Development of a swath-like imager implementation of an LES Cloud Remote Sensing Simulator

For a hyperangular imaging polarimeter like HARP (compared to heritage atmospheric imagers like MODIS) the number of solar geometries per pixel of observations is significantly greater because of its wide-field of view swath – but also particularly because each hyperangular observation constitutes its own pushbroom images. For a single granule of data every single pixel has it's own unique solar geometry so that even when rounded to the nearest degree there are thousands or more unique solar geometries (pairs of solar zenith and relative azimuth angles, SZA and RAZ).This poses a difficulty, because every radiative transfer call requires specification of a single solar geometry to generate observations. When you simulate clouds this can be particularly computationally expensive - especially for an LES cloud simulator that is defined to test remote sensing sensitivities to cloud inhomogeneity on 50 m scales over a size-resolved droplet size distribution. The domain of these LES simulations is ~10-20 km leading to nearly 20,000 radiative transfer simulations for each individual solar geometry. At the scale of HARP, this would be an intractable radiative transfer simulation task. However, in this project we take advantage of a feature of the GISS Polarized Doubling-Adding (PDA) RT algorithm that takes advantage of some key symmetries associated with determining the cloud BRDF to obtain a "radiative kernel" output from a single call that can be interpolated to an arbitrary solar geometry offline from a scene-based LUT. The output from this exercise is huge – approximately 5 Terrabytes per LES scene and spectral band, all at the native viewing angle sampling of 0.25 degrees.From this unique library of radiative transfer simulations, we are generating some of the most realistic swath-like observations of clouds ever produced in combination with the PACE Project’s example L1C orbital geometry products https://oceandata.sci.gsfc.nasa.gov/directdataaccess/Level-1C/PACE/2022/080/To mimic the 3-D nature of HARP2's observations during PACE we are forcing parallax projection to the surface - a feature that could have impacts on the advanced cloud remote sensing products of PACE, most specifically the polarimetric cloudbow retrieval of droplet size distribution properties. Then, we can test how different types of cloud regimes (right now it’s just stratocumulus) behave for the PACE Polarimetric cloud retrievals behave under the influence of uncorrected parallax, cloud top height inhomogeneity, and broken cloud fractions below the 5km aggregation resolution of the L1C data product.