Towards Operational GPS Meteorology

COST Action 716 Workshop

10 - 12 July, 2000
Soria Moria, Oslo, Norway

Summary by Sylvia Barlag

The workshop session on "Assimilation of GPS into NWP models and its impact" was the first session of the COST-716 workshop 'Towards operational GPS meteorology'. Two keynote presentations about the state of the art were given, one covering Europe and the other the US. My overall impression after the discussion was that although many problems remain to be resolved, especially w.r.t. understanding errors and error sources, the operational meteorological community is moving fast toward the integration of the GPS data into their observing systems.

In the session on "Climate research and climate monitoring" one keynote presentation was given. My impression from this presentation plus discussions is that the climate community is not ready (yet) to use the GPS data but is willing to learn to understand how to use GPS. More discussion with GPS experts and users will be needed.

Here is the more detailed account of the presentations plus discussions.

The first keynote presentation, by Mark Higgins of The Met Office of the UK, entitled: "Progress in 3D variational assimilation of total zenith delay at The Met. Office", introduced the mathematical formulation of data assimilation into a meteorological model. This formulation can be applied to many types of observations, but is only valid under certain assumptions about the behaviour of observation errors. Total Zenith Delay, Precipitable Water, Slant Water or Delay or other GPS derived values could be assimilated. In the UK regional model TZD's are assimilated being closest to the time observable. The presentation expanded a bit on the importance of knowledge of the observation errors and covariances, as well as of the horizontal correlations. Several examples of forecasts from the UK regional model were shown, and the differences between runs using and not using ground based GPS data from the UK were shown. Although some questions remain to be solved it was rather clear that most impact from GPS data can be expected in situations where the weather is not dominated by dynamics (wind).

The second keynote presentation by Alexander McDonald from the Forecast Systems Laboratory of NOAA (US) anticipated an improved precipitation forecast skill from the assimilation of GPS slant delay observables in mesoscale numerical weather prediction models. The presentation centered around Observing System Simulations with a 40 km spaced grid of GPS receivers and a 3-Dimensional variational analysis of slant water vapor using a 10 km grid mesoscale model. When using slant delays a trade-off between using signals from low satellite elevations, introducing additional errors in the observables, and receiver grid spacing is necessary to achieve the desired vertical resolution of the meteorological boundary layer. The dominant scales of humidity variations, e.g. 100 km in the horizontal direction, can be captured with the 40 km grid. This is not possible using the conventional radiosonde networks. The non-uniqueness of the analysis solution was shown to be resolved by regular vertical profile observations within the spatial domain. A severe weather forecast in the Oklahoma tornado belt was simulated using the 40 km GPS receiver grid data and compared to the regular forecast for that situation without the use of additional PW information. The results showed the potential of the additional GPS information. In the near future a network will be deployed in that region of approximately 100 receivers to perform real experiments. The perspective for a 5000 receiver network covering the US as part of the North Americal meteorological Observing System seems promising.

During the discussion that followed many questions were raised. The presented results were obtained using simple assumptions about the errors of the observables and derived quantities while no correlations were assumed. In reality the error sources and error correlations still have to be understood. This was emphasised by both authors. Also model background errors and structure functions should still be determined. Several questions about the impact of the lay-out of the network and the geography of the network domain were addressed, e.g. the influence on the presence of mountain ranges in the domain or the differences of the heights of the receivers w.r.t. model surface heights. The need for more vertically resolved observations over larger horizontal domains was addressed. The possibility of the use of aircraft moisture measurements along ascending or descending paths was mentioned as a means to constrain the GPS analysis solutions, as well the use of DIAL laser, 12-channel radiometer measurements and future satellite measurements (GIFT).

In the session on "Climate research and climate monitoring" Daniela Jacob of the Max Planck Institute in Hamburg (Germany) gave a keynote presentation entitled: "The role of water vapour in the atmosphere". She explained the need for long-term stable observations of water vapour for the purpose of climatology and climate change research. She showed that large discrepancies still exist between predictions of water vapour content of the atmosphere made with climate models and observations of water vapour, generally derived from satellite measurements due to their world-wide coverage. Climate models predict an important change of water vapour distribution, especially in polar regions, for a changing global climate. Hence, continuous and correct world-wide observation of water vapour is important. After discussion it remained unclear whether GPS data can help solve this challenge due to the mainly continental distribution of GPS receivers. First of all many networks can only derive zenith water vapour content at GPS stations in an indirect way, using pressure and temperature values from other sources. For climate research it is generally important to have independent observations. Secondly, at present climate models and modellers can not handle delay measurements, which are hard to compare to other water vapour observables. However, from the point of view of calibration and degradation of measurement devices the GPS system has advantages over many other observation systems in that it uses time as observable. During the discussion it was also pointed out that GPS Radio Occultations may provide important input to climate monitoring as this technique can provide accurate observation of tropopause heights.