Advantages of the new model of IRI (IRI-Plas) to simulate the ionospheric electron density: case of the European area

Satellite telecommunications, positioning and navigation systems require knowledge of the electron distribution in height Ne(h) to high-altitude orbits of satellites. One of the possibilities to construct such profiles is associated with the use of the ionospheric total electron content TEC. This paper is devoted to three advantages of the IRIPlas model. They include introduction of the topside basis scale height Hsc, expansion of the IRI model to the plasmasphere, ingestion of experimental values of TEC. Testing of this model according to different satellite experiments (CHAMP, DMSP) shows the high efficiency of this model. The method of adaptation of the IRI-Plas model to the plasma frequency at altitudes of CHAMP and DMSP satellites allows us to produce behavior of Ne(h)-profiles during the disturbances, as well as to refine the values of TEC, which determine the accuracy of positioning. Results were obtained using data of the European area.


Introduction
The operation of the various satellite communications, navigation, positioning systems depends on the state of the ionosphere and needs to know the electron distribution in height Ne(h) in near space.In application of radio and satellite communications, the empirical model of the ionosphere IRI (Bilitza, 2001) is most widely used, but it determines the Ne(h)-profile to a height of 2000 km.Ability to determine the profiles at higher altitudes is associated with the total electron content TEC.However, the IRI model gives a large discrepancy when compared with the experimental TEC because of the profile shape of the topside ionosphere, so that the model has been modified several times in this century (IRI2001, IRI2007, Bilitza, 2001;Bilitza, Reinisch, 2008) and the modification is continuing (in 2010, a new version of IRI2010 (Bilitza et al., 2010) was proposed).The next step can be connected with the IRI-Plas model based on papers of T. Gulyaeva (e.g. Gulyaeva, 2003, 2011).The main advantages of this model are introduction of the topside basis scale height Hsc, expansion of the IRI model to the plasmasphere, ingestion of experimental values of TEC.Previous versions allowed us to adapt the model to two experimental parameters of the ionosphere: the critical frequency foF2 and the maximum height hmF2.Now there are three parameters.This should allow to determine the more precise shape of Ne(h)-profiles.The aim of this work is: (1) determination of the behavior of Ne(h)-profiles when adapting the model to three parameters, (2) clarification of the values of TEC by means of adaptation of the model to the plasma frequency fne(sat) at altitudes of satellites CHAMP and DMSP.The IRI-Plas model was taken from site ftp://ftp.izmiran.rssi.ru/pub/izmiran/SPIM/.The values of the parameters foF2 and hmF2 are taken from the database (http://spidr.ngdc.nasa.gov/spidr/index.jsp),TEC values are taken from IONEX files of global maps of JPL, CODE, UPC, ESA (ftp://cddis.gsfc.nasa.gov/pub/gps/products/ionex/).Data of fne(sat) was taken from http://cindispace.utdallas.edu/DMSP/dmspdata at utdallas.html(Langmuir Probe Data), http://isdc.gfz-potsdam.de/.Juliusruh is chosen as a reference station.In addition, the results for other stations of the European area are presented.

Testing the model IRI-Plas according to different experiments
Testing by means of satellite data is carried out for 4 cases used in various applications: (1) the initial model IRI, and models adapted to: (2) the experimental value of foF2, (3) the experimental value of the TEC, (4) experimental values of foF2 and TEC, to demonstrate the difference between the results for these methods.Option 1 is used when there is no current information, and determines the average (quiet) A substantial difference between foF2(obs) and foF2(IRI) is seen.The profiles shown by black dots present the profiles for the initial model.The profiles shown by green triangles correspond to the traditional adaption techniques when the construction of N(h)-profile uses the experimental value of the TEC and the model value of foF2 (McNamara, 1985;Houminer andSoicher, 1996, Gulyaeva, 2003), as well as assimilation techniques involving TEC (e.g., Khattatov et al., 2005).The difference can be large.The best agreement corresponds to the new version.It allows us to illustrate how can change N(h)-profiles, adapted to the foF2 and TEC during the disturbances of various types.A positive deviation is seen for the daytime.At night, negative deviations dominate with lower gradients in the topside.Another example refers to a period of quiet Sun, which attracted a great attention of the scientific community because of the unusual duration (e.g.Liu et al., 2011;Zakharenkova et al., 2011).In terms of the perturbation this period is characterized by the absence of strong and moderate disturbances.The disturbances were observed in the form of positive bursts of TEC (Cander and Haralambous, 2011) synchronously covering the European zone from Chilton to Athens.Example of the TEC burst and corresponding variation of foF2 for the Juliusruh station for 22-24 January 2006 is displayed in Fig. 4. Behavior of Ne(h)-profiles for these days and UT = 8-18 is given in Fig. 5 along with profiles for medians.At the time UT = 8, Ne(h)-profiles are still little different from the medians.At the time UT = 10, only ionization of the topside part increased.At the time UT = 12, a gain propagated to lower ionosphere, reaching a maximum at UT = 14.At the time UT = 16, decrease in the ionization of the topside part occurred most rapidly.At the time UT = 18, peaks of TEC and foF2 have completely disappeared.This illustrates that development of disturbance is connected with transition of ionization from plasmasphere to the region near the F2 maximum.This behavior is characteristic for other mid latitude stations (Chilton, Ebre, Athens, Leningrad, Moscow), and for other cases.Different behavior of Ne(h)-profiles is visible for the Tromso station, as shown in Fig. 6.The perturbation starts earlier (at UT = 0) and passes the maximum phase in UT = 12.Some profiles seem unusual, but a comparison of profiles in the times when measurements of fne were in the satellite CHAMP (Fig. 7) shows a complete coincidence for adapted profiles.Since it is difficult to obtain the moments of coincidence between measurements of the satellite and the TEC, Fig. 7 shows the results for the nearest moments that are indicated in parentheses.The difference between profiles for four versions is great.

