Extended assimilation and forecast experiments with a 4D-Var assimilation system

Année de publication

1998

Results of four-dimensional variational assimilations, 4D-Var, in <br>cycling mode, over a few two-week assimilation periods are presented. <br>4D-Var is implemented in its incremental formulation, with a <br>high-resolution model with the full physical parametrization package to <br>compare the atmospheric states with the observations, and a <br>low-resolution model with simplified physics to minimize the <br>cost-function. The comparison of 4D-Var using several assimilation <br>windows (6, 12 and 24 hours) with 3D-Var (the equivalent of 4D-Var with <br>no time-dimension) over a two-week period shows a clear benefit from <br>using 4D-Var over a 6 or 12hour window compared to the static 3D-Var <br>scheme. It also exhibits some problems with the forecasts started using <br>4D-Var over a 24-hour window. The poorer performance of 4D-Var over a <br>relatively long assimilation window can be partly explained by the fact <br>that, in these experiments, the tangent-linear and adjoint models used <br>in the minimization are only approximations of the assimilating model <br>(having lower resolution and crude physics). The error these <br>approximations introduce in the time evolution of a perturbation affects<br> the convergence of the incremental 4D-Var, with larger discontinuities <br>in the values of the cost-function when going from low to high <br>resolution for longer assimilation windows. Additional experiments are <br>performed comparing 4D-Var using a 6-hour window with the 3D-Var system.<br> Two additional 2-week periods show a consistent improvement in <br>extratropical forecast scores with the 4D-Var system. The main 4D-Var <br>improvements occur in areas where the 3D-Var errors were the largest. <br>Local improvement can be as large as 35% for the root-mean-square of the<br> 5-day-forecast error, averaged over a two-week period. A comparison of <br>key analysis errors shows that, indeed, 4D-Var using a 6-hour window is <br>able to reduce substantially the amplitude of its fast-growing error <br>components. The overall fit to observations of analyses and short-range <br>forecasts from 3D-Var and 4D-Var is comparable. In active baroclinic <br>areas, the fit of the background to the data is considerably better for <br>the 4D-Var system, resulting in smaller increments. It appears that in <br>these areas (and in particular over the west Atlantic), 4D-Var is able <br>to better use the information contained in the observations. The ability<br> of 4D-Var to extrapolate some aircraft data in the vertical with a <br>baroclinic tilt is illustrated. Problems exist in the tropics and <br>mountainous areas due partly to a lack of physics in the tangent-linear <br>model. Possible improvements to the system (the introduction of more <br>physics; better behaviour of the incremental approach owing to a line <br>search at high resolution) are also discussed.</div>

  Accès à la notice sur le site du portail documentaire de Météo-France

  Liste complète des notices publiques