Rigorous estimates of the retrieval errors in diverse remote sensing applications provided by GRASP algorithm

Abstract:

The understanding of of the uncertainties in retrieval of the aerosol and surface properties is very important for adequate characterization of the processes that occur in the atmosphere.

However, the reliable characterization of the error budget of the retrieval products is a very challenging aspect that currently remains not fully resolved in most remote sensing approaches. For example, the level of uncertainties in the majority of satellite retrieval relies on post-processing validations and the dynamic errors are rarely provided. Therefore, implementation of fundamental approaches of the statistical estimation theory for practical retrievals and evaluation their efficiency remains of high importance.

This study describes and analyses the dynamic estimates of uncertainties of aerosol and surface retrieved properties from ground-based and satellite measurements using the GRASP (Generalized Retrieval of Atmosphere and Surface Properties) algorithm.

GRASP inversion algorithm described by Dubovik et al., (2011, 2014) is based on the concept of statistical optimization designed and provides dynamic errors estimates for all retrieved aerosol and surface properties. The approach takes into account the effects of both random and systematic uncertainties propagations. Similar concept has already been implemented in AERONET retrievals and discussed in previous studies by Dubovik et al., (2000) and Torres et al., (2014).

The algorithm provides error estimates for directly retrieved parameters included in the retrieved state vector, such as aerosol size distribution, refractive index, BRDF, etc. and for the characteristics derived from these parameters, such as total scattering, extinction, single dispersion albedo, etc. Moreover, GRASP algorithm provides full covariance matrices, i.e. not only variances of the retrieval errors and also correlations coefficients of these errors. The analysis of correlation matrix structure as can be very useful for identifying unobvious retrieval tendencies that appear to be useful approach for optimizing observation schemes and retrieval setups.

The illustrations of approach efficiency in ground-based and satellite applications will be provided.