International Reference Ionosphere Predictions Research Articles

Morphologies of the topside ionosphere observed by COSMIC at high-latitudes during the 17 March 2013 magnetic storm

The present study considers effects on the topside ionosphere at high-latitude (|MLat| ≥ 60°) during the magnetic storm on 17 March 2013. We divide the day into two segments, with magnetic conditions quiet during the first 6 h and disturbed thereafter, thus allowing comparisons of differences between these two states. The COSMIC data source consisted of 272 electron density profiles (EDPs) measured by COSMIC satellites using Abel inversions that assume the absence of horizontal plasma density gradients. We compared resultant EDPs with predictions of the International; Reference Ionosphere (IRI) model for the same altitudes along individual tangent-point lines. As conditions allowed, we also compared EDPs with densities measured by nearby DMSP satellites. During the pre-storm quiet period, 65 of the sampled 68 EDPs were highly correlated with IRI predictions. The remaining 3 EDPs showed significant irregularities along parts of the 800–200 km altitude range. We divided stormtime 204 COSMIC EDPs into three morphological categories. In the first group, 130 EDPs exhibit shapes similar to IRI-predictions. Of these, 72 (9) had larger (smaller) than predicted magnitudes. The second group consisted of 37 EDPs that showed little variation in magnitude over the designated altitude range. The remaining group had another 37 EDPs marked by significant irregularities. We found that the deepest EDP irregularity resulted from GPS signals passing through a deep plasma density trough observed by DMSP. The trough extended along the equatorward boundary of the evening-side, auroral oval. We suggest that EDP irregularities reflect density gradients along or near EDP tangent point lines.

A neural network-based foF2 model for a single station in the polar cap

A Neural Network (NN) model has been developed for the critical frequency of the F2 layer (foF2) at Resolute (74.70∘ N, 265.10∘ E) using data obtained from the Space Physics Interactive Data Resource, SPIDR (no longer available), for the period between 1975 - 1995. This model is a first step towards addressing the discrepancies of the International Reference Ionosphere (IRI) foF2 or peak electron density (NmF2) at high latitudes recently presented by Themens et al. [2014]. The performance of the NN model has been evaluated using foF2 data obtained from the Canadian Advanced Digital Ionosonde (CADI) at Resolute (74.75∘ N, 265.00∘ E) for the period between 2009 - 2013, in comparison with the IRI predictions. The 2012 version, and the International Union of Radio Science (URSI) option of IRI has been used. The NN nighttime monthly median foF2 variation demonstrates good agreement with observations compared to the IRI. The NN model is able to reproduce the enhancements in foF2 during the equinoxes, and also shows an improvement during disturbed days. Root Mean Square Errors (RMSE) were computed between hourly and monthly median model predictions and observations, and on the whole, the NN model seems to perform better during low solar activity and the equinoxes. The NN model shows an improvement in performance on average by 26.638% for hourly foF2 and 32.636% for monthly median foF2, compared to 7.877% obtained for the same station by the most recent NN model that attempted to predict foF2 at a polar cap station [Oyeyemi 24 and Poole, 2005].

Diurnal variations of the ionospheric electron density height profiles over Irkutsk: Comparison of the incoherent scatter radar measurements, GSM TIP simulations and IRI predictions

The long-duration continuous Irkutsk incoherent scatter radar (ISR) measurements allowed us to obtain the monthly averaged height-diurnal variations of the electron density in the 180–600km altitudinal range for 4 four seasons (winter, spring, summer, autumn) and for two solar activity levels (low and moderate). Considering these electron density variations as “quiet ionosphere patterns” we compared them with the Global Self-consistent Model of the Thermosphere, Ionosphere and Protonosphere (GSM TIP) simulations and the International Reference Ionosphere (IRI) predictions. It was found that some observational features revealed from the ISR measurements are reproduced nicely by both the theoretical and empirical models, and some features agree better with the GSM TIP than with IRI. None of the models is able to reproduce a detailed multi-peak behavior of the electron density observed by ISR at ∼300km and above for the spring and autumn under low solar activity, while for the spring the GSM TIP tends to reproduce the morning and daytime peaks at the same local times as they are seen from the ISR observations.

