Statistics of the vertical structure of PV anomalies are computed froma history of analyzed data over the 1999-2000 and 2000-2001 cool seasons, and these are used to define the vertical structure of anomalous PV, constrained by the vertical integral of anomalous PV that is derived from total ozone. The anomalous PV is then inverted, under the assumption of nonlinear balance, to obtain perturbations of velocity, temperature and pressure. We will illustrate results of the method and point out potential weaknesses. Furthermore, a method for obtaining the uncertainty of the retrieved variables will be suggested. This is essential to makeuse of retrieved fields in variational data assimilation systems. We apply the retrieval methodology to several cases of poor medium-range forecasts over North America. We trace forecast errors back to poor initialization over the Pacific Ocean and then investigate the ability of TOMS to improve the PV analysis and subsequent forecasts downstream.
Supplementary URL: