[1-1] Keynote

New applications of data assimilation: model and observation improvements and flux estimations

E. Kalnay (UMD), T. Miyoshi (RIKEN), D.Hotta (JMA), G.-Y. Lien (RIKEN), J. -S. Kang (KIAPS), T.-C. Chen (UMD), Y. Zhou (UMD), T. Sluka (UMD), S. Penny (UMD), J. Carton (UMD)

The data assimilation cycle used for numerical weather prediction (NWP) interpolates a short model forecast and the new observations to create new initial conditions (analysis) for integrating the model through the next cycle. Thus, improvements in NWP have been considered to require improvements in the model, observing systems and in the statistical interpolation scheme. Recently, however, it has become clear that with advanced methods of data assimilation, such as 4DVar and EnKF, the analysis cycle can be also applied in new areas: improving the models, observations, assimilation of new observing systems, and flux estimations. It is plausible that these may become the main tools to develop and improve Earth System models, observation forward models, quality control, and estimation of surface fluxes.

We will present examples of these new applications: optimal estimation of model parameters or surface fluxes by state augmentation, model improvements through the use of analysis increments, use of Ensemble Forecast Sensitivity to Observations (EFSO) to detect flawed observations that make the 6hr forecast worse (Proactive QC), and its application to estimate the observations Error Covariance R, and efficient operational implementation of new observing systems.