Global 3‐D ionospheric electron density reanalysis based on multisource data assimilation

We report preliminary results of a global 3‐D ionospheric electron density reanalysis demonstration study during 2002–2011 based on multisource data assimilation. The monthly global ionospheric electron density reanalysis has been done by assimilating the quiet days ionospheric data into a data assimilation model constructed using the International Reference Ionosphere (IRI) 2007 model and a Kalman filter technique. These data include global navigation satellite system (GNSS) observations of ionospheric total electron content (TEC) from ground‐based stations, ionospheric radio occultations by CHAMP, GRACE, COSMIC, SAC‐C, Metop‐A, and the TerraSAR‐X satellites, and Jason‐1 and 2 altimeter TEC measurements. The output of the reanalysis are 3‐D gridded ionospheric electron densities with temporal and spatial resolutions of 1 h in universal time, 5° in latitude, 10° in longitude, and ∼30 km in altitude. The climatological features of the reanalysis results, such as solar activity dependence, seasonal variations, and the global morphology of the ionosphere, agree well with those in the empirical models and observations. The global electron content derived from the international GNSS service global ionospheric maps, the observed electron density profiles from the Poker Flat Incoherent Scatter Radar during 2007–2010, and foF2 observed by the global ionosonde network during 2002–2011 are used to validate the reanalysis method. All comparisons show that the reanalysis have smaller deviations and biases than the IRI‐2007 predictions. Especially after April 2006 when the six COSMIC satellites were launched, the reanalysis shows significant improvement over the IRI predictions. The obvious overestimation of the low‐latitude ionospheric F region densities by the IRI model during the 23/24 solar minimum is corrected well by the reanalysis. The potential application and improvements of the reanalysis are also discussed.

Comparison of ionosphere characteristic parameters obtained by ionosonde with IRI-2007 model over Southeast Asia

In this work, the foF2 and hmF2 parameters at the conjugate points near the magnetic equator of Southeast Asia are studied and compared with the International Reference Ionosphere (IRI) model. Three ionosondes are installed nearly along the magnetic meridian of 100°E; one at the magnetic equator, namely Chumphon (10.72°N, 99.37°E, dip angle 3.0°N), and the other two at the magnetic conjugate points, namely Chiang Mai (18.76°N, 98.93°E, dip angle 12.7°N) and Kototabang (0.2°S, 100.30°E, dip angle 10.1°S). The monthly hourly medians of the foF2 and hmF2 parameters are calculated and compared with the predictions obtained from the IRI-2007 model from January 2004 to February 2007. Our results show that: the variations of foF2 and hmF2 predicted by the IRI-2007 model generally show the similar feature to the observed data. Both parameters generally show better agreement with the IRI predictions during daytime than during nighttime. For foF2, most of the results show that the IRI model overestimates the observed foF2 at the magnetic equator (Chumphon), underestimates at the northern crest (Chiang Mai) and is close to the measured ones at the southern crest of the EIA (Kototabang). For hmF2, the predicted hmF2 values are close to the hmF2(M3000F2OBS) during daytime. During nighttime, the IRI model gives the underestimation at the magnetic equator and the overestimation at both EIA crests. The results are important for the future improvements of the IRI model for foF2 and hmF2 over Southeast Asia region.

A modeling study offoF2andhmF2parameters measured by the Arecibo incoherent scatter radar and comparison with IRI model predictions for solar cycles 21, 22, and 23

[1] This work presents the results of a local empirical model that describes the behavior of the ionospheric F2 region peak. The model was developed using nearly 25 years of incoherent scatter radar (ISR) measurements made at the Arecibo Observatory (AO) between 1985 and 2009. The model describes the variability of the F2 peak frequency (foF2) and F2 peak height (hmF2) as a function of local time, season, and solar activity for quiet-to-moderate geomagnetic activity conditions (Kp < 4+). Our results show that the solar activity control of hmF2 and foF2 over Arecibo can be better described by a new proxy of the solar flux (F107P), which is presented here. The variation of hmF2 parameter with F107P is virtually linear, and only a small saturation of the foF2 parameter is observed at the highest levels of solar flux. The winter anomaly and asymmetries in the variation of the modeled parameters between equinoxes were detected during the analyses and have been taken into account by the AO model. Comparisons of ISR data with international reference ionosphere (IRI) model predictions indicate that both CCIR and URSI modes overestimate foF2 during the daytime and underestimate it at night. As expected, this underestimation is not observed in the AO model. Our analyses also show that the hmF2 parameter predicted by the IRI modes shows a saturation point, which causes hmF2 to be underestimated at high solar activity. The underestimation increases with higher levels of solar activity. Finally, we also found that IRI predictions of the seasonal variability of foF2 and hmF2 over Arecibo can be improved by using a small correction that varies with solar activity and local time.

