This raster dataset depicts percent canopy cover derived from 1-m conifer classifications when aggregated to 30-m cells.

Conifer features were classified from 2010, 2012, and 2013 NAIP Digital Ortho Quarter Quads (DOQQ) using the Feature Analyst 5.0 extension for ArcGIS 10.1. Tiles were organized and grouped by Nevada Department of Wildlife Population Management Unit (PMU) locations, plus a 10 km area beyond the PMU extent. Analysts visually identified conifers in the imagery using false color infrared settings and digitized multiple trees per tile as training locations for classification. After performing hierarchical learning and clutter removal with Feature Analyst to remove non-conifer features on output shapefiles, the conifer polygons underwent post-processing steps to improve results. This included removing small polygons and masking of vegetation that were frequently misclassified (riparian, mountain mahogany, and sagebrush). Cleaned feature classes were converted to 1 x 1 meter cell binary rasters, then mosaicked by PMU in Erdas Imagine.

Individual PMUs were aggregated to 30 m cells and populated with the calculated values for the percentage of canopy cover from all 1 m cells that occurred within them. These 30 m percent cover rasters were then mosaicked to create a single raster for the entire study area.

Cells values range from 0 to 100%.

This raster dataset depicts percent canopy cover derived from 1-m conifer classifications when aggregated to 30-m cells.

Conifer features were classified from 2010, 2012, and 2013 NAIP Digital Ortho Quarter Quads (DOQQ) using the Feature Analyst 5.0 extension for ArcGIS 10.1. Tiles were organized and grouped by Nevada Department of Wildlife Population Management Unit (PMU) locations, plus a 10 km area beyond the PMU extent. Analysts visually identified conifers in the imagery using false color infrared settings and digitized multiple trees per tile as training locations for classification. After performing hierarchical learning and clutter removal with Feature Analyst to remove non-conifer features on output shapefiles, the conifer polygons underwent post-processing steps to improve results. This included removing small polygons and masking of vegetation that were frequently misclassified (riparian, mountain mahogany, and sagebrush). Cleaned feature classes were converted to 1 x 1 meter cell binary rasters, then mosaicked by PMU in Erdas Imagine.

Individual PMUs were aggregated to 30 m cells and populated with the calculated values for the percentage of canopy cover from all 1 m cells that occurred within them. These 30 m percent cover rasters were then mosaicked to create a single raster for the entire study area.

Cells values range from 0 to 100%.