| Data from NASA's MODerate Resolution Imaging Spectroradiometer (MODIS), in polar orbit on the Terra and Aqua platforms, have provided surface bidirectional reflectance distribution function (BRDF) and albedo products (MCD43) that have been successfully validated during the growing seasons of various vegetated land surface types. This research, however, focuses on the quality of MODIS BRDF/albedo product retrievals during the more difficult periods of vegetation dormancy and snow cover by comparison with ground-based albedo measurements. Cropland, grassland, deciduous and conifer forest, and high latitude tundra (including recently burned) sites are considered. Low illumination angles and persistent cloudiness, as well as lower-quality atmospheric correction and cloud discrimination, limit the number of high quality retrievals that are obtained during snow-covered periods. Forest retrievals are challenging as underlying snow may be obscured by foliage or canopy shadows at high viewing and illumination angles. Neither satellite albedo retrievals nor ground measurements are considered reliable at solar zenith angles greater than 70°, which further complicates retrievals at high latitude locations. Moreover, changes due to snowfall or snowmelt can alter the albedo of a location significantly over a very short timescale. Therefore, the standard 500-m gridded BRDF/albedo products are also compared with results from both the MODIS daily Direct Broadcast BRDF/Albedo algorithm and the standard MOD10A daily snow albedo product. Using an integrated validation strategy, analyses of the representativeness of the surface heterogeneity under both dormant and snow-covered situations are performed to decide whether direct comparisons between ground measurements and 500-m satellite observations can be made or whether finer spatial resolution airborne or spaceborne data are required to scale the results at each location. Landsat ETM+ data are used to generate finer scale representations of albedo at each location to fully link ground data with satellite data. Results show that high quality MCD43 products do achieve a high accuracy during both dormant and snow-covered periods as compared to either spatially representative field data or field data that have been scaled up with Landsat imagery. However, a daily retrieval strategy is necessary to capture ephemeral snow events or rapidly changing situations such as the spring snow melt. |
