Abstract Maps of land cover class are more common, and generally more accurate, than maps of land use because “use” implies management intent that may not be directly sensible by earth-observing satellites. However, many monitoring frameworks related to sustainability require land use and land cover to be explicitly differentiated. This is particularly true for forests, where natural and human-caused dynamics in tree cover often occur independently of long-term land use changes that signal deforestation. We used an extensive multi-temporal, multi-variate sample of reference points across the United States to calibrate and validate 30 m mapped time series (1985–present) of land cover, land use, and vegetation condition change. These maps comprise the Landscape Change Monitoring System (LCMS) and are served through: an interactive, open-access app; Google Earth Engine; image services; and the FSGeodata Clearinghouse. Here, we provide methods, validation metrics, and a usage example highlighting the value of differentiating use from cover in the context of model-assisted estimation of forest area using U.S. Department of Agriculture, Forest Service inventory data.
Housman et al. (Tue,) studied this question.