The objectives of this proposal are the development and the evaluation of a generalized framework to assimilate observed microwave radiances emerging from raining areas into a global circulation model. Current systems for assimilating microwave radiances into numerical weather prediction models are suboptimal because of deficiencies in representing microphysical processes in the prediction models. These deficiencies are caused by the models' inability to numerically resolve small-scale processes at global scales, but also by limited knowledge of these processes. On the other hand, precipitation assimilation schemes are robust and less sensitive to the above-mentioned deficiencies in global numerical models. However, there might be information in the global models that can improve the satellite precipitation estimation if considered simultaneously with the global numerical modelling. In this project, we propose the incorporation of information from a high-resolution, high-quality precipitation-microwave radiance database used in precipitation retrievals into a microwave radiance assimilation framework. The framework will generalize existing radiance and precipitation assimilation schemes by allowing the estimation of precipitation variables at scales not resolved by the model in agreement with the model variables, and observed and predicted radiances. Its implementation will be based on a forward radiative transfer model and its adjoint used currently for precipitation estimation and the variational framework for precipitation assimilation used in an existing global circulation model. The framework's impact on global atmospheric analyses and forecasts will be tested using data from the TRMM Tropical Imager (TMI) and other satellite instruments.
Specifically, the proposed research consists of the following elements:
- Development of a variational assimilation framework to incorporate radiance information from passive microwave sensors into a numerical weather prediction model. The radiances will be assimilated into a one-dimensional (1D) nonlinear column model with tendencies prescribed by the 3D full model integration. The analyzed variables from the 1D assimilation will be incorporated into the full model either as tendency correction or retrieved state variables. Additional information derived from the precipitation-radiation databases currently used in precipitation retrievals will be considered in the time-dependent 1D assimilation framework to account for deficiencies in the column model.
- Comparisons of results from the new assimilation technique with results from existing formulations.
- Application and further evaluation of the newly developed assimilation framework.
This research is directly responsive to one of NASA's strategic goals, i.e. the advancement of precipitation assimilation techniques for the benefit of global atmospheric analyses and forecasts.