Neural Network (NN) Retrievals of Low Level Liquid Cloud Properties from Multi-Angle Polarimetric Observations During ORACLES 2016-2017

Simultaneous retrievals of aerosol and cloud optical properties are being developed (e.g. MODIS, OMI), but are still challenging, especially for passive, single viewing angle instruments. By comparison, multiangle polarimetric instruments like RSP (Research Scanning Polarimeter) show promise for detection and quantification of aerosol above clouds (AAC), however, there are no operational retrieval algorithms available yet that combine both. Here we describe a new algorithm to retrieve low level liquid cloud optical properties from observations by RSP flown on the ER-2 in 2016 and the P-3 in 2017 during the ORACLES (ObseRvations of Aerosols above CLouds and their intEractionS) campaign. The algorithm is based on training a NN, and is intended to retrieve aerosol and cloud properties simultaneously. However, the first step, which is presented here, was to establish the retrieval scheme for low level StratoCumulus (Sc) cloud optical properties. The NN training was based on simulated RSP total and polarized radiances for a range of cloud optical depths (COD), effective radii (Reff), and effective variances (Veff), spanning 7 wavelength bands and 152 viewing zenith angles. We incorporated an uncertainty model for each of the measurable quantities (total reflectance and polarized reflectance) in our simulations to achieve a more realistic representation of the signals. We will present a summary of our cloud property retrievals from ORACLES 2016-2017 campaigns, which show good agreement compared with the standard RSP low-level cloud retrieval method that has been validated against in situ observations, and will discuss the results in relation with their spatial trends and above clouds aerosol amounts. We will also discuss next steps in developing the combined aerosol-cloud retrieval scheme using the NN approach.