Key points are not available for this paper at this time.
The equivalent brightness temperature Tb recorded by geosynchronous infrared (geo-IR) "window" channel (10.7–11.5 μm) satellite sensors is shown to depend on the zenith angle (local angle from the zenith to the satellite for a pixel's ground location) in addition to the mix of clouds and surface that would be observed from a direct overhead viewpoint (nadir view). This zenith-angle dependence is characterized, and two corrections are developed from a collection of half-hourly geo-IR pixel data that have been parallax corrected and averaged to a 0.5° × 0.5° latitude/longitude grid for each geosynchronous satellite separately. First, composites of collocated Tb over tropical regions from the Geostationary Operational Environmental Satellite (GOES)-8/GOES-10 and the Meteosat-5/Meteosat-7 satellite pairs are used to produce robust estimates of isotropic zenith-angle corrections as a function of zenith angle and grid-box-averaged Tb. The corrections range from zero for a zenith angle of ∼26.5° to increases of more than 20 K at Tb = 230 K near the limb. Near-limb corrections in clear and very cold thick overcast conditions are smaller but are still positive. This empirical result depends on the surface–tropopause temperature differences, so a second correction was developed. Using collocations from the same two satellite pairs from 60°N to 60°S, differences in uncorrected Tb divided by differences in the corresponding corrected values were accumulated as a function of latitude and season. The resultant ratios smoothly vary from ∼1 in the Tropics to ∼0.5 at 60°N and 60°S, with a quicker decrease in the winter hemisphere. In comparison with the uncorrected geo-IR data, there is a 50% reduction in the root-mean-square differences between collocated values from adjacent satellites by applying the latitude/season-adjusted zenith-angle corrections. Histograms of corrected geo-IR Tb at large zenith angles closely match histograms from collocated near-nadir-view values. Residual difference maps are smooth and indicate intersatellite differences.
Joyce et al. (Sun,) studied this question.