In situ and satellite-based estimates of cloud properties and aerosol-cloud interactions over the southeast Atlantic Ocean

Abstract

In situ cloud probe data from the NASA ObseRvations of Aerosols above CLouds and their intEractionS (ORACLES) field campaign were used to estimate the effective radius (Re), cloud optical thickness (τ), and cloud droplet concentration (Nc) for marine stratocumulus over the southeast Atlantic Ocean. The in situ Re, τ, and Nc were compared with co-located Moderate Resolution Imaging Spectroradiometer (MODIS) retrievals of Re and τ and MODIS-derived Nc. For 145 cloud profiles, a MODIS retrieval was co-located with in situ data with a time gap of less than 1¯h. On average, the MODIS Re and τ (11.3¯μm and 11.7) were 1.6¯μm and 2.3 higher than the in situ Re and τ with Pearson's correlation coefficients (R) of 0.77 and 0.73, respectively. The average MODIS Nc (151.cm-3) was within 1¯cm-3 of the average in situ Nc with an R of 0.90. The 145 cloud profiles were classified into 67 contact profiles where an aerosol concentration (Na) greater than 500¯cm-3 was sampled within 100¯m above cloud tops and 78 separated profiles where Na less than 500¯cm-3 was sampled up to 100¯m above cloud tops. Contact profiles had a higher in situ Nc (by 88¯cm-3), higher τ (by 2.5), and lower in situ Re (by 2.2¯μm) compared to separated profiles. These differences were associated with aerosol-cloud interactions (ACI), and MODIS estimates of the differences were within cm-3, 0.5, and 0.2¯μm of the in situ estimates when profiles with MODIS Re>1μm or MODIS τ>25 were removed. The agreement between MODIS and in situ estimates of changes in Re, τ, and Nc associated with ACI was driven by small biases in MODIS retrievals of cloud properties relative to in situ measurements across different aerosol regimes. Thus, when combined with estimates of aerosol location and concentration, MODIS retrievals of marine stratocumulus cloud properties over the southeast Atlantic can be used to study ACI over larger domains and longer timescales than possible using in situ data.