Professional Documents
Culture Documents
Khlawhon Gis512 Final Project
Khlawhon Gis512 Final Project
Khlawhon Gis512 Final Project
2.
3.
Source
URL
Format
Metadata
LandSat 8 Scene,
Southern California
August 29, 2014
US Geological
Survey
http://earthexplorer.usgs.g
ov/browse/landsat_8/2014/
040/037/LC804003720142
73LGN00.jpg
GeoTIFF
LandSat 8 Scene,
Southern California
September 30, 2014
US Geological
Survey
http://earthexplorer.usgs.g
ov/browse/landsat_8/2014/
040/037/LC804003720142
41LGN00.jpg
GeoTIFF
Digital Elevation
Model-ASTER
Global DEM
US Geological
Survey
http://earthexplorer.usgs.g
ov/browse/aster_dem_v2/
N33/ASTGDEMV2_0N33
W118.jpg
Arc/Info
1-Arc Second
Fire Perimeter
Shape File
National Wildlife
Coordinating Group
http://inciweb.nwcg.gov/inc
ident/4105/
GeoDataBase (.shp
file)
Digitized at 1:12,000
Methods
Before and After False Color Composite comparison
Image enhancements: Pan-Sharpen, Sub-Set, Topo Norm, Standard Deviation
Stretch
Unsupervised Classification helped determine spectral friction points and inform
Supervised Classification
Results of Supervised Classification of Before and After clipped to spatial
extent of the fire in ArcGIS.
Raster Calculator used to convert pixel count to actual spatial extent (Acres)
Before and after attribute tables compared to determine which classes gained or
lost pixels (acreage).
Major Spatial Operations: Supervised/Unsupervised Classification, Raster
Calculation
Additional Spatial Analysis Tools Used: ArcMap- Extract by Mask and Raster
Calculator
Process
Pan
Sharpen
Topo
Normalize
Sub-Set
Std Dev
Stretch
Supervised
Unsupervised
Extract By
Mask
ArcGIS
Field
Calculator
Results
Key Points
Interpretation
A Total of 956-958 acres of forest were consumed by the blaze
Between 556 and 765 acres of Tree Cover and Healthy Herbaceous areas were
consumed by the wildfire.
Only 23 acres of Tree Cover was consumed. The remainder was wild land grasses.
Differentiating between bare earth and unhealthy vegetation within the burn
perimeter proved challenging during classification. Scorching makes it nearly
impossible to delineate features that were once highly reflective.
For a fire this smaller, imagery with higher spatial resolution would have yielded
better results. Even with Pan-Sharpening, and Topo Normalization, resolution was
an issue during classification. Data from IKONOS, WORLDVIEW, QUICKBIRD, or
other high resolution platforms may have improved results.
Temporal Resolution was important in this case. Finding images shortly before and
after the fire allowed for a reasonably accurate estimate of the burned area. The
after image was specifically selected during a period that allowed all the smoke
from the fire to clear. This improved results remarkably.