Modal view of the flow dependency of forecast errors

The normal-mode framework is developed for the representation of time-averaged and time-dependent structures of the global forecast errors. The applied methodology provides an attractive way to quantify the balance by splitting the forecast-error variances into parts projecting on the balanced and inertio-gravity (IG) circulations. The approach is particularly suitable for the tropics where the IG circulation dominates on all scales and where the short-range forecast errors are largest. Results from operational and idealized NWP data assimilation experiments show some important characteristics of the current forecast errors. The ECMWF 4D-Var ensemble is characterized by around 50% of the global forecast-error variance being unbalanced. In particular, the variance growth between 3-hr and 12-hr range is substantially different in the balanced, eastward-propagating and westward-propagating IG modes. The balanced mid-latitude variance growth dominates while the variance growth in the IG modes is most significant in the large-scale equatorial Kelvin waves. The results from the operational NWP system are contrasted with the outputs from a perfect-model ensemble data assimilation experiment which assimilated globally homogeneous observations of dynamical variables by using the ensemble adjustment Kalman filter in the Community Atmosphere Model (DART/CAM).