Using remotely-sensed burn severity data from modern reference ecosystems as a guide for land management: Describing fire regimes, identifying burn severity levels, and quantifying patchiness

Background/Question/Methods

The high spatial complexity and variable burn severity of mixed-severity fire regimes makes quantifying fire frequency, extent, and severity with traditional methods challenging. In these systems, the landscape pattern created by fire cannot be understood based only on the boundaries of the fire; variation of severity within fires also creates and maintains landscape heterogeneity. Remotely-sensed burn severity data is becoming a commonly-used tool for managers and scientists to describe and predict the severity and spatial pattern of fires. We describe the application of remotely-sensed burn severity indices, specifically those based on the near- and mid- infrared bands of 30-meter Landsat data, to better describe reference conditions for mixed-severity fire regimes.

The fire regimes of modern reference ecosystems—areas with intact fire regimes due to limited historical fire suppression, or to recent increases in area burned—can be used to inform land management actions and decisions. We provide an overview of two main approaches of quantifying the spatial pattern of severity using remotely-sensed burn severity data from reference ecosystems. The first approach focuses on the geographical and ecological context of the fire severity, and the second describes the spatial patterns of burn severity typical for a given ecosystem.

Results/Conclusions

Using burn severity data from mixed-severity fire regimes in the Pacific Northwest, we show that (1) the scale and extent of the area analyzed, (2) methods of classifying severity, and (3) the choice of spatial pattern metrics can all lead to different, but valid, interpretations of the spatial pattern of burn severity for a given ecosystem. Thus, these methodological issues are barriers to comparing results of different studies and transferring those results into management guidelines. We suggest that quantifying the continuous severity distribution and focusing on the spatial pattern of the lowest and highest burn severity classes minimizes uncertainty regarding burn severity classifications. Spatial pattern metrics that are based on the definition of a “patch” instead of the adjacencies of neighboring “pixels” may provide more transferable information regarding the spatial structure of severity within burns. Despite methodological challenges and inherent spectral, spatial, and temporal limitations, remotely sensed burn severity data provides an inexpensive and data-rich resource for understanding the severity and spatial pattern of mixed-severity fire regimes.