IRI 2003 Workshop

Meeting Report / Dieter Bilitza


With the 2003 Workshop the International Reference Ionosphere (IRI) team convened for the first time on the African continent.  The meeting was held at Rhodes University in Grahamstown, South Africa from October 6 to 10. The primary topic of the workshop was  “Quantifying Ionospheric Variability”. The workshop agenda included 38 papers authored by scientists from Argentina, Australia, Austria, Brazil, Bulgaria, Canada, Cuba, Czech Republic, India, Italy, Japan, Nigeria, South Africa, Spain, U.K., and USA. The talks were grouped into sessions entitled “Lower and Bottomside Ionosphere”, “Ionospheric Variability”, “Total Electron Content”, “IRI Applications and Latest Results”, “Topside Ionosphere”, “Equatorial Ionosphere”, and “Final Discussions”. 

The local organizing committee (L.-A. McKinnell, A. Poole) provided excellent support with all aspects of the meeting including transportation and accommodation arrangements. A special thank you goes to L.-A. McKinnell who made this workshop possible through her dedication and untiring efforts before, during, and after the meeting. An excursion to the Schotia Game Reserve was one of the many highlights of this memorable workshop.

       Financial support for the workshop was provided by Rhodes University, the Grintek Ewation Company, the Abdus Salam International Centre for Theoretical Physics (ICTP), the International Union of Radio Science (URSI), the Committee on Space Research (COSPAR), Hermann Ohlthaver Institute for Aeronomy (HOIA), the South African National Research Foundation (NRF) and the US National Science Foundation (NSF). ICTP is an organization funded by the United Nations Education, Scientific, and Cultural Organization (UNESCO), the International Atomic Energy Agency (IAEA), and the Italian Government.

       What follows is a brief summary of the workshop presentations and discussions with special emphasis on the decisions made regarding the next version of the IRI model.




Several speakers presented results from the IRI Task Force Activity that meets annually at the ICTP and that in recent years has focused on the development of a variability model for IRI. Bilitza (USA) provided a progress report and pointed out some of the roadblocks that still have to be overcome. Currently the relative inter-quartile range (RIQ=[upper quartile – lower quartile]/median) is used as measure of monthly variability and the ionospheric parameters considered include foF2, hmF2, foF1, foE, M3000F2, B0, B1, D1, and TEC. RIQ alone, however, is not sufficient since it does not describe the often highly unsymmetrical distribution of data with respect to the monthly median. Upper and lower quartiles and/or deciles could be used to better characterize the data distribution. Bradley (U.K) pointed to his earlier studies that had shown that the data distribution could  be quite well represented based on the upper and lower deciles. Mosert (Argentina) and Amarante (Spain/Italy) reported about their variability studies using South American and European ionosonde data, respectively, and also about the efforts of Cuban colleagues (Alazo, Lazo, and Calzadilla) using data from the ionosonde CD-ROM of the National Geophysical Data Center (NGDC). Generally they find that RIQ is larger at night than during daytime reaching the highest values during sunrise, that it decreases with increasing solar activity, and that it increases towards high latitudes with a latitudinal maximum often found near the crests of the equator anomaly. Significant deviations from these general trends, however, are found for individual stations and need to be resolved before a variability model can be established. These discrepancies may be also related to the very disturbing data problems (missing data, repeated values, extra data, etc.) that were reported by Weatherhead (USA). She had encountered these problems during her statistical studies of long-term trends involving large amounts of ionosonde data from NGDC. A prime focus of such long-term studies is to look for the ionospheric imprint of the tropospheric greenhouse effect. Cannon’s (U.K.) study of 50 years of E and F peak parameters collected by the Tromso (Norway) ionosonde finds a decreasing trend for all parameters as would be expected from the greenhouse effect (cooling of the mesosphere). Other earlier studies involving many stations have found positive as well as negative trends. Recognizing the importance of data quality control for such long-term studies as well as the IRI-related variability studies, the IRI team will urge NGDC to address these issues.

