Brett Castro, Aaron Johnson, Xuguang Wang
The land surface often plays an important role in deep moist convection initiation (CI), especially on cases with weak synoptic scale forcing and the presence of boundary layer mesoscale circulations such as drylines. In such cases, improvements in the skill of mesoscale convection-allowing forecast systems such as RRFS are expected to be achievable through the strongly coupled data assimilation (SCDA) of screen level and soil moisture observations. Effective SCDA requires that the coupling between soil state and atmospheric state variables is well represented in the ensemble-based background error covariance. Such coupled covariances can be strongly impacted by how the initial perturbations to the soil states are generated and by sampling errors related to the finite ensemble size. However, previous studies have placed little emphasis on methods and impacts of generating ensemble perturbations to soil states, especially in the context of spinning up the land-atmosphere coupled covariances for mesoscale SCDA. The present study will explore the sensitivity of ~12 h forecasts of dryline CI cases to initial soil state perturbations. The sensitivity tests will then be used to optimize the selection and application of ensemble soil state perturbations downscaled from global ensemble forecasts for the purpose of generating and interpreting suitable land-atmosphere coupled background error covariance in a RRFS-like model. After evaluating the impacts of the optimized soil state initial perturbations, the ensemble-based background error covariances will be evaluated over a range of different ensemble sizes to better understand the role sampling error in the ensemble-based land-atmosphere coupled covariance. Process level studies will also be made to understand the mechanism of soil state on deep moist CI.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner