The accuracy of satellite-derived albedo for northern alpine and glaciated land covers
Back to previous page
| Authors: | Williamson, SN; Copland, L; Hik, DS |
| Year: | 2016 |
| Journal: | Polar Sci. 10: 262-269 Article Link (DOI) |
| Title: | The accuracy of satellite-derived albedo for northern alpine and glaciated land covers |
| Abstract: | Alpine and Arctic land cover can present a challenge for the validation of satellite-derived albedo measurements due, in part, to the complex terrain and logistical difficulty of accessing these regions. We compared measurements of albedo on transects from northern mountain land covers (snowfield, glacier ice, tundra, saline silt river delta), and over a large elevation range to the coincident 8-day MODIS (MCD43) albedo product. We also compared field measurements at snow covered sites to the coincident daily MODIS (MOD10A1) snow albedo product. For each transect, we measured a range of albedo values, with the least variability on the silt river delta (range = 0.084) and the largest over mid-elevation glacier ice (range = 0.307). The highest elevation snowfield (0.170) had nearly the same range of albedo values as tundra (0.164). The MODIS shortwave White Sky Albedo product (MCD43A3) was highly correlated with the field transect albedo (R-2 = 0.96), with a Root Mean Square Error (RMSE) of 0.061. The MODIS shortwave Black Sky Albedo product was similarly correlated with field transects (R-2 = 0.96; RMSE = 0.063). These results indicate that remote observation of albedo over snow covered and alpine terrain is well constrained and consistent with other studies. Albedo varied by 15% both spatially and temporally for the high elevation snowfields at the point in the season where albedo variation should be at its minimum. There were several instances where MCD43A3 albedo was not produced over snow and was instead classified as cloud covered, despite field observations of cloud free skies. There were also several instances where daily MOD10A1 albedo was produced during the coincident 8-day period at these locations. This suggests that the cloud mask in the MCD43 product is overly conservative over snow. Spatial variation in albedo within the MODIS grid cell (500 m), especially for snow and glacier ice, combined with the uncertainty associated with positional accuracy of MODIS, indicates that the accuracy of MODIS albedo will be dependent on both land cover type and the period of observation. (C) 2016 Elsevier B.V. and NIPR. All rights reserved. |
Please send suggestions for improving this publication database to
sass-support@sfu.ca.
Departmental members may update their publication list.