Mapping directional emissivity at 3.7 m using a simple model of bi-directional reflectivity
Algorithms for Land Surface Temperature determination from satellite data need incorporating pixel-wise emissivity values to take care of the variability of this parameter over land and achieve a high accuracy in the surface emissivity. This parameter should be determined at the pixel scale and thus directly from the sensor's data. This work addresses the issue of extracting the land emissivity from AVHRR data. The necessary emissivity-temperature decoupling is achieved thanks to a method (Becker and Li 1990) that uses a combination of day/night channels 3, 4 and 5 data. Single channel atmospheric corrections are performed using MODTRAN and atmospheric profiles from outputs of GCMs (MétéoFrance AVISO database). The channel 3 emissivity is extracted in two steps: first the channel 3 bi-directional reflectivity is retrieved using the emissivity-temperature decoupling; second, emissivity is related to the reflectivity thanks to Kirchoff's relation. Accuracy assessment indicates that the expected overall error on channel 3 emissivity ranges between 3% and 6% at most (for low emissivity value). Series of day/night AVHRR images are processed, resulting in a wide range of view angles. A simple empirical model is used to fit the angular variation of the bi-directional reflectivity, from which a form factor can be calculated, allowing directional emissivity to be obtained. A clear distinction is observed in the angular behaviour between bare soils and vegetation covered surfaces where backscattering appears dominant. Directional channel 3 emissivity shows a large dynamical ranges, from above 0.95 for fully vegetation covered surfaces, to below 0.6 for desert surfaces. Amplitude of emissivity angular variation is small for vegetation, whereas it may be up to 10% between nadir and 60°-view angle for desert areas.
No Reference information available - sign in for access.
No Citation information available - sign in for access.
No Supplementary Data.