Estimating dry grass biomass residues using AVIRIS image analysis

The amount of dry grass residue remaining in grasslands in the autumn is indicative of land management practices, especially grazing, and for predicting future fire and erosion potentials. This study has examined factors necessary to provide a remote assessment of grassland condition (biomass, patch distribution), based on a linear spectral mixing analysis of Advanced Visible Infrared Imaging Spectrometer (AVIRIS) images calibrated to surface reflectance for an 80 km² area east of Lake Berryessa, California. The fine-scale spectral features that distinguish soils and dry plant material in the shortwave infrared (SWIR) region are thought to be primarily due to the presence of cellulose and lignin which are absent from soils. The authors have used the endmember fractions of dry grass (mixed annual grasses), green vegetation (Heteromeles arbitufolia, Toyon), and soils (Sehorn Clay series), contained in a spectral library of common plant and soil materials of the area for the analysis. Estimated abundance of dry grass biomass were also estimated by lignin and cellulose absorption features around 1.72 μm and compared to endmember fractions and against ground measured biomass samples. The endmember fractions were combined with topographic data, slope, aspect, and cumulative drainage in a geographic information system (GIS) and used to produce a maximum likelihood classification of vegetation of the region. The region contains grazed and ungrazed grasslands, oak woodlands, chaparral, riparian woodlands, agricultural cultivation, and other vegetation types. The prediction accuracy of the vegetation map was estimated to be 0.58 overall, based on comparison against a vegetation map of the University of California´s Stebbins Cold Canyon Reserve (SCCR) which was derived through field-based and photogrammatic interpretation a small area in the AVIRIS image.