The workshop provided a good review of prior efforts to quantitatively describe ionospheric variability, especially for radio wave propagation applications (Bradley, U.K.; Wilkinson, Australia; Mosert, Argentina). Specifically mentioned were two models that provide relative deciles for different times and location: (1) the tabulation of values by Davis and Groome (1964) and (2) the decile factors recommended by ITU-R (1997). Comparing these models with measurements from an Australian ionosonde Wilkinson (Australia) finds slightly better agreement with the ITU-R (1997) model.

Bradley (U.K.) presented results from a European study (Kouris and Fotiadis, Greece; Stanislawska and Juchnikowski, Poland) that looked at variability of different ionospheric parameters and the correlation between these variabilities. They find that the variability of foF2 clearly exceeds the variability of foE, foF1 and M(3000)F2 and that no strong correlations exist between the variabilities of these parameters. Codrescu (USA) presented the results of a study by Araujo-Pradere and Fuller-Rowell (USA) who used ionosonde data from 43 storm periods to study the changes in variability with varying magnetic activity.  Studying the variability in the bottomside ionosphere at different heights Amarante (Spain/Italy) and colleagues confirmed the variability maximum just below the F2 peak (a result of the very steep gradients in this region).

Truhlik (Czech Republic) used data from the ACTIVE mission to study the variability of the ion composition and the electron temperature in the altitude range from 500 to 2500 km. The variability is generally greater during night than during day and increases with latitude and altitude. Codrescu (USA) reminded the participants about the strong variability of the equatorial ion drift and used a coupled thermosphere-ionosphere-plasmasphere model to investigate possible sources for this variability (neutral wind changes and their resultant dynamo action).   


Electron Density


D-Region: With IRI-2000 two new options were introduced for this region (models by Friedrich, Austria, and Danilov, Russia). But users have had considerable technical problems with this part of the IRI-2000 program because of the large BLOCKDATA statements and the differences of various FORTRAN compilers in dealing with these statements. An additional problem is the fact that only the older D-region model provides a seamless merging with the rest of the profile, whereas the two new options do not. It was decided to generate a special D-region version of IRI with all three options and to only include the older D-region model in the full model version. A new and promising Neural Networks (NN) approach to modeling the auroral D- and E-region was presented by McKinnell (South Africa).


E-Region:  Nighttime remains a problem because of the scarcity of reliable data. Theoretical computations (e.g., Titheridge, 2003) predict higher nighttime values than IRI. Abdu (Brazil) found that he also needed to increase the nighttime IRI foE values (E-region conductivity) to get agreement between measured and computed values of the F2 peak height and ion drift.


F Peak and Bottomside: The shape of the IRI bottomside profile is defined by the F2 bottomside thickness (B0) and shape (B1) parameters and by the F1 layer thickness parameter D1. Ionosonde measurements from Pruhonice (Czech Republic) show overall good agreement with IRI (Mosert, Argentina, and Buresova, Czech Republic). Data from the Ebro (Spain) station are used to study the variation of D1 with solar activity and season (Buresova, Czech Republic, D. Altadill, Mosert, Argentina, and Amarante, Italy/Spain) as a starting point for a better description of this parameter in IRI.

Evaluating IRI with data from Brazilian ionosonde stations near the magnetic equator and near the anomaly crests, Batista and Abdu (Brazil) find that in general the URSI maps provide better foF2 predictions than the CCIR maps. The largest discrepancies are found at nighttime near the equator. This could be due to the fixed epoch used for the magnetic field in IRI whereas the Brazilian data stretch over 15 years.

The utility of the Neural Networks (NN) technique for ionospheric modeling was illustrated with NN models for foF1 (Jacobs and Poole, South Africa) and for foF2 (Oyeyemi, Nigeria and Poole, South Africa). Fully developed these could be candidate models for future replacements of the CCIR-type peak parameter maps.

The storm effects on the different IRI F peak and bottomside parameters at mid-latitudes were studied by Buresova (Czech Republic), Mosert (Argentina), Altadill (Spain) and Amarante (Spain/Italy). They find the strongest effects for foF2 and TEC and a close correlation of the temporal variations of the effects on these two parameters. Similarly the effects on B0 and on the slab thickness are closely correlated and of similar magnitude. The storm effects in the F1 region are weaker than in the F2 region and almost always negative. Independent of the sign of the F2 region magnetic storm effect, the effect on the electron density at F1 region heights is prevailingly negative and shows substantial seasonal dependencies at European  middle latitudes.

