Vegetation Classification and Mapping at Fredericksburg and Spotsylvania National Military Park

Taverna, Kristin

This report documents the classification and mapping of vegetation and other land-use classes at Fredericksburg and Spotsylvania National Military Park. Twenty-one map classes representing 19 USNVC associations, two nonstandard, park-specific classes, and three Anderson Level II land-use categories were used to map the 3,157 ha (7,800 ac) park. Vegetation map classes were determined through extensive field reconnaissance, data collection, and analysis. Aerial photography from 2002 served as the base map, and field sampling was conducted from 2002–2005. Spatial data were digitized onscreen over digital orthophoto mosaics created from scanned color infrared, stereo pair 1:6,000 scale aerial photography using a 0.5-ha minimum mapping unit. Ten map classes represent later successional forests and cover approximately 48% (1,510 ha [3,731 ac]) of the park. These map classes can be broadly characterized based on different environmental settings, such as upland forests, alluvial floodplain forests, and non-alluvial wetlands. Early successional or transitional vegetation covers 31% (966 ha [2,387 ac]) of the land in the park. Cultural map classes cover 21% (680 ha [1,680 ac]) of the park, and include the Anderson land-use categories and other man-made or maintained areas in the park. A vegetation map of Fredericksburg and Spotsylvania National Military Park was created following the USGS-NPS Vegetation Mapping Program protocols. Vegetation map classes were crosswalked to the Natural Communities of Virginia and to the USNVC in order to provide a regional and global context for the park’s vegetation. The thematic accuracy of the vegetation map was assessed using a stratified random sampling approach, distributing the sampling effort across the map classes. The number and distribution of accuracy assessment sampling points was determined by map class abundance and frequency as recommended in the USGS-NPS Vegetation Mapping Program protocols, with one regional modification to accommodate the vegetation pattern seen in the park. Two hundred and seventeen randomly selected accuracy assessment sampling points were collected throughout the park. The map was edited based on initial accuracy assessment results to better represent the vegetation in the park. The Kappa Index for the final vegetation map was 78.3%±4.8%, with an overall accuracy of 81.1%.