Model selection in atmospheric remote sensing with application to aerosol retrieval from DSCOVR/EPIC. Part 1: Theory

The retrieval of aerosol and cloud properties such as their optical thickness and/or 1 layer/top height requires the selection of a model that describes their microphysical properties. 2 We demonstrate that, if there is not enough information for appropriate microphysical model 3 selection, the solution accuracy can be improved if the model uncertainty is taken into account 4 and appropriately quantified. For this purpose, we design a retrieval algorithm accounting for the 5 uncertainty in model selection. The algorithm is based on (i) the computation of each model solution 6 using the iteratively regularized Gauss-Newton method, (ii) the linearization of the forward model 7 around the solution, and (iii) the maximum marginal likelihood estimation and the generalized 8 cross-validation to estimate the optimal model. The algorithm is applied to the retrieval of aerosol 9 optical thickness and aerosol layer height from synthetic measurements corresponding to the EPIC 10 instrument onboard the DSCOVR satellite. Our numerical simulations have shown that the heuristic 11 approach based on the maximum solution estimate, which is frequently used in the literature, is 12 completely unrealistic when both the aerosol model and surface albedo are unknown.