McCoy [200 i] has provided background for the Global Assimilating Ionospheric Model (i.e., GAIM), which was sponsored by the U.S. DoD. Specifically it was a Multidisciplinary University Research Initiative (MURI) managed jointly by the Office of Naval Research (ONR) and the Air Force Office of Aerospace Research (AFOSR). The 5-year program began in 1999. Two awards were given: the Utah State University team headed by Robert Schunk, and the University of Southern California team headed by Chunming Wang. Elements of both programs were discussed at IES2002 in Alexandria, VA and were included in the conference Proceedings [Goodman, 2002], While the MURI component of GAIM is virtually over, the initiative has developed a life of its own, and a number of separate research activities based on GAIM technology are underway.
The purpose of GAIM-MURI was the development of a new generation of ionospheric model based upon near-real-time data assimilation. The idea was the joint vision of Robert McCoy at ONR and Paul Bellaire of AFOSR, derived from the recognition of considerable progress in the last 50 years or so by meteorological investigators in the use of various filtering, data assimilation and variational methods. McCoy points out that more data is always a good thing, but that skill in forecasting requires a precise knowledge of the source of errors. He also notes that raw data assimilation is preferable to the assimilation of secondary products that are derived from data sets. An example of a secondary product would be (vertical) TEC derived by conversion of oblique GPS-TEC measurements. In any case, it is felt that GAIM and its follow-on technologies are certainly a step in the right direction. The telecommunications community eagerly awaits the outcome of this research. It is hoped that versions of the code, or Internet access to specified output data files, can be made available to the public. Schunk et al.  and Wang et al.  outline the two independent DoD-sponsored MURI efforts to develop a GAIM model. Additional work by the groups is described by Scherliess et al.  and Hajj et al. ,
It has been recognized for some time that empirical models, with an initial value for a driving parameter, can be updated by forcing the output to match data, thereby deducing an altogether new value of the driving parameter. Using the new value of the driving parameter, we find that the model should do a better job at matching the real world, provided correlation distances are sufficiently long and the raw data is not error prone. This general procedure has been used for HF communication networks for many years [Goodman, 1991]. But GAIM is far more elegant and less simplistic, although the starting point in the GAIM methodology exhibits certain similarities to some of the HF updating methods.
GAIM, as a MURI program, was completed in 2004; however aspects of the program have transitioned to a quasi-operational phase. Work is continuing at NOAA-SEC and the Air Force Weather Agency (AFWA). Fuller-Rowell et al.  and Mintner et al., 2004] of the SEC group have examined the data assimilation of neutral thermospheric species during geomagnetic storms. This research is important since it is well known that the ratio of certain thermospheric species can be the most important driver in foF2 variations during magnetic storms. The research of the SEC group includes a comparison of Kalman filtering with nudging, and it was concluded that the former method is superior. Nudging is the simplest approximation to the Kalman filter in that it simply ingests raw data into the model without attempting to correct for observational errors. Mintner et al.,  maintain that this is equivalent to setting the Kalman gain equal to unity, with full acceptance of recent data, and neglecting the propagation state.
It should be noted that specialized data assimilation efforts have been undertaken, as derivatives of GAIM. Sojka et al.  have examined data assimilation for ARGOS LORAAS tomographically reconstructed electron density profiles near the equator. Keskinen and Dymond  have described data assimilation techniques for mesoscale space weather forecasting. They consider Kalman filters, direct data insertion, the so-called nudging processes, and variational methods. Of special interest is an application associated with spread-F bubbles or plumes.
Was this article helpful?