Examples of the behavior of Ne(h)-profiles during the disturbances
The above examples show that the profiles adapted to both of the foF2 and TEC parameters can be an important tool for studying the behavior of the ionosphere.N(h)-profiles adapted to one of the parameters may be far away from the real profiles.

Refinement of the TEC values from satellite experiments
There is a variety of the TEC values, which can be used to adapt the model.These are: (1) measurement at individual receivers, the values of which can differ up to ∼ 10 TECU of each other (e.g.Choi et al., 2010), (2) the global maps, the values of which can differ up to 2-3 times (Arikan et al., 2003), (3) tomographic measurements (Chartier et al., 2012) and others.A typical example of differences was done in Fig. 7 of the paper (Arikan et al., 2003).This These examples and statistics from many stations and geophysical conditions show that the values of TEC corresponding fne(DMSP) may lie within the range of maps below the weighted average IGS values (Hernandez-Pajares et al., 2009), but may be below that range.

Conclusions
TEC is the most important parameter of the ionosphere used in scientific and engineering applications, so there is a need of its model.One such model is the IRI.This paper deals with the IRI-Plas version of this model, which has advantages of a rigorous account of the plasmaspheric part of the profile, the introduction of Hsc parameter to determine the shape of the upper part of the profile, ingestion of the TEC.We used   one of the parameters, or even three parameters (foF2, hmF2, TEC) does not always lead to the Ne(h)-profile corresponding plasma frequencies fne(sat), and (3) an adaptation of the model to fne(sat) can lead to new values of TEC, which can lie within the existing set of values, and may be below that set.We can point to two possible ways to use these values.If we can specify a value of "true" TEC, the difference can be attributed to the plasmaspheric part and thus we get the way of correction of this part, which is still of some interest (e.g., Cherniak et al., 2012).If this value cannot be specified, then the values of TEC for profiles adapted to fne(sat), can be considered as an improved value of the TEC.For applications in navigation and positioning, an empirical ionospheric model should be used to minimize the ionospheric errors.The Klobuchar model (e.g., Klobuchar, 1987) widely used for this aim provides a 50 % correction on a global basis.It is possible to suppose that the IRI-Plas model can compensate the ionospheric error with much higher accuracies.These results and conclusions were obtained for the European region.It is not difficult to obtain such results for any other region of the globe.

Fig ( 1 )Fig. 1 .
Fig (1) Fig. 1.Comparison of Ne(h)-profiles corresponding to the initial IRI model (black circles) with Ne(h)-profiles of the model adapted to experimental values of foF2 (red squares), experimental values of TEC (green triangles) and joint values of foF2 and TEC (violet asterisks).The red circles show values of the plasma frequency of the CHAMP satellite.The left panel illustrates case of the positive disturbance, the right panel applies to negative one.

Fig ( 2 )Fig. 2 .
Fig (2) Fig. 2. The behavior of the plasma frequency fne(CHAMP) near noon (left panel) and midnight (right panel) along with medians of these values to detect periods of positive and negative disturbances in the topside ionosphere.

Fig ( 3 )Fig. 3 .
Fig (3) Fig. 3.The behavior of Ne(h)-profiles during disturbances in April 2001.Black circles display the median profiles obtained by the model adapted to experimental medians of foF2 and TEC.The other profiles apply to days indicated by corresponding marks.The left panel concerns to time near noon (UT = 13), the right panel -time near midnight (UT = 1).

Figure 2
Figure2shows the behavior of the plasma frequency fne(sat) according to CHAMP satellite in April 2001 over the Juliusruh station, along with the median to detect cases of disturbances in the upper ionosphere.In the left panel are cases that fall near noon, on the right side -near midnight.Figure3shows Ne(h)-profiles for the Juliusruh station together with Ne(h)-profiles calculated for the experimental medians of foF2 and TEC (the lines shown by black dots).A positive deviation is seen for the daytime.At night, negative deviations dominate with lower gradients in the topside.Another example refers to a period of quiet Sun, which attracted a great attention of the scientific community because of the unusual duration (e.g.Liu et al., 2011;Zakharenkova et al., 2011).In terms of the perturbation this

Fig ( 5 )Fig. 5 .
Fig (5) Fig. 5.The behavior of Ne(h)-profiles during the burst of TEC on 22-24 January 2006 over the Juliusruh station.The Ne(h)-profiles shown by black circles were obtained by the model adapted to experimental medians of foF2 and TEC.The Ne(h)-profiles shown by red circles apply to the day of the burst (23 January 2006).

FigFig. 6 .
Fig (6) Fig. 6.The behavior of the Ne(h)-profiles during the burst of TEC on 22-24 January 2006 for the Tromso station.Designations are the same as on Fig. 5.

Fig ( 7 )Fig. 7 .
Fig (7) Fig. 7. Ne(h)-profiles during the burst of TEC on 22-24 January 2006 in the Tromso station at moments, close to passages of the satellite.

Fig ( 8 )Fig. 8 .
Fig (8) Fig. 8. Differences of TEC calculated from the various global maps (left panel) and TEC obtained by adaptation of the model to the plasma frequency fne of the DMSP satellite (right panel).

Fig ( 9 )Fig. 9 .
Fig (9) Fig. 9. New TEC values in comparison with data of global maps of JPL, CODE, UPC, ESA and weighted average IGS values.Results are given for days (not shown in the figure) and time UT of the satellite passages indicated on the x-axis.