ASSESSMENT OF A MULTI-SENSOR APPROACH FOR NOISE REMOVAL ON LANDSAT-8 OLI TIME SERIES USING CBERS-4 MUX DATA TO IMPROVE CROP CLASSIFICATION BASED ON PHENOLOGICAL FEATURES

Abstract

In this work we investigated a method for noise removal on Landsat-8 OLI time series using CBERS-4 MUX data to improve crop classifi cation. An algorithm was built to look to the nearest MUX image for each Landsat image, based on an user defi ned time span. The algorithm checks for cloud contaminated pixels on the Landsat time series using Fmask and replaces the contaminated pixels to build the integrated time series (Landsat-8 OLI + CBERS-4 MUX). Phenological features were extracted from the time series samples for each method (EVI and NDVI original time series and multi sensor time series, with and without fi ltering) and subjected to data mining using Random Forest classifi cation. In general, we observed a slight increase in the classifi cation accuracy when using the proposed method. The best result was observed with the EVI integrated fi ltered time series (78%), followed by the fi ltered Landsat EVI time series (76%).