A Comparison of Cloud and Aerosol Measurements from OCO-2 and CALIPSO

The launch of OCO-2 and its inclusion in the A-train has provided new opportunities to conduct intercomparison studies among the different A-train sensors. The goal of OCO-2 is to use hyperspectral measurements of reflected near-infrared sunlight to retrieve carbon dioxide concentrations with high accuracy. This is currently only possible, however, for measurements of very clear scenes. To filter out scenes contaminated by clouds and aerosols, OCO-2 relies primarily on two fast cloud-screening algorithms, one using the O2-A band, and one using two near-infrared CO2 bands. In this study, we compare OCO-2 cloud-screening results to CALIPSO observations co-located in time and space. CALIPSO cloud and aerosol optical depths and their vertical profiles were used for validation. We find that for high optical depth CALIPSO measurements, the O2-A band cloud-screening algorithm correctly identifies most (~80%) scenes as cloudy. For low optical depth CALIPSO measurements, it correctly identifies most (~70%) scenes as clear. Additionally, we found that the O2-A band cloud-screening algorithm misidentifies up to ~80% of scenes containing low, thick clouds, while correctly identifying nearly all scenes containing high clouds. The majority of these unidentified low clouds, however, can be filtered out using information from the CO2 bands. These results confirm previous simulation-based work for the first time. Future studies include investigating the sensitivity of these results to cloud phase, the optical depth detection limits of OCO-2, and potential solar zenith angle biases.