engleski

Spectral Blending Initialization of ALADIN Model by Incremental Digital Filter

Increasing of the LAM resolution enlarges the role of the fine initialization methods by introducing the more sensible methods for control and reduction of unrealistic amplitudes of internal gravity waves from the initial meteorological fields produced by coupling numerical models. The idea of the "Blending type" of initialization method relies on assumption that the analysis of the coupling numerical models provide the correct large scale waves, but the short wave features are artificial and lead to the unbalanced initial state. In the aim to overwhelm misrepresented properties and to introduce the fine scale structures, the blending of the correct large scale waves (from the resolution of the coupling model analysis) and short wave first guess forecasts from coupled LAM models was introduced. The blending of the spectral fields was obtained by Digital Filter Initialization, in the two steps: blending assimilation with the long cut-off analysis and blending production with short cut-off analysis of the coupling model. The blending of the long waves and the short waves was preformed at the lower truncation ratio, where the DFI was toll to separate the processes of the shortest time and space scales. The blending tests and tuning were preformed at ARPEGE/ALADIN model system over the Central European domain at RC LACE Prague Center. The long period of parallel tests of ordinary DFI initialization and spectral blending initialization, under the summer and winter atmospheric conditions showed the benefits of the blending method. The strongest blending impact was evident on reduction of internal gravity waves errors and the diminishing of spin-up effects of water vapor fluxes and MSLP. Blending keeping the well-developed meso-scale structures in the initial state of the forecast compared to the case when an external DFI was used. The blending initial conditions are more balanced, with realistic divergence fields as well as the sharper intensities of the frontal systems. The biases of the MSLP and surface wind were reduced, while the finer structure and the more accurate positions of the orographic precipitation were introduced. Blending type of the initialization is robust system that is operationally applied on the different ALADIN model configuration. The blending initialization ensures the well-balanced initial state for the outgoing ALADIN 3D-VAR research.

NWP; data analysis; initialization; blending; Digital Filter

nije evidentirano