A13G-3265:
Mapping Fuel Load and Its Dynamics in Shrubland and Grassland Using eMODIS Data

Monday, 15 December 2014
Hua Shi, InuTeq, USGS EROS Center, Sioux Falls, SD, United States, James E Vogelmann, USGS EROS Center, Sioux Falls, SD, United States, Xiaoyang Zhang, Geographic Information Science Center of Excellence, Brookings, SD, United States, Todd J Hawbaker, US Geological Survey, Lakewood, CO, United States, Matthew C Reeves, US Forest Service Missoula, Forestry Sciences Laboratory, Missoula, MT, United States and Zhengpeng Li, Cooperative Institute for Climate and Satellites University of Maryland, College Park, MD, United States
Abstract:
A challenge for mapping live fuels and their dynamics in rangeland ecosystems is separating fuel conditions and changes from background variation. Rangelands are characterized by seasonal and inter-annual spectral variation associated with phenology and other factors. Remote sensing systems for assessing these ecosystems, such as the Moderate Resolution Imaging Spectroradiometer (MODIS), enable characterization of intra- and inter-annual spectral conditions across all seasons. Fuel conditions in rangelands are influenced by major disturbances such as fires, and phenological changes are strongly influenced by weather and climate. Fuel load data are key components to simulating areas that are more likely to burn and burn severity, and to provide regional estimates biomass burning emissions. We developed an approach for mapping live fuel load, biomass density, dynamics using enhanced MODIS (eMODIS) data at a spatial resolution of 250m. Generalized allometric models were developed to estimate branch biomass and aboveground biomass based on independent variables of maximum foliage biomass. Temporal leaf area index (LAI) data were converted from weekly eMODIS Soil Adjusted Vegetation Index (SAVI) data for various vegetation types and fitted to a vegetation growing model to reduce uncertainty. Results from the study area (Owyhee and Twin Falls) will be examined and validated by using high-resolution aerial photography (WorldView) and field data. Additionally, we analyzed and summarized the correlations among fire occurrence (frequency), burn severity, live fuel load, and climatic conditions. Our results demonstrate that mapping fuel load and its dynamics in shrubland and grassland with remotely sensed time series data can capture spatiotemporal heterogeneity in nonforest live fuel types relevant to fire disturbances and climate/weather variation. This work provides a useful and powerful mapping approach that can improve and augment existing LANDFIRE fuels data in shrubland and grassland. This approach can be used to monitor changing fuel conditions and meet the needs in response to management activities and climate change.