Regarding spread-F probability models it was decided to include the South American model (Abdu, Brazil) in the next version of IRI. The hope is that this will initiate comparisons with other longitude zones and the development of a global model.


Topside and Plasmasphere:  Efforts continue to improve the IRI model in the region above 700 km where IRI overestimates topside sounder measurements. Different approaches have been reported at earlier meetings based on constraining specific model parameters and are now being studied for future inclusion in IRI (Coisson, Italy; Iwamoto, Japan; Bilitza, USA; Reinisch, USA; Gulyaeva, Russia). The long-term goal is to establish a database of topside scale height measurements combining space and ground data and to develop a new global model for this parameter. The improvement of the upper topside profile is also of critical importance for merging IRI with a plasmasphere model  (e.g. GCPM model of Gallagher et al., 2000). GPS-related modeling efforts were discussed by Cilliers (South Africa) for the South African region and by Oyama (Japan) based on the nearly 1000 GPS receivers in Japan.



Ion Composition, Temperatures, and Drift


Ion Composition: Watanabe (Japan) presented a model for the ion transition height (O+ to light ions) that he had developed with Marinov and Kutiev (Bulgaria) based on a Chapman-type profile analysis of Alouette and ISIS topside sounder data. The model describes variations with local time, season, geomagnetic latitude, solar flux (F10.7), and geomagnetic activity (Kp). Significant differences are found in comparisons with IRI, which need to be further investigated with other data sources.  


Electron Temperature:  Using Akebono data in conjunction with theoretical simulations Oyama and Watanabe (Japan) studied plasmaspheric processes during quiet and disturbed times. Both studies relied on the empirical model for the plasmaspheric electron temperatures that they had developed together with Kutiev (Bulgaria) based on the Akebono data. It was decided to include this very useful model in the next version of IRI. 


Ion Temperature:  First results from an effort to develop an ion temperature model based on ACTIVE data were reported by Triskova (Czech Republic). She presented a first version of a global Ti model for high solar activity and finds that at high latitudes the current IRI model systematically overestimates the ion temperature measurements.


Ion Drift:  Scherliess (USA) described two drift models that might be of interest for IRI: (1) The disturbance model describes the storm-time effects on the equatorial ion drift including short-lived prompt penetration electric fields as well as longer-lived disturbance dynamo electric fields. (2) The second model is a description of the global ion drift in terms of spherical harmonics based on satellite and incoherent scatter data. Both models were accepted for inclusion into the next version of IRI.




Abraham (USA) discussed the use of IRI for the data analysis of future ocean salinity satellite missions (Aquarius, SMOS). Ionospheric densities are required for the corrections of the L-band signal as well as for the elimination of ionospheric emissions. Coetzee (South Africa) reported about the many applications of IRI by South African agencies including, for example, direction finding, HF frequency predictions and path analysis. He stressed the important role of these techniques in the fight against crime, drug trafficking, and political instability on the African continent.

       Three new members were elected into the IRI Working Group: Lilijana R. Cander (Rutherford Appleton Laboratory, U.K.), Ludger Scherliess (Center for Atmospheric and Space Sciences, Utah State University, USA), and Michael Codrescu (Space Environment Center, NOAA, USA).

There will be an IRI-related session during the General Assembly of the Committee on Space Research (COSPAR) in Paris, France  (July 18-25, 2004). The special topic will be “Advances in Modeling the Ionospheric Temperatures and Ion Composition”. The 2005 IRI Workshop will be held at the Ebro Observatory in Roquetes, Spain, as part of its 100-year anniversary celebrations. Alberca (Spain) provided the Working Group with an excellent overview over the observatory location, facilities, and research areas and discussed transportation and lodging issues. The GPS group from the Polytechnical University of Catalonia in Barcelona (Hernandez-Pajares, Juan, Sanz) will also help with coordinating the meeting.     


* Back to IRI workshops page
* Back to IRI homepage