The new model for the F2 bottomside parameters B0 and B1 was presented (Bilitza, USA), and discussed in great detail. This model was developed as a result of the earlier ICTP task force activities. It is based on data from African, Argentinian, Peruvian and Chinese stations that were acquired and processed in preparation for the task force activities. The most important improvement can be seen at low and (magnetically) equatorial latitudes were the new model provides values much larger than the old IRI and in much better agreement with the measurements than the old IRI. The special focus of the 1999 task force activity was on resolving contradictory results from various stations. At low latitudes considerable differences were found for the monthly average noon value during high solar activity for Ougadougou and Ibadan. This is most likely due to the limited number of profiles available from Ibadan. New data were presented for Kohrogo, Ivory Coast (Adeniyi, Nigeria) and Jicamarca, Peru (Reinisch, USA). These data together with future Jicamarca data (during the upcoming solar maximum) will be used to study future improvements of the B0, B1 models especially for the solar maximum period. For mid-latitudes differences were found in the diurnal variation of the B0 and B1 parameters (Buresova, Czech Republic). It was found that this was to a large part the result of using different ionogram reduction programs (Artiste, Polan) and the differences between these two methods in defining the nighttime starting height. Data from Pruhonice, Czech Republic (Buresova) and from El Arenosillo, Spain (Mirò, Spain) were re-evaluated with the improved method during this task force activity. The final results agree much better with the diurnal variation of the B0 and B1 parameters as known from other mid-latitude stations.

New input data for the B0 and B1 modeling effort were also presented by Zhang (China). He has studied several year’s worth of data from the Japanese MU radar facility and has developed a time and season dependent model of B0 for this mid-latitude station. Other IRI parameters compared with MU measurements included the F maximum height and the F region electron temperatures. For hmF2 good agreement was found for low solar activity and systematic discrepancies at high solar activity. An anti-correlation with solar activity was found for the electron temperature contrary to what is currently included in IRI.

Two topics were discussed: (1) the ARP data base established for this activity and (2) a Windows NT/95/98 interface to IRI.

(1) The Average Representative Profile (ARP) technique developed at the University of Massachusetts Lowell (UML) is a method for deducing a representative average profile from a collection of profiles, e.g. for obtaining a monthly average noon profile from the profiles measured each day at noon (Reinisch, Huang, USA). Using this technique a data base of representative monthly average profiles has now been established including data from Tucuman, San Juan, Buenos Aires and Ushuaia in Argentina (Mosert, Argentina), from El Arenosillo and Tortosa/Ebro in Spain (Mirò, Spain) and for Ougadougou in Burkina Faso (Adeniyi, Nigeria). This is a growing data base with more data and stations being added in the future. Huang (USA) described in great detail the latest version of the UML programs NHPC (inversion of ionograms to electron density profile) and CARP (computation of ARP profiles). The bottomside profile parameters B0, B1 and D1 obtained through these methods and data base will be a treasure chest for future improvements of the IRI electron density profile in the middle ionosphere. D1 is the profile parameter used for the revised representation of the intermediate region. From a first evaluation of available D1 values the UML team proposed the following formula:

SS is the time of sunset and SR of sunrise.

(2) Huang (USA) presented and demonstrated his Windows interface to the IRI programs. This will be a great help for IRI users and ideally complements the IRI Web interface that is maintained by NASA’ s National Space Science Data Center (NSSDC). IRI has a long tradition of making electronic access to the model as easy as possible. A PC/DOS interface developed at NSSDC in the eighties and distributed on floppy disk was widely used at that time. The new Windows interface replaces this earlier system with a much more sophisticated and very user friendly GUI and the added capability of producing line plots and contour plots of IRI parameters. It also includes an algorithm for predicting the Rz12 and IG12 indices based on the past indices. The Windows version of IRI already includes the revised profile description for the intermediate region.

