Joint Centre for Mesoscale Meteorology (JCMM)
JCMM INTERNAL REPORT NO.108
The use of water vapour imagery and PV inversion to adjust initial conditions in NWP
by Sean J Swarbrick
It is considered desirable to make greater use of satellite imagery in the generation of NWP analyses. This project has explored one possible method for doing this, using the Meteosat water vapour channel. The technique attempts to utilise a proposed relationship between water vapour imagery and upper-tropospheric PV fields. This relation, if properly understood, could facilitate the detection of errors in the initial state of a forecast.
The basis of the PV - water vapour methodology employed by this study is as follows. The objective is to reduce errors in forecasts of N Atlantic cyclonic weather systems. The PV field on an appropriate isentropic or isobaric surface, at the initial time of the forecast, is compared with the corresponding water vapour image in order to reveal mismatches between the image and the PV field. The PV field is then adjusted to correct these mismatches. The PV adjustments may involve scaling and/or displacement of PV elements, on a range of pressure levels. (Note that scaling factors are arrived at only by guesswork, since the PV/water vapour comparison provides no scaling information.) PV inversion is used to derive wind and temperature increments from the PV adjustments. These wind and temperature increments are applied to the initial forecast fields in a way that preserves the dynamical balance. The forecast is then rerun from the adjusted initial state.
A number of case studies have been carried out using this methodology. No unequivocal positive impacts have been achieved. In some cases the impact was positive in certain respects, but negative in others. The case studies have established that there is often a direct relationship between the strength of an upper-tropospheric PV anomaly and the depth of the corresponding surface low; scaling up the anomaly will deepen the low. In addition, positional adjustments applied to a PV anomaly will affect the trajectory of the surface low, but the way in which the trajectory is affected is unpredictable. The results overall have not demonstrated that the PV – water vapour methodology has the potential to be useful for operational forecast initialisation, but rather tend to suggest the opposite.
The most significant scientific difficulty with which the project has had to contend is the fact that the PV – water vapour image relation is not yet defined in a quantitative sense. It is reasonably well understood in a qualitative sense, and this understanding has been successfully applied by other investigators in the context of subjective verification, but this level of understanding appears to be insufficient for the purposes of forecast initialisation. No adequate quantitative relation between PV and water vapour images has yet been established, although this is being pursued by various workers.
The greatest difficulty of all which has impeded progress in this project is a technical one – the fact that the only means available to modify PV fields is by manual editing of pp files. This procedure is extremely crude and laborious, and has severely limited the scope of possible PV adjustments. Until adequate technical means for PV adjustment are available, it will not be possible to explore the PV – water vapour methodology fully. In order to establish such a technique for PV adjustment, it would first be necessary to establish a quantitative PV - water vapour image relation by one means or another.