Improvement of HF coherent radar line-of-sight velocities by estimating the refractive index in the scattering volume using radar frequency shifting

[1] Measurements of ionospheric drift velocities using HF coherent scatter radars, such as SuperDARN, are generally underestimated because the refractive index in the scattering volume has not been taken into account. Refractive index values evaluated from electron density measurements, international reference ionosphere predictions, or elevation angle measurements have been applied to SuperDARN velocities in past studies. However, the SuperDARN velocities so obtained were, on average, statistically lower than velocities measured by other instruments. One possible explanation for this underestimation is that HF coherent scatter preferentially occurs in regions of the ionosphere where the scattering cross section is largest, and such regions are characterized by small-scale structures which have higher-than-average electron densities. This was not accounted for in past studies because the refractive index estimates used were from large scale and therefore smoothed estimates of electron density. In this paper, a new method of estimating the actual electron density (or plasma frequency) at the location of SuperDARN scatter (instead of the larger-scale background electron density) is presented. This method takes advantage of the frequency shifts which occur in normal SuperDARN operations. If it is assumed that, on average, the actual ionospheric drift velocity and plasma frequency are roughly constant before and after a shift in frequency, any change in measured velocity as SuperDARN changes frequency is due to a change in refractive index. An analysis of the change in the measured velocity resulting from each shift in frequency gives an experimentally based estimate of the electron density in the scattering volume. A statistical analysis of essentially all frequency shifts by SuperDARN and the estimated electron densities in the scattering volume has been performed. The resulting electron densities are appreciably higher than previous methods to estimate electron density predict. Application of this new method to velocity comparisons between SuperDARN and other instruments results in agreement between the HF radar and non-HF radar velocities for the first time. This new method allows for the first direct measurements of electron densities in the exact locations where the cross section for SuperDARN scatter maximizes.

Comparison of ionospheric F2 peak parameters foF2 and hmF2 with IRI2001 at Hainan

Monthly median values of foF2, hmF2 and M(3000)F2 parameters, with quarter-hourly time interval resolution for the diurnal variation, obtained with DPS4 digisonde at Hainan (19.5°N, 109.1°E; Geomagnetic coordinates: 178.95°E, 8.1°N) are used to investigate the low-latitude ionospheric variations and comparisons with the International Reference Ionosphere (IRI) model predictions. The data used for the present study covers the period from February 2002 to April 2007, which is characterized by a wide range of solar activity, ranging from high solar activity (2002) to low solar activity (2007). The results show that (1) Generally, IRI predictions follow well the diurnal and seasonal variation patterns of the experimental values of foF2, especially in the summer of 2002. However, there are systematic deviation between experimental values and IRI predictions with either CCIR or URSI coefficients. Generally IRI model greatly underestimate the values of foF2 from about noon to sunrise of next day, especially in the afternoon, and slightly overestimate them from sunrise to about noon. It seems that there are bigger deviations between IRI Model predictions and the experimental observations for the moderate solar activity. (2) Generally the IRI-predicted hmF2 values using CCIR M(3000)F2 option shows a poor agreement with the experimental results, but there is a relatively good agreement in summer at low solar activity. The deviation between the IRI-predicted hmF2 using CCIR M(3000)F2 and observed hmF2 is bigger from noon to sunset and around sunrise especially at high solar activity. The occurrence time of hmF2 peak (about 1200 LT) of the IRI model predictions is earlier than that of observations (around 1500 LT). The agreement between the IRI hmF2 obtained with the measured M(3000)F2 and the observed hmF2 is very good except that IRI overestimates slightly hmF2 in the daytime in summer at high solar activity and underestimates it in the nighttime with lower values near sunrise at low solar activity.

Comparing TOPEX TEC measurements with IRI predictions

TEC values obtained from TOPEX satellite were compared with the International Reference Ionosphere (IRI) 2001 model estimates. The present work also shows results of the IRI model with the option of a new topside electron density distribution (NeQuick model). TOPEX TEC measurements, which include years of high and middle to low solar activity (2000 and 2004), were analyzed by binning the region covered by the satellite (±66°) every five degrees of modip. In general, there is good agreement between IRI predictions and Topex measurements. Cases with large disagreements are observed at low and high latitudes during high solar activity. Comparing the model predictions using the default IRI2001 model and the NeQuick topside option show that the default IRI 2001 version represents the observed data in a more realistic way, but appears to be less reliable at high and low latitudes in some cases.

Evaluation of the IRI model using CHAMP observations in polar and equatorial regions

This study compares the electron density and temperature observed by the CHAMP satellite and those predicted by the International Reference Ionosphere (IRI) model. The comparison has revealed a general agreement between CHAMP and IRI in the seasonal variation of the electron density and temperature. However, in polar regions, the model tends to overestimate the electron density and underestimate the electron temperature. In addition, the CHAMP observes a weaker winter anomaly in the southern hemisphere than in the northern hemisphere, while the IRI predictions are similar in both hemispheres. In the equatorial region, the model describes the dayside equatorial ionization anomaly fairly well, but underestimates its depth in the post-sunset local time sector by about 50%. In polar regions, two prominent spatial structures observed by CHAMP are found to be missing in the IRI prediction. Namely, a band of low electron density along the nightside auroral region and a band of elevated electron temperature near the cusp. The high-Te band near the cusp exhibits clear seasonal variation, with the largest latitude and local time coverage in summer and the smallest in winter.

