Ft. Carson is a 137,000-acre military installation located in the high desert of southern Colorado for which F4 Tech assisted with GIS and Remote Sensing based land cover classification.
Ft. Carson consists of two land ownerships, the northern most area being near Colorado Springs and the southern in Pueblo County, with the primary land cover being rangeland and mixed juniper uplands. The northern cantonment area has significant topography and elevation changes with landscapes ranging from fractured red-rock bluffs to sensitive estuarian ecosystems. The southern training area has less topographic change and is mainly mixed juniper uplands. This variety in topographic relief and ecosystem types as well as the large land area to be classified made this a good candidate for software-based supervised classification.
To produce land cover classification on this region, several remotely-sensed data sources were utilized including LandSat (8 bands), National Agricultural Inventory Program (NAIP, 4 bands), and previously collected Lidar. F4 Tech acquired the remotely-sensed data and did all geographic and radiometric processing required to get continuous raster data sets for analysis. From this processed data, supervised training datasets were generated by F4 Tech staff utilizing both imagery and previously collected inventory data. These training sets were used in Esri ArcMap along with the processed imagery to classify forest cover using supervised pixel based classification methods.
The resulting raster, as well as a raster classifying height class from Lidar data, was aggregated using nearest neighbor algorithms and converted to polygons of similar forest type, height class, and canopy cover. These polygons were then ground verified using new onsite data collection and quality checked by F4 Tech staff.
Finally, previously collected inventory data, as well as ground-truth data, were used to generate volume tables to estimate, by strata, the standing fuelwood for each polygon.
Deliverables included land cover classification raster, land cover classification polygons, all corrected and mosaiced imagery, and a summary report of methods.
Project Status: This project began in Nov 2013 and ended in July 2014.