COSPAR Session C41 Updating Ionospheric Models with Ground and Space Data


Session Report, Dieter Bilitza


A 2-day session on the Updating of Ionospheric Models with Ground and Space Data was held during the 2008 Scientific Assembly of the Committee on Space Research (COSPAR) in Montreal, Canada. The session was organized by the joint COSPAR-URSI Working Group on the International Reference Ionosphere (IRI) and was divided into oral and poster sessions with the oral sub-sessions being focused on ‘Updating Ionospheric Models’, ‘TEC Data and Models’, F-peak Mapping’, Topside Modeling’, ‘Storm Modeling’, ‘New Inputs for IRI’, and ‘Solar Cycle Effects’. A business meeting of the IRI Working Group was also convened during the COSPAR meeting. We briefly report on the papers presented and the resulting decisions for the next update of the IRI model.


Electron Density Profile


The IRI electron density profile is normalized to the F-peak density and height and the profile above and below is determined by several characteristic bottomside and topside parameters. A major improvement of the representation of the IRI bottomside profile will be achieved with the introduction of the analytical model of Altadill et al. (Ebro Observatory, Spain). Their model is based on ionosonde data and shows a more accurate and continuous description of annual and seasonal variations than the current IRI model which uses step-wise interpolation and only a limited set of ionosonde inputs.

            Topside modeling is benefiting from the large volume of new topside profiles that has become available through the Alouette/ISIS data restoration effort at GSFC (Benson, Bilitza) and also from the Vary-Chap approach developed by Reinisch and Huang (U Mass Lowell, USA). This approach uses a height-varying scale height function that describes the transition from the O+ dominated F-region to the H+&He+ dominated upper topside. Important parameters in this approach are the starting scale height at F-layer heights (Hm) and the transition height ht. Reinisch et al. (UML, USA) deduce Hm from ISIS-2 topside sounder profiles and noted only small diurnal variations contrary to what is found for the bottomside scale height; the values for Hm derived from topside and bottomside side profiles differ by a factor between 1 and 3 depending on season and time of day. Using COSMIC occultation data Liu et al. (IGG, Beijing, China) study scale heights at 400km altitude and find a noontime maximum most notably near the magnetic equator. They also note the presence of a significant longitudinal structure in the equatorial sector. Ratovsky et al. (ISZF, Irkutsk, Russia) studied scale heights derived from Irkutsk incoherent scatter radar (ISR) measurements and also see significant diurnal structure similar to the Millstone Hill and Arecibo ISRs. The new IRI-2007-Corrected option reproduces the daytime variation but overestimates nighttime scale heights. They also observed significant differences to the F-peak scale heights deduced from digisonde bottomside measurements.


Mapping the F-peak Density and Height


A special IRI task force team is working towards new models for the F-peak density and height. The models currently used in IRI and other models are more than two decades old, the most widely used model, the CCIR model, dates in fact back to the sixties. A new modeling effort is long overdue and can take advantage of a much increased volume of ionosonde data, although still heavily biased towards the Northern mid-latitudes. The team has made an effort to locate and encourage all data providers to participate in the mapping effort. As base model the group will use the Neural Network model developed by McKinnell (HMO, Hermanus, South Africa) and Oyeyemi (University of Lagos, Nigeria). Newest results of this model with special focus on the equatorial ionosphere and the Equatorial Ionization Anomaly (EIA) were presented by McKinnell and Oyeyemi. The mapping group intends to work closely with the Radio-propagation working group of the International Union of Telecommunications (ITU) that is supporting the worldwide radio and telecommunication community and agencies with advice on ionospheric predictions.     


New Inputs for IRI


Of special interest is a better description of the day-to-day ionospheric variability. This item is very high on the priority list of IRI members and users. A first-order model based on a large volume of ionosonde data was developed by Araujo-Pradere et al.  (U Colorado, Boulder, USA) describing the variability during quiet-time as well as storm-time conditions. With the inclusion of satellite data it is planned to extend his model to the topside and widen the solar activity range. Zhang and Holt (MIT, USA) determined the variability for the whole profile at mid-latitudes based on the long record of ISR measurements from Millstone Hill and St. Santin and developed a model that describes the variability at different altitudes, seasons, and solar activities. The STROM model of Fuller-Rowell et al. (U Colorado, Boulder, USA) is used for the description of storm effects on the F2 peak density. Results of a comparative analysis of European ionosonde data show that the STORM model captures quite effectively the negative phases of the summer ionospheric storms, while electron density enhancement during winter storms and the changeover between the different storm phases is reproduced with lower accuracy (Buresova et al., IAP, Prague, Czech Republic).

