for crop identification by Terra/MODIS 250- m data

Abstract

We studied and compared the Bayesian classifier (BC) and the Estimates-based algorithm (EBA) in this research for crop identification by a set of Terra/MODIS 250-m data. Methods are based on time-series calculated by satellite images. EBA has an undoubted advantage. It uses time series even with missing values whereas BC requires reconstructing missing values of time series a priori. Experiments are based on data obtained for the Samara region in 2014-2016 years. The numerical results are given for the different length of time series to compare two methods throughout the growing season.