Aerosol Impact on Satellite Radiance Assimilation and Numerical Weather Prediction at the U.S Naval Research Laboratory

The data assimilation (DA) of hyperspectral IR-sensors (HIS) has proven to be crucial for numerical prediction, particularly, in substantially reducing forecast errors. However, IR satellite observations are sensitive to dust and sea salt aerosols, which are widespread throughout the globe, and have the potential of inducing radiance biases of up to 10K in channels used for operational numerical weather prediction. The impact of these induced temperature biases have not been thoroughly investigated and considered in operational HIS DA. For the most part, aerosol “contaminated” fields get rejected before getting assimilated, significantly reducing the amount of satellite data that is ingested. The irony of this scenario is that most of these rejected fields are located in regions of utmost importance, for example, over large cities, and in the vicinity of severe-weather genesis regions. Not only the impact of this rejection on operational HIS DA is uncertain, it is still unclear if it is worth the computational resources to include aerosol data in numerical modeling cycles for correcting HIS radiances before the DA step. That is, until now.

Here, we describe the implementation of aerosol radiance assimilation in the U.S Navy Data Assimilation System (NAVDAS-AR). We utilize the Naval Aerosol Analysis Prediction System (NAAPS) to identify aerosol laden pixels for the HIS sensors that get assimilated operationally (AIRS, IASI, CrIS, METOP, HIMARI and GOES) and assimilate them. The newly introduced hyperspectral radiance signatures utilize vertical distributions from NAAPS and optical properties from OPAC for the radiative transfer. Furthermore, we investigate and quantify aerosol-induced biases as a function of aerosol properties and loading for the aforementioned satellite sensors, iterating two case studies over 3-6 month periods (hurricane seasons 2016-2017). We analyze satellite data rejection, sensitivity and innovations, revise channel selection, errors and quality control. We also study the impact on the analysis and forecast over the primary oceanic regions (tropical Atlantic and Caribbean, northern Indian Ocean and western Pacific Ocean). The results are discussed within the context of aerosol influence on NWP outputs, particularly important as the community propels forward towards all-sky data assimilation and inline aerosol coupling.