Production Method of Reference Data Focusing on Quality Assessment for Global Land Cover Map Validation using the Degree Confluence Project Information

Abstract

A reliable reference dataset is required for the validation of global land cover products. Aerial photographs, high resolution satellite images, and ground truth datasets have been used to produce reference data for the validation of global land cover maps. Recently, validation methods were proposed using land cover information from the Degree Confluence Project (DCP). In this study, a method is developed to produce high-quality reference data, using the information from DCP to validate global land cover products from the International Geosphere-Biosphere Programme (IGBP) land cover class scheme. Quality level of reference data is determined by assigning each label to three interpreters and combining their common interpretations with four land cover types obtained from Google Earth views. The MODIS IGBP global land cover products' validation results indicate that the level of quality level 1 agreement is the highest of the three quality levels and accuracy decreases with the quality level.