Delineation of Vegetation Using Satellite Imagery, Clark County
DigitalGlobe’s QuickBird satellite high-resolution multispectral imagery was classified by using Visual Learning Systems’ Feature Analyst feature extraction software to produce land-cover data sets for the Red Rock Canyon National Conservation Area and the Coyote Springs, Piute-Eldorado Valley, and Mormon Mesa Areas of Critical Environmental Concern in Clark County, Nevada. Over 1,000 vegetation field samples were collected at the stand level. The field samples were classified to the National Vegetation Classification Standard, Version 2 hierarchy at the alliance level and above. Feature extraction models were developed for vegetation on the basis of the spectral and spatial characteristics of selected field samples by using the Feature Analyst hierarchical learning process. Individual model results were merged to create one data set for the Red Rock Canyon National Conservation Area and one for each of the Areas of Critical Environmental Concern. Field sample points and photographs were used to validate and update the data set after model results were merged. Non-vegetation data layers, such as roads and disturbed areas, were delineated from the imagery and added to the final data sets. The resulting land-cover data sets are significantly more detailed than previously were available, both in resolution and in vegetation classes.
Two publications have been released as part of this study:
Smith, J.L., Damar, N.A., Charlet, D.A., and Westenburg, C.L., Land-cover mapping of Red Rock Canyon National Conservation Area and Coyote Springs, Piute-Eldorado Valley, and Mormon Mesa Areas of Critical Environmental Concern, Clark County, Nevada: U.S. Geological Survey, Scientific Investigations Report 2014–5076, 42 p., http://dx.doi.org/10.3133/sir20145076.
Charlet, D.A., Damar, N.A., and Leary, P.J., 2014, Vegetation database for land-cover mapping, Clark and Lincoln Counties, Nevada: U.S. Geological Survey, Data Series 827, 18 p., http://dx.doi.org/10.3133/ds827
The Bureau of Land Management (BLM) administers more than half of the approximately 2 million hectares (5 million acres) of land in Clark County, Nevada. Federal lands are often designated for specific uses or protection, requiring management plans tailored to those designations. In 1990, the Red Rock Canyon National Conservation Area was designated as Nevada’s first National Conservation Area to protect and preserve geological, archaeological, ecological, cultural, scenic, scientific, wildlife, riparian, wilderness, endangered species, and recreation resources. The Coyote Springs, Piute-Eldorado Valley, and Mormon Mesa Areas of Critical Environmental Concern (ACEC) were designated to protect critical habitat for the desert tortoise (Gopherus agassizii).
Regional and national land-cover and vegetation data sets are available that include these areas. These data sets are based on mid-resolution (30-meter) satellite imagery and delineate generalized vegetation communities. The available resolution and level of vegetation detail are not sufficient to support local resource management in these areas.
In 2006, USGS, in cooperation with the BLM, began a 4-year study of the Red Rock Canyon NCA to produce a land-cover data set from DigitalGlobe’s QuickBird high-resolution (2.4-meter) satellite imagery and field vegetation data. In 2010, the study was extended to include the Clark County portion of Mormon Mesa, and Coyote Springs and Piute-Eldorado Valley ACECs. Visual Learning Systems’ feature extraction software, Feature Analyst, was used to classify the imagery on the basis of detailed field data (Visual Learning Systems, Inc., 2008). A minimum mapping unit of 0.1 hectares (0.25 acres) was used, or about one-quarter the size of previously available data. Vegetation data collection included more than 1,000 field samples that were used as possible training sites in the feature extraction process. Sampling protocol and naming conventions for the field data collection followed the National Vegetation Classification Standard, Version 2 (NVC; Federal Geographic Data Committee, 2008). The imagery and field data were used to map vegetation communities at the stand level, in which the dominant species in each canopy layer are identified. Non-vegetated data layers were also derived from the imagery.