The vertical temperature profile of the atmosphere has an influence on thewidth and intensity of gaseous absorption lines. In the visible and nearinfrared part of the spectrum, this poses a problem for the fast forwardsimulation of the radiative transfer, needed in algorithms for the retrievalof any atmospheric or surface-related parameter from satellite measurements.We show that the main part of the global variability of temperature profilescan be described by their first 2 to 6 eigenvectors, depending on theaccuracy requirement, by performing a Principal Component Analysis (PCA) on aglobal set of temperature profiles from the Global Forecast System (GFS).Furthermore, we demonstrate the possibility to approximate the atmospherictransmittance in the O A band for any temperature profile with almostperfect accuracy by a linear combination of the transmittances attributed toeach of the significant temperature eigenvectors. For the retrieval ofsurface pressure from O A band measurements, this reduces the global rootmean square error from >30 hPa to better than 1 hPa by strongly reducingthe regional bias of surface pressure, retrieved on the assumption of anaverage temperature profile. The technique can be applied under scatteringconditions to eliminate temperature-induced errors in, e.g., simulatedradiances. In principal, the method can be useful for any problem includinggaseous absorption or emission with a significant influence of thetemperature profile, such as the retrieval of total water vapour contentor sea surface temperature.
展开▼