Aerosol emission estimation using SPEXone observational capabilities and Observing System Simulation Experiments (OSSEs)

A top-down approach for aerosol emission estimation from polarimetric retrievals of aerosol amount, size, and absorption is employed . The method uses a fixed-lag ensemble Kalman smoother (LETKF-Smoother) under the framework of Observing System Simulation Experiments (OSSEs), in order to evaluate the observational capabilities of a satellite with near perfect global coverage as well as of the future multi-angle polarimeter instrument, SPEXone. ECHAM-HAM is used for the nature runs (NATs), the control (CTL) and the data assimilation (DAS) experiments. The ensemble is composed by 32 simulations where the default aerosol emissions for all species are perturbed with factors taken from a Gaussian distribution. Synthetic observations, specifically Aerosol Optical Depth at 550nm (AOD₅₅₀), Angstrom Exponent 550nm to 865nm (AE_550-865) and Single Scattering Albedo at 550nm (SSA₅₅₀) are assimilated in order to estimate the aerosol emission fluxes of desert dust (DU), sea salt (SS), organic carbon (OC), black carbon (BC) and sulphates (SO₄), along with the emission fluxes of two SO₄ precursor gases (SO₂, DMS). The synthetic observations are sampled from the NATs according to two satellite observing systems, with different spatial coverage capabilities. The first, is an idealized sensor that retrieves observations over the whole globe in 2days even under cloudy conditions, hence is named PERFECT. The second, is the sensor SPEXone, a hyperspectral multi-angle polarimeter with a narrow swath (100km), that will be a part of the NASA PACE mission. The assimilated observations sampled using the PERFECT sensor, estimate the emission of all aerosol species with a global relative Mean Absolute Error (MAE) equal or lower than 5% (except SO₄). Despite its limited coverage, the SPEXone sampling bares similar results, although MAE is a bit larger for Dust and Sea Salt. Further, experiments show that doubling the measurement error on the assimilated observations, increases additionally the global relative MAE by less than 10%. In addition, the role of biased meteorology on emission estimation was quantified by using two different datasets (ERA5 and ERAi) to nudge the U and V wind components of the model. The results reveal that when the wind of NAT and DAS are nudged to different datasets the global relative MAE of SS grows by 24%, while the estimated emissions of DU, OC, BC and SO₂ are negatively affected to a smaller extent (~10%). The upcoming SPEXone sensor will provide observations related to aerosol amount size and absorption, with sufficient coverage and accuracy, in order to estimate aerosol emission accurately.