Improved fire severity mapping in the North American boreal forest using a hybrid composite method

Abstract

Fire severity is a key driver shaping the ecological structure and function of North American boreal ecosystems, a biome dominated by large, high-intensity wildfires. Satellite-derived burn severity maps have been an important tool in these remote landscapes for both fire and resource management. The conventional methodology to produce satellite-inferred fire severity maps generally involves comparing imagery from 1 year before and 1 year after a fire, yet environmental conditions unique to the boreal have limited the accuracy of resulting products. We introduce an alternative methodthe 'hybrid composite'based on deriving mean severity over time on a per-pixel basis within the cloud-computing environment of Google Earth Engine. It constructs the postfire image from satellite data composited from all valid images (i.e., clear-sky and snow-free) acquired in the time period immediately after fire through the early growing season of the following year. We compare this approach to paired-scene and composite approaches where the post-fire time period is from the growing season 1 year after fire. Validation statistics based on field-derived data for 52 fires across Alaska and Canada indicate that the hybrid composite method outperforms the other approaches. This approach presents an efficient and cost-effective means to monitor and explore trends and patterns across broad spatial domains, and could be applied to fires in other regions, especially those with frequent cloud cover or rapid vegetation recovery.

Evaluating a New Relative Phenological Correction and the Effect of Sentinel-Based Earth Engine Compositing Approaches to Map Fire Severity and Burned Area

“…Therefore, the values of a spectral index can vary considerably from slight changes in image dates, resulting in substantially different burn severity maps [21,22,54]. This can affect image accuracy in mapping fire perimeters [62] and fire severity [19,58]. Moreover, it could also influence the widely reported lack of transferability of fire severity and burned area thresholds between seasons and ecosystems [20,[65][66][67].…”

“…In addition, in contrast to the more widely analyzed Landsat satellite, there are relatively few studies examining the temporal variations in Sentinel spectral indices' accuracy in mapping burned area [68] and fire severity [22]. Furthermore, most of those relatively scarce studies have used paired-image techniques, contrasting with a scarcity of studies addressing temporal and phenological variations in fire severity spectral indices' accuracy using composite images [19,22]).…”

“…This approach consists in subtracting from the uncalibrated dNBR image a phenological constant offset, generally calculated as the average dNBR from the unburnt area surrounding the fire perimeter [70,72,73]. Because of the relative simplicity of this method, some steps have been taken towards its automatization for Landsat [19,41,74] and Sentinel images [75]. Nevertheless, in order to be representative, the average offset should be calculated in similar vegetation types to those where the fire occurred [41,74,76].…”

“…In spite of the great progress in fire severity and burned area evaluations using medium-resolution sensors, several challenges remain for the development of more standardized, automated methods of post-fire assessment. Among others, these challenges include: (i) the assessment of spectral indices, particularly for Sentinel-2, to map burned area perimeters and measure fire severity and fire effects by strata, namely vegetation and soil; (ii) the evaluation of the effect of compositing periods and techniques, particularly on automated platforms such as Google Earth Engine (e.g., [19]), on the accuracy for mapping initial assessments of fire severity and burned area perimeters; (iii) the quantification of the effect of non-fire-related temporal and spatial variations in spectral indices [20][21][22] and the development of simple, readily automated methods of phenological correction that account for the spatial and temporal heterogeneity of such effects [23].…”

“…Nevertheless, the image composite techniques analyzed in the literature have largely varied, depending on the study goal, location, and period analyzed. For example, mean composites have been used for extended assessments of fire severity [19,41,44,45] and mean, median, or medoid [46,47] composites have been also used to map burned area and recovery [47][48][49][50]. Likewise, minimum, or maximum composites [26,27,42,[51][52][53] have been used for mapping fire perimeters and trends of burned area [54] or fire severity [55].…”