Sampling for Thematic Map Accuracy Testing

Abstract

The cumulative binomial distribution is used in thematic map accuracy testing to develop (1) the minimum sample size needed to validate the accuracy for each category with specified confidence, and (2) the critical level, which is used as the criterion for determining whether the identification from remotely sensed data of a thematic category meets a specified accuracy. The algorithm used in map accuracy testing employs a stratified, systematic, unaligned sampling technique based on the map as a whole, and an additional random sample of points for under-represented categories from all points in that category. Such a technique is implemented by a computer program prepared to select the sample from the Geographic Information Retrieval and Analysis System file. The estimated overall accuracy of the map, and the estimated accuracy for each category of the map, are considered, with associated confidence limits.