Examining the Fidelity of Cloud Optical Properties and Overlap Statistics through Comparisons of Radiative Transfer and Cloud Resolving Models against AIRS Observations

Clouds play an important role in determining the energy budget of the atmosphere. The largest uncertainty in predicting the climate is also caused by clouds. Most radiative transfer models can match satellite observations under clear sky conditions; however, this is not true for cloudy atmospheres. Uncertainties due to cloud effects arise not only from lack of adequate knowledge of the cloud optical properties, but also from the unknown vertical liquid/ice cloud distributions. In this study, we use a cloud resolving model (CRM) to create a large number of scenarios with different vertical cloud distributions. Subsequently, we compare the results of two radiative transfer models—High-performance Atmospheric Radiation Package (HARP) and Principal Component-based Radiative Transfer Model (PCRTM)—against observations made by the Atmospheric InfraRed Sounder (AIRS) aboard NASA's Aqua satellite.

We will test two hypotheses about the ability of the HARP and PCRTM models to reproduce spectra observed by AIRS. Hypothesis 1 is that HARP predictions are superior to PCRTM predictions in reproducing AIRS spectra. Hypothesis 2 is that HARP predictions are statistically indistinguishable from AIRS observations. We will employ statistical permutation tests to generate the required null distributions to carry out these hypothesis tests using a variety of test statistics, e.g., correlations between spectra.