Vegetation Classification, Distribution, and Mapping Report Tumacácori National Historical Park

Drake, Sam; Buckley, Steve; Villarreal, Miquel; Studd, Sarah; Hubbard, Andy

In 2007–2008, the National Park Service, Sonoran Desert Network Inventory and Monitoring Program, in cooperation with the Arizona Remote Sensing Center (University of Arizona, Offi ce of Arid Lands Studies), carried out classification and mapping of vegetation at Tumacácori National Historical Park (NHP), as part of the national U.S. Geological Survey–National Park Service Vegetation Characterization Program. The primary objective of the program is to produce high-quality, standardized maps and associated data sets of vegetation and other land cover occurring within the parks (http://science.nature.nps.gov/im/inventory/veg/index.cfm). In particular, the aim of this project was to create a vegetation map at the National Vegetation Classification alliance level or finer, with a minimum mapping unit of 0.5 hectares, thematic accuracy of 80% or better per map class, and spatial accuracy meeting U.S. National Map Accuracy Standards. Project scoping, initiated in October 2007, involved the project team and Tumacácori NHP staff . Quick-bird satellite imagery, acquired in August 2006, was used for this project, which covered the three subunits of the park plus a 100-m buffer around each one, for a total area of 238 hectares. Image preprocessing and initial interpretation to the vegetation formation level were done at the Arizona Remote Sensing Center. The draft formation-level map was produced through limited field reconnaissance and visual interpretation of the pan-sharpened imagery and heads-up digitizing in ArcGIS to delineate polygons based on vegetation physiognomy. Vegetation classification and mapping to the alliance level were done simultaneously, through intensive field work from November 2007 to April 2008. The field team, including the image interpreters, took the formation map as a starting point and verified or modified formation boundaries; split formation polygons into alliance- or fi ner-level, floristically based polygons; and characterized each alliance type with quantitative (plot- and polygon-based) data on species composition, relative cover, and other factors. Because Tumacácori is a small park, a census of the entire park was done in this way. Nineteen vegetation types were identified and mapped. Thematic accuracy was assessed by the park’s resources manager/archeologist after a thorough briefing on the methods used for classification and mapping, as well as on the map classes. The assessor was provided with tools similar to those used by the mappers, including a GPS-linked handheld computer loaded with draft map polygon boundaries, and printed map sections with each polygon labeled with a simple identification number. The assessor completed a census of every mapped polygon, walking through each one to evaluate the lifeform and canopy cover of the dominant species, and choosing the best-fitting map class name for the polygon. Initial accuracy was 90.2%, with four classes below 80%. Because of the small area of the park and the small number of polygons, each discrepancy between the accuracy assessment data and the map data was investigated and resolved, resulting in 100% thematic accuracy. Spatial accuracy was assessed against 2006 digital orthophoto quarter quadrangle imagery, using 20 test points for each of the park’s three units. The maximum absolute error measured was less than 2 m ground distance, and maximum root mean square error was 1.03 m, well within the limits of the National Map Accuracy Standards. While the main products of this project are the vegetation classification and the vegetation map database, a number of ancillary digital geographic information system and database products were also produced that can be used independently or to augment the main products.

Vegetation Classification, Distribution, and Mapping Report Tumacácori National Historical Park

Series

Collections