Accuracy Assessment Report - the USGS

IntroductionIn an effort to catalog and map the biodiversity of the United States, in 1994 the National ParkService (NPS) and the U.S. Geological Survey (USGS) embarked on a collaborative vegetativemapping project with the goal of mapping 230+ National Park units (ESRI et al. 1994). As partof this national mapping initiative, a digital vegetation map of Kings Mountain National MilitaryPark (KIMO) was created in 2004 by the University of Georgia Center for Remote Sensing andMapping Science (Jordan and Madden 2008), in consultation with NatureServe. The mappingeffort included collection of field data, aerial photograph interpretation, and polygon attributionto GIS maps.Kings Mountain National Military Park is located in Cherokee and York counties, SouthCarolina (just southwest of Gastonia, North Carolina), on rugged terrain that served as animportant Revolutionary War battlefield. The park’s approximately 1,597 hectares (3,946 acres)are composed primarily of acidic, second growth oak forests on steep ground and successionalforests on old agricultural land. Many of the forested stands are young and currently recoveringfrom past disturbances such as cultivation and logging (White and Govus 2005). KingsMountain National Military Park occurs in the Kings Mountain subregion of the Piedmontecoregion (Griffith et al 2002). Vegetation at Kings Mountain was mapped and classified to theassociation level using the United States National Vegetation Classification (Grossman et al.1998, White 2005), following NPS guidelines. The minimum mapping unit (MMU) was 0.5hectare.The accuracy assessment assigns a measure of validity to the map product and allows users tounderstand the reliability with which the mapped vegetation classes capture conditions on theground. Knowing the accuracy of the map will enable potential users to determine the suitabilityof the map for any particular application (ESRI et al. 1994). This report describes the methodsused in the accuracy assessment and the results for each map class.MethodsOur team assessed the thematic accuracy of the map by comparing the vegetation type shown onthe map to the vegetation type identified on the ground from a representative sample ofevaluation points. When polygons representing vegetation types are mapped and labeled with thecorrect community types, then the map has high thematic accuracy.For each map class, we evaluated both producer’s and user’s accuracy. User’s accuracy is aprediction of the percentage of points mapped as a certain type, which are confirmed to belong tothat mapped vegetation type when visited in the field. In other words, user’s accuracy is ameasure of the reliability of the map to predict what is found on the ground (i.e. how likely themap user is to encounter correct information while using the map). Producer’s accuracy is thepercentage of points observed to be of a given vegetation type in the field that are correctlymapped to that type. In other words, producer’s accuracy is a measure of the reliability of theaerial photo interpretation to distinguish the vegetation types (i.e. how well the map maker wasable to represent the ground features). In addition to the user’s and producer’s accuracy, thisNatureServe KIMO AA 7