PARENT SESSION
Poster Session 2: Forest Ecology
Monday, August 8, 5:00 PM - 6:30 PM, Exhibit Hall 220 A-E, Level 2, Palais des congrès de Montréal
Assessing the spatial patterns of forest fuel using spectral unmixing of vegetation components in the Front Range of Colorado.
Jia, Gensuo*,1, Burke, Ingrid1, Kaufmann, Merrill2, 1 Forest, Rangeland, and Watershed Stewardship, Fort Collins, CO 80523, USA2 USDA Forest Service, Fort Collins, CO 80526, USA
ABSTRACT- Montane coniferous forest and woodland are the dominant ecosystems in the Colorado Front Range. Fuel loading and carbon storage, both related to wildland fires, are heterogeneously distributed in these ecosystems. Considering the size of the patches and the mosaic of fuel materials, high spectral and spatial resolution estimates of vegetation components and fuel types are needed to improve fire risk assessment and regional carbon balance assessment, especially around the urban-wildland interface. We used high altitude AVIRIS imagery, and performed mixture tuned matched filtering (MTMF) and spectral fitting techniques to spectrally unmix major forest components and fuel types in the Colorado Front Range by discriminating the fractional covers of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV) and bare soil at a sub-pixel level. Based on the fractional cover of these fuel components, we were able to assess the spatial patterns of fuel loading and carbon budgets at landscape scale. In the study areas, PV fractions are dominant (48.7%), followed by NPV (28.8%) and soil (22.5%). There are high fractions of PV on north facing slopes and in old growth forest areas, while bare soil fractions are high in low elevations and in recently burned areas such as those around the Cheesman Lake. There are generally high fractional covers of live fuels in all the experimental areas. However, due to microclimate, vegetation disturbances such as fire and forest management practices, the spatial distribution of fractions are highly heterogeneous. Fuel treatments have greatly reduced the fraction of PV and canopy cover, and have resulted in relatively high fractions of NPV and bare soil. With accurate estimates of spatial distribution of each fuel components, our results greatly improve the accuracy of regional carbon assessment efforts. Spatially-explicit images also permit us to evaluate the effects of current fuel treatment efforts throughout the region, an otherwise extremely labor intensive process.
Key words: fuel, spectral unmixing, conifer forest, Front Range