Comprehensive Remote Sensing. Article title: 092 Accuracy and Area Estimation

The quality, quantity, and availability of remote sensing data and products are greater now than in any other point in time. Free data policies and increasing computing power have made it easy and inexpensive to acquire satellite imagery and make maps. Still, while the need to better understand a changing Earth is greater than ever, if remote sensing is to reach its full potential, greater attention to map quality, bias, and uncertainty using protocols rooted in statistical inference theory is required. In this context, maps are not simply subject to an accuracy assessment where the end-product is a measure of map accuracy but integral components of an inference framework that yields estimates of study area parameters such as areas of thematic categories and area bias and uncertainty. Hence, maps are essential in the protocols that generate estimates of parameters rather than providing such parameters directly. This chapter outlines the development of statistical inference within the remote sensing science and application communities. It reviews various approaches to statistical inference, accuracy assessment, and area estimation.