Hyperspectral remote sensing of fire induced water repellent soils

Lewis, S.A. 2003. Hyperspectral remote sensing of fire induced water repellent soils. Master of Science in Engineering thesis. Pullman, WA: Washington State University. December 2003. 111 p.

Keywords: hydrophobic soil, water repellent soil, hyperspectral, water repellency, Hayman Fire, Colorado

Links: PDF [3.1 MB]

Abstract: The summer of 2002 was one of the worst fire seasons in history, especially for the state of Colorado where the Hayman Fire burned nearly 60,000 ha. Almost a third of the fire was classified as a high severity burn, which often translates into a high erosion potential. After a fire, soils are frequently rendered water repellent due to the burning of vegetation and surface organic matter. Two methods of testing water repellent soils were performed on the Hayman Fire, the traditional water drop penetration time (WDPT) test and a new mini-disk infiltrometer test. The ability of these two methods to identify water repellent soils in relation to burn severity was tested as well as the compatibility between the tests. The moderately burned sites exhibited the strongest and most persistent water repellency according to both WDPT and infiltrometer tests. The WDPT and infiltrometer values were correlated for each individual burn severity class as well as overall. As the infiltrometer is still in experimental stages, we recommend both tests be used for method comparison and evaluation.

After evaluation of the ground data, a hyperspectral image that was acquired after the fire was analyzed in an attempt to remotely identify soil water repellency. Remote detection of organic matter in soils has been studied extensively and wavebands that were previously identified as useful were analyzed for spectral features indicative of water repellency. No features were found to correlate well with soil water repellency, mostly because of the overall dampening the blackness from the fire had on soil spectral signatures. A supervised classification was run with the mean spectral signatures of low, moderate and high water repellency soils as endmembers. The accuracy of the classification for identifying the degree of severity of soil water repellency was low, but the ability to identify the presence of soil water repellency was nearly 80 percent. According to the classification, approximately 20 percent of the Hayman Fire was at a high risk for soil erosion, and these are the areas on which erosion mitigation should be focused.

Moscow FSL publication no. 2003m