GPS-VTEC measurements and IRI predictions in the South American sector

Vertical total electron content (VTEC) measurements obtained with GPS satellite signals during 1999 are used to check the validityof the international reference ionosphere (IRI) to predict this ionospheric variable in the South American sector. Measurements obtained during June solstice and September equinox at nine stations are used. The considered latitude range extends from 18.4 to −64.7 and the longitude ranges from 281.3 to 297.7. The deviation between modelled and measured values was obtained. The results show that, in general, for both seasons IRI over-estimatesVTEC at night-time and sunrise hours. Better predictions were obtained in the latitude range (18.4°, −33.1°) for the period 8–22 LT. For high latitude stations, IRI does not give good predictions. Additional studies covering more stations and conditions and using ionosonde data will be useful to complete the IRI validation.

Equatorial F2-peak parameters in the IRI model

We have used measurements of an ionosonde station near the magnetic equator in Ouagadougou, Burkina Faso to evaluate the ability of the International Reference Ionosphere (IRI) model to correctly represent ionospheric F2 peak parameters in this region. The data represent conditions of high and low solar activity. Comparing the URSI and CCIR option for the F2 plasma frequency, foF2, we find that for low solar activity both options agree quite well with the ionosonde foF2 values and overall the CCIR maps show a slightly better fit. During high solar activity discrepancies are found during nighttime and the URSI maps providing the overall better fit. The measured F2 peak height values, hmF2, are compared on one hand with IRI predictions that are obtained based on the CCIR model for the propagation factor M(3000)F2 and on the other hand with the ionosonde-measured M(3000)F2 values. As expected using the measured values results in more accurate predictions. It is important to note that with the measured M(3000)F2 values IRI predicts the characteristics post-sunset that is seen in the measurements but not in the IRI predictions with the CCIR-M(3000)F2 model.

Combining GPS measurements and IRI model values for space weather specification

We will discuss various ways in which the International Reference Ionosphere (IRI) model and ionospheric data deduced from GPS measurements can be combined to improve ionospheric determinations. A number of research groups are analyzing GPS data products and providing global maps of vertical Total Electron Content (TEC) on a regular basis. IRI predictions can guide the interpolation of regional TEC estimations, computed from GPS data, to obtain global TEC maps. GPS measurements, on the other hand, can be used to update the IRI monthly averages to actual conditions. This can be done by using the GPS-derived TEC maps or by using the actual GPS measurements of the electron content along the signal path from satellite to ground receiver. We will discuss the updating results using the actual GPS measurements.

Total electron content in Australia corrected for receiver/satellite offset bias and compared with IRI and PIM predictions

Abstract A procedure is presented for removing receiver and satellite (transmitter) offset bias from total electron content (TEC) data obtained using Global Positioning System (GPS) satellites from 1991 to 1998. Samples of TEC data from 1991 to 1995 for Australia have been corrected for bias. Corrected TEC values show substantial agreement with model predictions from the International Reference Ionosphere (IRI) and the Paramaterized Ionospheric Model (PIM).

Predictions of HF communication MUF in the region of the South China Sea

In this paper, an artificial-neural-network (ANN) method is employed to predict the HF-communication MUF (maximum useable frequency) in the region of the South China Sea. Historical observation data of f/sub 0/F/sub 2/ from six ionospheric observatories around the South China Sea are used to train the network. The comparison of ANN predictions and International Reference Ionosphere (IRI) models with the observations shows that the ANN predictions are closer to the observed results than are the IRI predictions.

Improved IRI predictions for the GEOSAT time period

Single-frequency altimeter measurements require an estimate of the signal retardation effect in the ionosphere. Ionospheric models like the International Reference Ionosphere (IRI) are being used to obtain these correction factors. Incorporation of worldwide ionosonde measurements is discussed as a way towards more accurate predictions. Using measurements of the F peak critical frequency (foF2) from more than 70 ionosondes we have derived (hour and month dependent) regional and global ionospheric-effective solar indices. Replacing the solar index in IRI with these new indices we could improve the IRI estimates for the GEOSAT time period (1986–1989) by a few percent.

Comparison of San Diego observations (March 1992) with international reference ionosphere predictions

During the month of March 1992, the new ionosonde at NRaD in San Diego, CA U.S.A., was used to obtain ionograms at five minute intervals. The new ionosonde is similar to the two channel HF sounder developed at NOAA in Boulder, CO U.S.A. in the late 1970s /i/, however, the NRaD machine is the first fully four channel version of that machine. The four channel architecture permits faster and simpler operation since antenna multiplexing used for the two channel machine is no longer required.