An approach for quickly labeling land cover types for multiple epochs at globally selected locations

This paper presents a sampling-based land cover labeling approach for efficiently producing high accuracy land cover reference data over large areas. The accuracy of the interpreted land cover data is maximized by associating the sample locations with multiple independent sources of input, such as Landsat images acquired in different seasons and epochs, high-resolution images from Google Maps, and in-situ “ground” photos from web sources. The approach can be adopted for large areas-e.g., countries, continents, or the entire globe-to provide reference data samples for classification model training and validation. An online land cover labeling system has been prototyped at the Global Land Cover Facility (GLCF) at the University of Maryland.