            New efforts are underway to improve the description of the auroral E-region ionosphere. Data from the TIMED GUVI (Zhang et al., JHU/APL, USA) and SABER (Mertens et al., NASA/LRC, USA) instruments are used to describe auroral boundaries, their changes with magnetic activity, and the influence of auroral protons and magnetic storms on the E peak density and height. Comparisons during a TIMED pass over Sondrestrom showed that the SABER-deduced densities agree well with the ISR measurements. An new and essentially continuous data source for this region and altitude range was presented by Lattek, Singer et al. (LIAP, Kühlungsborn, Germany) noting that electron density profiles in the altitude range 55 km to 90 km can be obtained with the MF Doppler radar near Andenes, Norway.

            A systematic study of the Equatorial Ionization Anomaly (EIA) with CHAMP data at 440 km altitude revealed a strong intensification with increasing solar activity in the post-sunset sector, but only moderate variations during noontime. In addition, it was found that the crest regions are generally more sensitive to solar activity than the equatorial trough, and that the crest’s variation is seasonally dependent. IRI reproduces the noontime variations very well for all solar activity levels, but underestimates the EIA strength in the post sunset sector (Stolle et al., GFZ, Potsdam, Germany). Watanabe et al. (U. Hokkaido, Sapporo, Japan) discussed the longitudinal 4-peak structure as seen with CHAMP data that may be the result of diurnal atmospheric tides driven by weather in the tropics. IRI shows longitudinal maxima but not all 4 peaks are well developed.

An adaptive ray-tracing model was discussed by Al-Ubaidi (Baghdad University, Baghdad, Iraq).


Comparisons and Model Evaluation


De La Beaujardiere et al. (AFRL,USA) evaluated the impact of GNSS data on data assimilation models including the PBMod developed at AFRL and the GAIM developed by the University of Southern California /Jet Propulsion Lab. Their results show that both models overestimated TEC relative to the Jicamarca ISR and that the GAIM derivations of NmF2 and hmF2 can be degraded rather than improved, when COSMIC occultation data are assimilated. Other papers during the IRI sessions also noted the problems in deducing accurate peak parameters from the integral GNSS measurements. Comparisons with ionosonde data show that the techniques are getting better but still have significant error margins in areas of steep ionospheric gradients.

            Comparisons with GPS-TEC maps over Taiwan indicate that IRI-2007 overestimates TEC when F10.7 is less than 150, and underestimates TEC when F10.7 is larger than 200 (Kakinami et al., NCU, Chung-Li, Taiwan). A regional ionospheric model for the Brazilian sector was developed by Souza et al. (INPE, Brazil) based on results from the Sheffield University Plasmasphere Ionosphere Model (SUPIM). The model includes the F3 layer often observed at equatorial latitudes. The model will be helpful in improving IRI at equatorial latitudes.   


Plasma Temperatures


A major IRI task still pending is the inclusion of solar activity variations for the plasma temperature models. A large volume of insitu satellite measurements is being applied by Bilitza (GMU/GSFC, USA) and Truhlik (IAP, Prague, Czech Republic) towards this goal.  Most difficult is the F-region altitude range during daytime where the electron temperature increases, decreases, or stays constant with increasing solar activity depending on the season, time of day and latitude. During equinox the temperature decreases at mid-latitudes but increases everywhere else. During summer mid-latitude temperatures increase while winter temperatures decrease and the temperatures at low latitude are almost constant during solstice. A first model based on these results was presented benefiting also from input from Richards (GMU, USA) Field-Line Interhemispheric Plasma (FLIP) model for times and areas where no data where available or where data were inconclusive. Significant discrepancies were, for example, found between simultaneous DMSP in situ and Millstone Hill incoherent scatter radar measurements of electron temperature (Truhlik et al., IAP, Prague, Czech Republic). At low solar activities (= low electron densities) the DMSP temperatures exceeded the ISR measurements by up to a factor of 2.


