Hyperspectral Forest Fuel Assessment in the South Platte, Colorado
Abstract
There is strong evidence that many current forests are denser and more vulnerable to catastrophic fires than they were historically. In 2002 fire season approximately 3,000 structures were damaged or destroyed in nearly 67,000 forest fires that burned nearly 2.67 million ha in the U.S. Those fires emphasized their importance in the urban/wildland interface and their dramatic effects on ecosystem carbon storage. Therefore it is important to assess regional fire probability in order to appropriately minimize these risks with fuel treatments. Fire risk is controlled by a series of factors, and among which, fuel loading is a critical one that could be managed through fuel removal. Fuel loading describes the characteristics and quantities of biomass, which can be broken down in to several size-classes, normally by vegetation types and forest canopy cover levels. Fuel loading is widely mapped using aerial photos and multi-band satellite images, with intense field sampling. However, the Airborne Visible and Infrared Imaging Spectrometer (AVIRIS) data, with detailed spectral signatures on plant features such as liquid water thickness, vegetation greenness and cellulose contents provide great potential for separating dominant vegetation, live/dead material and canopy cover based on the variations in plant chemicals. In this study we used high altitude AVIRIS data and field spectral measurements to separate fuel loading classes using various hyperspectral unmixing techniques in the Upper South Platte, Colorado. First, we performed minimum noise fraction (MNF) transformation and pixel purity index (PPI) analysis to pickup endmembers with pure spectral signatures. Then, we examined the resulting classes with their key spectral features, and compared them with field spectral measurements. Finally we classified fuel loading in the area into 17 categories, which include 5 canopy cover levels for Ponderosa pine and Douglas-fir. It was indicated in our AVIRIS derived fuel loading map that 13.2% of the study area is over 60% canopy cover with high fuel loading. We validated our remote sensing analysis in an USFS fuel treatment area. It was shown that AVIRIS derived liquid water thickness and vegetation index were strongly related to green leaf biomass and canopy cover respectively in conifer stands, while cellulose content reflected total carbon (live or dead) stored in vegetation.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2003
- Bibcode:
- 2003AGUFM.B41B0878J
- Keywords:
-
- 0933 Remote sensing;
- 1851 Plant ecology;
- 9350 North America