Satellite Visible Radiance Data Assimilation: A Review

Data assimilation (DA) of satellite visible radiance observations holds great potential to improve the forecasting skills of numerical weather prediction (NWP) models, yet several challenges remain. This review synthesizes recent progresses related to the visible radiance DA and critically assesses their impacts on the analyses and forecasts of cloud and precipitation. As two major components of DA, significant efforts have been devoted to both observation operators (e.g., CRTM and RTTOV) and DA methods (e.g., variational and ensemble-based methods). On one hand, improvements on observation operators, particularly in radiative transfer solvers and cloud optical parameterizations, have enhanced both computational accuracy and efficiency for Top-of-Atmosphere (TOA) visible radiance simulations. Meanwhile, there are still challenges to generate representative visible images due to uncertainties in cloud optical parameterizations and simplifications to three-dimensional (3D) radiative effects. On the other hand, various DA methods are constrained by the quasi-static assumption in background error covariances, whereas ensemble Kalman filter (EnKF)-based methods offer greater flexibility at the current stage. Nevertheless, both methods are limited by the nonlinear and non-Gaussian nature of moist physics processes. EnKF-based methods also suffer from limitations in vertical localization. Although particle filters are theoretically well suited to nonlinear and non-Gaussian problems, the strong nonlinearity in observation operators severely limits the representativeness of resampled particles. Future efforts should focus on incorporating 3D radiative effects, refining cloud optical parameterizations, designing specific DA methods for visible nonlinear and non-Gaussian problems, and designing adaptive vertical localization strategies. Addressing these challenges will facilitate more effective application of visible radiances DA in cloud and precipitation forecasting.