Accuracy Assessment Resources

This page provides links to accuracy assessment data sets that may be useful to those looking to benchmark remote sensing classifications.  Each data set has a summary page explaining the data attributes, and a link to the raw data.  In some cases, imagery is also provided.

Name (and link to summary data set):  AVIRIS Indian Pine
Area Covered:  2.5 x 2.5 km
Type of scene: Agricultural
Imagery Available: 20m, 220 band AVIRIS data
Contact Person:  Larry Biehl, Purdue University
Comments: AVIRIS data link also provided
How to obtain data:  Downloadable

Name (and link to summary data set):  Urban and Rural Land Cover
Area Covered: Allegheny County, PA; Anne Arundel County, MD; Baltimore City, MD; Baltimore County, MD; Howard County, MD; Jefferson County, VA; Lancaster County, PA; Montgomery County, MD; Philadelphia, PA; Prince George’s County, MD; Syracuse, NY; Washington, DC;
Type of scene:  Mixed urban and rural
Imagery Available: 1-m, 4-band NAIP imagery; lidar
Contact Person: Jarlath O’Neil-Dunne, University of Vermont Spatial Analysis Laboratory
Comments: VermontView maintains a database of high-resolution land cover datasets that have been developed in collaboration with the USDA Forest Service to support Urban Tree Canopy (UTC) Assessments throughout the United States. Complete land cover generated though OBIA then subjected to detailed, manual editing. Overall accuracy ranges from 94% to 99%
How to obtain data:  Downloadable
  

Name (and link to summary data set):   SW Regional Gap Analysis Reference Data
General Location: Arizona, Colorado, Nevada, New Mexico, Utah
Area Covered: 5-state region
Type of scene: Western US
Imagery Available: Landsat imagery from USGS EarthExplorer
Contact Person: Doug Ramsey, Utah State University Remote Sensing/GIS Laboratory
Comments: 93,000 SWReGap field training sites. Data were collected by each collaborating state for their respective geographic area. Field data were collected between 2000 and 2004 throughout a 5-state region.
How to obtain data: Downloadable