Three-way comparison between OMI and PARASOL cloud pressure products

The cloud pressures determined by three different algorithms, operating on reflectances measured by two spaceborne instruments in the ‘‘A’’ train, are compared with each other. The retrieval algorithms are based on absorption in the oxygen A-band near 765 nm, absorption by a collision induced absorption in oxygen near 477 nm, and the filling in of Fraunhofer lines by rotational Raman scattering near 350 nm. A Lambertian reflector as cloud model is assumed in the retrievals. The first algorithm operates on data collected by the POLDER instrument on board PARASOL, while the latter two operate on data from the OMI instrument on board EOS-Aura. The satellites sample the same air mass within about 15 min. We compare the retrieval algorithms using synthetic spectra to give the comparison realistic baseline expectations. It appears that these cloud pressures are not the pressure of the cloud top, but of a level inside the cloud. This is corroborated by comparisons with MODIS and CloudSat data: while the top of the cloud is seen by MODIS using emitted IR radiation, both OMI and PARASOL algorithms retrieve a pressure near the midlevel of the cloud. The three cloud pressure products are compared using 1 month of data. The cloud pressures are found to show a similar behavior, with correlation coefficients larger than 0.85 between the data sets for high effective cloud fractions. The average differences in the cloud pressure are small, between 2 and 45 hPa, for the whole data set, with an RMS difference of 65 to 93 hPa. This falls within the science requirement for the OMI cloud pressure to have an accuracy of 100 hPa. For small to medium effective cloud fractions, the cloud pressure distribution found by PARASOL is very similar to that found by OMI using the O2 –O2 absorption. Somewhat larger differences are found for very high effective cloud fractions.