The ECMWF operational implementation of four-dimensional variational assimilation. I: Experimental results with simplified physics

This paper presents results of a comparison between four-dimensional variat ional assimilation (4D-Var), using a 6-hour assimilation window and simplif ied physics during the minimization, and three-dimensional variational assi milation (3D-Var). Results have been obtained at 'operational' resolution T 213L31/T63L31. (T defines the spectral triangular truncation and I, the num ber of levels in the vertical, with the first parameters defining the resol ution of the model trajectory, and the second the resolution of the inner-l oop.) The sensitivity of the 3D-Var performance to different set-ups is inv estigated. In particular, the performance of 4D-Var in the Tropics revealed some sensitivity to the way the adiabatic nonlinear normal-mode initializa tion of the increments was performed. Going from four outer-loops to only o ne (as in SD-Var), together with a change to the 1997 formulation of the ba ckground constraint and an initialization of only the small scales, helped to improve the 4D-Var performance. Tropical scores then became only margina lly worse for 4D-Var than for 3D-Var. Twelve weeks of experimentation with the one outer-loop 4D-Var and the 1997 background formulation have been stu died. The averaged scores show a small but consistent improvement in both h emispheres at all ranges. In the shea range, each two- to three-week period has been found to be slightly positive throughout the troposphere. The bet ter short-range performance of the 4D-Var system is also shown by the fits of the background fields to the data. More results are presented for the At lantic Ocean area during FASTEX (the Fronts and Atlantic Storm-Track Experi ment), during which 4D-Var is found to perform better. In individual synopt ic cases corresponding to interesting Intensive Observing Periods, 4D-Var h as a clear advantage over 3D-Var during rapid cyclogeneses. The Very short- range forecasts used as backgrounds are much closer to the data over the At lantic for 4D-Var than for SD-Var. The 4D-Var analyses also display more da y-to-day variability. Some structure functions are illustrated in the 4D-Va r case for a height observation inserted at the beginning, in the middle or at the end of the assimilation window. The dynamical processes seem to be relevant, even with a short 6-hour assimilation period, which explains the better overall performance of the 4D-Var system.