| Given the meteorological and optical parameters of a layer-structured atmosphere and the surface temperature, and emissivity or reflectance, the atmospheric transmittance and radiance distribution can be accurately calculated with a plane-parallel radiative transfer model. A combination of atmospheric data used in the LOWTRAN6 code with an azimuth-dependent radiative transfer computation is an efficient theoretical model for atmospheric correction. We use such a model to generalize the multichannel method for satellite measurement of sea surface temperature from thermal infrared data into land surface temperature measurement from space. An empirical formula derived from statistical analysis of theoretical simulations over snow cover and sands can estimate, within viewing angles up to 60(DEGREES) off-nadir, land surface brightness temperatures with a standard deviation of 0.2 K and a maximum error less than 1 K under clear-sky conditions, or a standard deviation less than 3 K and a maximum error less than 10 K under slightly adverse air conditions (including cirrus clouds, fogs, and rural aerosols), given satellite measurements in the three infrared channels of the NOAA 5-channel Advanced Very High Resolution Radiometer (AVHRR) and the surface elevation. If emissivity is known, it is also possible to obtain land surface thermodynamic temperatures with the same accuracy. From simulation data of a hypothetical four infrared channel instrument, one in the 3.5-4 (mu)m window and three in the 8-12.5 (mu)m window, it is possible to improve the accuracy to a standard deviation of about 2 K and a maximum error of about 6 K under slightly adverse air conditions. |
