Development of Advanced Radiative Transfer Capabilities for Polarimetric Remote Sensing

Spaceborne observations by polarimeters contain rich information about atmospheric constituents (e.g., clouds and aerosols). Due to the large volume of remote sensing observational data, an efficient retrieval algorithm is needed to infer atmospheric constituent properties. The accuracy and efficiency of the retrieval algorithm depends in large part on the forward model approach. An in-line radiative transfer model (RTM) can incorporate atmospheric profiles and surface properties directly into the retrieval system to improve accuracy over a pre-computed look-up table approach, but it is challenging to satisfy the efficiency requirements in remote sensing applications.

To fully exploit the capability of polarimetric instruments, we have developed an accurate and fast in-line vector RTM that is capable of simulating the full Stokes vector observed at the top of the atmosphere (TOA) and at the terrestrial surface by considering absorption, scattering, and emission. Gas absorption is parameterized in terms of gas concentration, temperature, and pressure using a regression method that can be incorporated into multiple scattering computations. This approach is more than four orders of magnitude faster than a line-by-line RTM and has errors less than 0.1 K in terms of TOA brightness temperature. A two-component approach is utilized to solve the vector radiative transfer equation (VRTE), which separates the phase matrices of aerosol and cloud into forward and diffuse parts. The forward solution can be expressed by a semi-analytical equation based on the small-angle approximation. The diffuse part is solved by the adding-doubling method, which is computationally efficient because the diffuse component needs much fewer generalized spherical function expansion terms than the case of the original phase matrix. Because of the temperature variation in a homogeneous scattering layer, generally only the adding method has been used in past approaches to compute cloud and aerosol thermal emission. In this RTM, however, with proper assumptions, we develop an efficient doubling method to compute the cloud and aerosol thermal emission. To show the effectiveness of the new combined method, we compare the RTM simulations with POLDER and MODIS observations.