Constraining the Uncertainties of Skin Temperature and Surface Emissivity for the Assimilation of Simulated Surface-Sensitive Radiance Observation Over Land in an Idealized 1D-EnVar

The assimilation of surface-sensitive radiance observations over the land surface has the potential to improve the skill of weather forecasts at Environment and Climate Change Canada. However, the benefit of assimilating such observations is known to be limited by large uncertainties of skin temperature and surface emissivity. In this study, the following approaches will be investigated to reduce their uncertainties when assimilating various combinations of low-peaking AMSU-A channels: 1) Examine the impact on the skin temperature error; 2) Include surface emissivity as a state variable and examine the ability to independently correct errors in skin temperature and surface emissivity; 3) Examine the influence of including background-error correlation between skin temperature, surface emissivity and atmospheric state variables. In order to facilitate these experiments, an idealized 1D-EnVar system will be used to assimilate simulated observations, reducing the complexity of the research problem by eliminating the spatial effect, the need for bias correction, and also eliminating any inconsistency between the errors of the background and observations and their corresponding specified error covariances. The background state and observations are simulated based on a prescribed “true” atmosphere. The realism of the simulation is preserved by using the operational background state as the “true” atmospheric state and employing the ensemble derived background error covariance matrix to both simulate background error and to specify the B-matrix in the 1D-EnVar. For each experimental configuration, the 1D-EnVar analysis corrections will be evaluated for their ability to reduce errors in skin temperature, surface emissivity and atmospheric variables.