Eskes, H. J., A. J. M. Piters, P. F. Levelt, M. A. F. Allaart, and H. M. Kelder
(Journal of the Atmospheric Sciences, Vol. 56, pp. 3560-3572, 1999).

Click here to download: compressed postscript file, 650 k
(unix extension .gz, use gunzip to decompress)

A four-dimensional data-assimilation method is described to derive synoptic ozone fields from total-column ozone satellite measurements. The ozone columns are advected by a 2D tracer transport model, using ECMWF wind fields at a single pressure level. Special attention is paid to the modeling of the forecast error covariance and quality control. The temporal and spatial dependence of the forecast error is taken into account, resulting in a global error field at any instant in time that provides a local estimate of the accuracy of the assimilated field. We discuss the advantages of the 4D-variational approach over sequential assimilation schemes. One of the attractive features of the 4D-Var technique is its ability to incorporate measurements at later times t>t0 in the analysis at time t0. This significantly improves the off-line analyzed ozone fields.