A scale-up method for reference data for validation of global land cover maps using ALOS/AVNIR-2 satellite data

Abstract

Validating the accuracy of land cover products using a reliable reference dataset is an important task. Recently, the amount of ground truth data provided by volunteers has increased. Although ground truth data can provide information that can produce reliable reference data, the information is only correct within the observable landscape. It is necessary to scale up reference data derived from ground truth data to match the spatial resolution of the global land cover product. We propose a scale-up method that confirms expanding land cover characteristics for a target position using the occurrence ratio of pixels that meet the criteria for the target position to the total pixels for the target scale. The results of applying the scale-up method to test sites showed that the occurrence ratio method was a better judge of expanding target land cover types than the average method.