Large Area High-Resolution Albedo Retrievals from Remote Sensing to Assess the Impact of Wildfire Soot Deposition on High Mountain Snow and Ice Melt.

Abstract:

Deposition of wind-borne soot on snow and ice surfaces decreases albedo, thereby changing their energy budget and accelerating melt rates. Investigating the effect of wildfires on high mountain snow and ice albedo is essential to understanding potential changes in runoff generation in the glaciated headwaters of western North America’s major rivers. Several studies have addressed the effect of soot deposition on snow and ice albedo, but there is a lack of studies measuring the spatial distribution of albedo changes over large areas at high resolution. This study evaluates the spatial patterns of albedo decrease caused by soot deposition from the Western Canada 2017 and 2018 intense wildfire seasons on the Columbia Icefield (151 km²), Canadian Rockies. Albedo retrievals at 20 m resolution from 14 Sentinel-2 scenes were corrected for anisotropic properties of snow and ice using a novel application of a snow-kernel model based on MODIS. Albedos were compared to the number of cumulative smoky days (CSD) since July 1, 2016, as a proxy of upwind fire activity. The first image without fresh snow (July 29, 2017 - 7 CSD) was used to compute the reference albedo (snow: 0.614, ice: 0.372) and associated shortwave radiative forcing (SWF) of subsequent scenes. Comparison of albedo retrievals to station-measured albedo on the Athabasca Glacier revealed a mean bias of 0.024 and RMSE of 0.050. The lowest spatially-averaged albedos were 0.509 for snow (July 29, 2018 - 27 CSD) and 0.257 for ice (Sept. 5, 2019 - 48 CSD). Smoke activity ceased at the end of the 2018 summer at 48 CSD, but the ice albedo continued to decrease in 2019, likely due to algae growth as observed in field visits in the 2019 summer. The impact of lower albedo on SWF was calculated using the terrain-corrected irradiance of July 29, 2017, and varying the albedo retrievals of the remaining scenes. The greatest spatially-averaged SWFs were 118 W/m² for snow and 145 W/m² for ice, coinciding with the lowest albedos for each surface. The extensive spatial coverage and temporal persistence of lowered albedo revealed over the complex Columbia Icefield surfaces required high-resolution albedo retrievals that cover large areas. These retrievals demonstrate the importance of using bidirectional remote sensing models to assess the effect of wildfires on snow and ice energetics.