Analyzing variation in snow albedo across spatial scales of observation in the Alaskan boreal forest
Loading...
Authors
Fitts, Allyson
Issue Date
2024
Type
Thesis
Language
en_US
Keywords
Remote Sensing , Snow Albedo , Snow Hydrology
Alternative Title
Abstract
Snow albedo plays a critical role in the Earth's energy balance due to its high reflectivity in the visible wavelengths. On average, more than 50% of the land area above 20° North latitude is snow covered by December. In a warming climate where temperatures are estimated to rise and less precipitation will fall as snow in high latitude regions, it is crucial to understand variation in snow albedo across changing spatial scales. The albedo of snow is a primary control on the timing and rate of snowmelt and is affected by grain size and light absorbing impurities, such as black carbon shed from burned trees after a wildfire. Like the western United States, the boreal forests in Alaska are experiencing an increase in wildfire activity, drastically altering both the regional and global energy balance. After a wildfire, light absorbing particles such as black carbon and woody debris are deposited on the snow surface. These particles darken the snow surface, decreasing snow albedo and increasing the shortwave energy absorbed at the surface. From a landscape perspective, removing the dark forest canopy cover by fires increases the landscape albedo, yet the snow albedo is decreased. This research strives to untangle the paradox between the point and landscape scale across varying vegetation cover. To do so, we compare multispectral and hyperspectral remote sensing instrumentation across scales of observation from ground-based transects to airborne and satellite-based measurements. Using airborne data from Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-NG), we conducted a spectral unmixing analysis to characterize the fraction of snow across transects. Assuming ground measurements are entirely snow, we compared field spectrometer measurements to airborne and satellite reflectance data, which have a coarser spatial resolution. We computed the difference in reflectance values for ground-based vs. airborne and ground-based vs. satellite data. Results show that there is a linear relationship between reflectance differences and the fraction of snow cover in a mixed pixel. As the fraction of snow cover in a pixel increases, the cross-scale reflectance differences decrease in a linear manner. This means that if we know the snow fraction in a pixel, and we have measured the reflectance at one scale, then we can linearly interpolate to estimate the reflectance that would be measured at a finer or coarser scale. Moreover, if we can assume that albedo generally tracks with reflectance, we can estimate landscape albedo at different spatial scales. We conclude that the land cover classification and spatial scale of observation must be considered when estimating snow and landscape albedo. Instruments tend to agree well in snow covered, open areas compared to forested areas. Fine resolution airborne data, such as AVIRIS-NG, tends to agree with ground observations in open and burned areas. We encourage the snow remote sensing community to consider instrument optics and their spectral, spatial and temporal resolution as it relates to area classification when utilizing snow reflectance data products.