Effects of Slope and Aspect Variations on Satellite Surface Temperature Retrievals and Mesoscale Analysis in Mountainous Terrain

Abstract

Surface temperature retrieval in mountainous areas is complicated by the high variability of temperatures that can occur within a single satellite field of view. Temperatures depend in part on slope orientation relative to the sun, which can vary radically over very short distances. The surface temperature detected by a satellite is biased toward the temperatures of the sub-field-of-view terrain elements that most directly face the satellite. Numerical simulations were conducted to estimate the effects of satellite viewing geometry on surface temperature retrievals for a section of central Colorado. Surface temperatures were computed using a mesoscale model with a parameterization of subgrid variations in slope and aspect angles. The simulations indicate that the slope-aspect effect can lead to local surface temperature variations up to 30°C for autumn conditions in the Colorado mountains. For realistic satellite viewing conditions, these variations can give rise to biases in retrieved surface temperatures of about 3°C. Relative biases between retrievals from two satellites with different viewing angles can be over 6°C, which could lead to confusion when merging datasets. The bias computations were limited by the resolution of the available terrain height data (?90 m). The results suggest that the biases would be significantly larger if the data resolution was fine enough to represent every detail of the real Colorado terrain or if retrievals were made in mountain areas that have a larger proportion of steep slopes than the Colorado Rockies. The computed bias gradients across the Colorado domain were not large enough to significantly alter the forcing of the diurnal upslope-downslope circulations, according to simulations in which surface temperature retrievals with view-dependent biases were assimilated into time-continuous analyses. View-dependent retrieval biases may be relevant to climatological analysts that rely on remotely sensed data, given that bias-induced errors are systematic.