Improved fire radiative energy estimation in high latitude ecosystems

Abstract:

Scientists, land managers, and policy makers are facing new challenges as fire regimes are evolving as a result of climate change (Westerling et al. 2006). In high latitudes fires are increasing in number and size as temperatures increase and precipitation decreases (Kasischke and Turetsky 2006). Peatlands, like the large complexes in the Alaskan tundra, are burning more frequently and severely as a result of these changes, releasing large amounts of greenhouse gases. Remotely sensed data are routinely used to monitor the location of active fires and the extent of burned areas, but they are not sensitive to the depth of the organic soil layer combusted, resulting in underestimation of peatland greenhouse gas emissions when employing the conventional ‘bottom up’ approach (Seiler and Crutzen 1980).

An alternative approach would be the direct estimation of the biomass burned from the energy released by the fire (Fire Radiative Energy, FRE) (Wooster et al. 2003). Previous works (Boschetti and Roy 2009; Kumar et al. 2011) showed that the sampling interval of polar orbiting satellite systems severely limits the accuracy of the FRE in tropical ecosystems (up to four overpasses a day with MODIS), but because of the convergence of the orbits, more observations are available at higher latitudes.

In this work, we used a combination of MODIS thermal data and Landsat optical data for the estimation of biomass burned in peatland ecosystems. First, the global MODIS active fire detection algorithm (Giglio et al. 2003) was modified, adapting the temperature thresholds to maximize the number of detections in boreal regions. Then, following the approach proposed by Boschetti and Roy (2009), the FRP point estimations were interpolated in time and space to cover the full temporal and spatial extent of the burned area, mapped with Landsat5 TM data. The methodology was tested on a large burned area in Alaska, and the results compared to published field measurements (Turetsky et al. 2011).