Impacts of Observation Forward Operator on Infrared Radiance Data Assimilation with Fine Model Resolutions

Abstract

Abstract All-sky radiance assimilation often has non-Gaussian observation error distributions, which can be exacerbated by high model spatial resolutions due to better resolved nonlinear physical processes. For ensemble Kalman filters, observation ensemble perturbations can be approximated by the linearized observation operator (LinHx) that uses the observation operator Jacobian of ensemble mean rather than the full observation operator (FullHx). The impact of observation operator on infrared radiance data assimilation is examined here by assimilating synthetic radiance observations from channel 1025 of GIIRS with increased model spatial resolutions from 7.5 km to 300 m. A tropical cyclone is used, while the findings are expected to be generally applied. Compared to FullHx, LinHx provides larger magnitudes of correlations and stronger corrections around observation locations, especially when all-sky radiances are assimilated at fine model resolutions. For assimilating clear-sky radiances with increasing model resolutions, LinHx has smaller errors and improved vortex intensity and structure than FullHx. But when all-sky radiances are assimilated, FullHx has advantages over LinHx. Thus, for regimes with more linearity, LinHx provides stronger correlations and imposes more corrections than FullHx; but for regimes with more nonlinearity, LinHx provides detrimental non-Gaussian prior error distributions in observation space, unrealistic correlations, and overestimated corrections, compared to FullHx. Significance Statement Assimilating satellite radiances has been essential for numerical weather prediction. All-sky radiance assimilation can improve the analyses and forecasts of tropical cyclones, but it often has non-Gaussian observation error distributions. With increased model resolutions, nonlinear physical processes can be better resolved, which leads to non-Gaussian error distributions of state variables. Nonlinearity and non-Gaussianity impose great challenges for data assimilation (DA), since most DA theories assume linear processes and Gaussian error distributions. As the spatial resolutions of observations and numerical models will keep increasing, the potential issues of assimilating high-spatial-resolution observations given fine model resolutions need to be examined. The linearized observation forward operator, as an alternative to remedy non-Gaussianity for radiance DA, is investigated. With model resolution increasing from 7.5 km to 300 m, the linearized observation operator has advantages over the full observation operator for assimilating clear-sky radiances, but the opposite is true for assimilating all-sky radiances.