Ion Composition and Drift


Park et al. (KAIST, Dae-jeon, Korea) compared DMSP F15 measurements of O+ and H+, of the low latitude nighttime topside ionosphere during the period of 2000-2004 with IRI predictions. The IRI variations with solar activity show a saturation effect at high solar activities not seen in the data and most likely induced by the correlation with electron density (IRI describes the percentage ion composition and assumes charge neutrality: total ion density = electron density). The IRI model overestimates the hydrogen ion density and underestimates the oxygen ion density and these effects are more pronounced when F10.7 is high.  Combining DMSP data at 840 km with Komsat data at 600 km Min (KAIST, Daejeon, Korea) studied the seasonal and solar activity variations of O+, H+, and He+ densities over a wider altitude range. They note that during nighttime topside H+ densities do not show significant variations with solar activity but are strongly influenced by season, which is in contrast to the oxygen ions, which show a more pronounced variation with F10.7 than with season.

            Su and Chen (NCU, Chung-Li, Taiwan) studied the solar flux effect on longitudinal/seasonal variations of ion density structures at the 600-km low-latitude ionosphere as observed by the ROCSAT-1 satellite.

            Abdu et al. (INPE, Brazil) studied the effects of solar activity on the equatorial vertical ion drift with digisonde data from Sao Luis and Fortaleza pointing out discrepancies with the current IRI drift model. A similar effort by Oyekola and Oluwafemi (Covenant University, Nigeria) used ionosonde data from Ouagadougou, Burkina Faso. On average, the values of daytime and nighttime ionosonde-derived vertical drifts are smaller by about a factor of four than measurements in other equatorial regions using different experimental techniques. Both efforts note that the diurnal variation is dominated by the characteristic morning peak and evening prereversal enhancement (PRE) velocities. Seasonal and solar cycle effects are prominent near the dusk sector with an increase of PRE from solar minimum to maximum. The average equinoctial evening prereversal enhancement increases by almost a factor of three from low to high flux.


New Members, Publications, and Future Workshops


Four new members were elected to the IRI Working Group: (1) Eduardo Araujo-Pradere, of the University of Colorado in Boulder, USA is one of the authors of the STROM model now used in IRI  and is very active in modeling ionospheric variability based on the long record of ionosonde measurements; (2) Elijah Oyeyemi of the University of Lagos in Lagos, Nigeria is working on the new NN model for the F-peak density; (3) Dalia Buresova of the Institute of Atmospheric Physics in Prague, Czech Republic was one of the main local organizers of the very successful 2007 IRI-COST Workshop in Prague and her scientific interest is in the ionopsheric variability of the region below the F-peak; (4) Andrzej Krankowski of the University of Warmia and Mazury in Olsztyn, Poland is an expert in GPS-TEC mappings and is currently the Chair of the Ionosphere Working Group of the International GNSS Service (IGS).

            Refereed papers from the IRI meetings are slated for publication in the Journal of Advances in Space Research. The latest issue was published as Number 4 of Volume 42 including papers from the IRI session during COSPAR-2006 in Beijing, China and from the 2006 IRI Workshop in Buenos Aires, Argentina. Editing work is under way for the papers from the Prague IRI Workshop. Because of the large number of contributions the papers will be distributed over two issues of Advances in Space Research.

            The next IRI Workshop is now planned for the week of November 2 – 6, 2009 in Kagoshima, Japan and will be organized by Shigeto Watanabe of Hokkaido University in Sapporo, Japan. There will also be a special IRI Task Force Activity on the Real-Time IRI in Boulder (or Colorado Springs) on May 4-6, 2009; this is the week following the Space Weather Week. The meeting will be organized by Tim Fuller-Rowell and Eduardo Araujo-Pradere of the University of Colorado in Boulder, USA. The IRI team has submitted a session proposal for the 2010 COSPAR General Assembly in Bremen, Germany on the topic of the Representation of the Auroral and Polar Ionosphere in IRI.

            Access to IRI is now also available through the Community Coordinated Modeling Center (CCMC) at, and the Virtual Model Repository (VMR) is using IRI for data-model comparisons. More information about the IRI project and access to the IRI Fortran code, web interface, and related links can be found on the IRI homepage at

* Back to IRI workshops page
* Back to IRI homepage