The presentations highlighted the shortcoming of the present IRI topside profile and provided status reports from several modeling efforts. Reinisch and Huang (USA) have developed a method for obtaining the topside scale height from ionosonde measurements assuming a Chapman profile with a constant topside scale height. The reliability of this method was demonstrated in comparison with Total Electron Content (TEC) measurements at several locations. Radicella (ICTP) has combined ground ionosonde measurements, the DGR model, and Intercosmos 19 topside sounder data to produce a model representation of the whole ionosphere. Using Alouette and ISIS topside sounder data Bilitza (USA) is working on a new model for the topside profile based on two characteristic anchor points (the inversion height and the ion transition height). The COST 251 model for the topside profile is based on work by Leitinger (Austria) and Titheridge (New Zealand).

Comparing IRI predictions to a large database of Alouette and ISIS topside sounder data, Bilitza (USA) found good agreement in the lower topside and an overestimation of the sounder data by a factor 2 to 8 in the upper topside (most prominent at equatorial and high latitudes). Ezquer and Mosert (Argentina) have used Hinotori insitu satellite measurements (low latitude, 600 km altitude) and almost simultaneous ionosonde soundings to investigate the reliability of the IRI topside model. They find that IRI underestimates the satellite measurements and that this is a shortcoming of the topside shape function and parameters, and not of the F peak parameter models. Taken together with the results of the Alouette/ISIS study (Bilitza) this indicates that in the current IRI profile function the height gradient is first too large (around 600 km) and then too small (800k m and above). Pulinets (Russia) described the data base that has been accumulated by the Intercosmos 19 topside sounder experiment from 1979 to 1981 (inclination: 72 degrees; orbit period: 99 min; 500 — 1000 km) and indicated how these data could help in improving the IRI topside model. He also presented a very interesting talk about the topside sounder experiments that is now set up on the MIR space station (altitude: 350 km; inclination: 45 degree). He reported first results from a measurement campaign over Taiwan.

One of the prime data sources for the Total Electron Content (TEC) is the fleet of GPS satellites. Several groups have developed algorithms for deducing TEC from the GPS measurements. It had been suggested to use the IRI model to test the reliability of the various methods. This could be done by using an IRI specification of the global ionosphere and by computing the expected GPS measurements in such an ionosphere. One could then test a specific method by calculating TEC from the IRI-simulated GPS data and compare the calculated TEC to the actual IRI TEC. Ciraolo (Italy) reported the results of such a study for the IROE (GP-TEL) algorithm. The IRI test was good for identifying specific conditions for which the method needed improvements. For these IRI applications it is critically important to extend IRI to plasmaspheric heights since the GPS satellites orbit at geostationary altitudes.

The IROE data base of NNSS TEC observations was described by Spalla (Italy). Comparisons of slant TEC data with IRI simulations show generally good agreement. But often a small discontinutiy was found in the latitudinal variation of the IRI slant TEC. Studying this problem in more detail during the afternoon computer session, it was found that the discontinuity is due to discontinuties in the B0 bottomside thickness parameter. This shows the importance of a reliable B0 model not only for the shape of the bottomside profile but also for the integral electron content. A test with the new B0 model still showed a discontinuity, although a much smaller jump in value. The discontinuty, it was found, was not because of a discontinuity in the B0 model but because of discontinuity in the way the F- and E region profiles were merged. Under certain special conditons the current IRI cannot merge the two profiles and instead implements an artificial linear transition region. The modified merging algorithm proposed by Huang and Reinisch (USA) during an earlier Task Force Activity (and now accepted for the next version of IRI, IRI-2000) overcomes this shortcoming.

Several topics were discussed as focus for the Task Force Activity in the year 2000 and beyond. Of particular interest are (1) a better description of the nighttime E region and (2) a quantitative models for the variability of ionospheric parameters, e.g. a model that would describe the standard deviation (from a monthly mean) in electron density (or plasma frequency) for different altitudes, latitudes, longitudes, hours, seasons, solar and magnetic activity indices.