- Email: [email protected]
FORMOSAT-3, Irkutsk incoherent scatter radar, Irkutsk Digisonde and IRI model electron density vertical profiles
Accepted Manuscript Comparative study of COSMIC/FORMOSAT-3, Irkutsk incoherent scatter radar, Irkutsk Digisonde and IRI model electron density vertical profiles K.G. Ratovsky, A.V. Dmitriev, A.V. Suvorova, A.A. Shcherbakov, S.S. Alsatkin, A.V. Oinats PII: DOI: Reference:
S0273-1177(16)30748-7 http://dx.doi.org/10.1016/j.asr.2016.12.026 JASR 13025
To appear in:
Advances in Space Research
Received Date: Revised Date: Accepted Date:
13 May 2016 13 December 2016 17 December 2016
Please cite this article as: Ratovsky, K.G., Dmitriev, A.V., Suvorova, A.V., Shcherbakov, A.A., Alsatkin, S.S., Oinats, A.V., Comparative study of COSMIC/FORMOSAT-3, Irkutsk incoherent scatter radar, Irkutsk Digisonde and IRI model electron density vertical profiles, Advances in Space Research (2016), doi: http://dx.doi.org/10.1016/ j.asr.2016.12.026
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Comparative study of COSMIC/FORMOSAT-3, Irkutsk incoherent scatter radar, Irkutsk Digisonde and IRI model electron density vertical profiles
K.G. Ratovskya, A.V. Dmitrievb, A.V. Suvorovac, A.A. Shcherbakovd, S.S. Alsatkine, A.V. Oinatsf a
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, tel: +7 3952 564539, fax: +7 3952 425557, [email protected] b
National Central University, Jhongli District, Taoyuan City 32001, Taiwan; Scobeltsyn Institute of
Nuclear Physics, Moscow State University, Moscow, 119991, Russia, [email protected] c
National Central University, Jhongli District, Taoyuan City 32001, Taiwan; Scobeltsyn Institute of
Nuclear Physics, Moscow State University, Moscow, 119991, Russia, [email protected] d
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, [email protected] e
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, [email protected] f
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, [email protected]
Abstract
The long-duration continuous Irkutsk incoherent scatter radar observations allowed us to collect 337 electron density vertical profiles obtained almost simultaneously with the radar and the COSMIC in the radar vicinity. The COSMIC electron density profiles were compared with those from the radar, Digisonde, and the IRI model. The comparison included 4 seasons and 2 solar activity levels (low and moderate).The number of simultaneous cases was ~10 times more than in the previous incoherent scatter radar comparisons. In the case of the bottomside characteristics (peak density and bottomside electron content), the deviations between the COSMIC and the ground-based facilities data may be interpreted as the COSMIC measurement errors without significant systematic biases and with root-mean-square values that are ~1.4-1.6 times smaller those that from the IRI model prediction. In the case of the topside characteristics (topside electron content and ionospheric electron content), the IRI model overestimates
the COSMIC data by 0.6-0.8 tecu on average, and the COSMIC overestimates the Irkutsk incoherent scatter radar data by 1.0-1.1 tecu on average. The percentage differences between the radar and COSMIC in the topside electron content can reach 80%. In terms of the root-mean-square deviation, the COSMIC and the radar agree better than each of them agrees with the IRI model.
Keywords: topside ionosphere, COSMIC/FORMOSAT-3, incoherent scatter radar, Digisonde.
1. Introduction
The Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC or FORMOSAT-3) (Schreiner et al., 2007) provides vertical electron density profiles globally and thus represents a unique system for studying the ionospheric electron density on a global scale and using these data for different application and empirical model assimilation. The validation of the COSMIC profiles through comparison with ground-based radio-sounding facilities data has been performed in many papers (e.g. Lei et al., 2007; Stolle et al., 2004; Wu et al., 2009; Chu et al., 2010; Krankowski et al., 2011; Cherniak and Zakharenkova, 2014; Hu et al., 2014; Mikhailov et al., 2014). The radio facilities considered in the above-cited papers are ionosondes (Wu et al., 2009; Chu et al., 2010; Krankowski et al., 2011; Hu et al., 2014) and/or incoherent scatter radars (Lei et al., 2007; Stolle et al., 2004; Cherniak and Zakharenkova, 2014; Mikhailov et al., 2014). In case of ionosondes, the comparisons were made for the bottomside characteristics (F2 peak density (NmF2) and the peak height (hmF2)), while the comparisons with incoherent scatter radars gave an additional opportunity to validate of the topside electron density. The main objective of this paper is a comparative analysis of the electron density vertical profiles obtained with the COSMIC satellite and the Irkutsk incoherent scatter radar (IISR) (52.9N, 103.3E) (Potekhin et al., 2008) during the periods of long-duration continuous radar observations presented in Table 1. For these periods we collected 337 COSMIC measurements that have tangent points at hmF2 within ±5º by latitude and longitude of the IISR location and the time within ±10 min of the IISR measurements (Table 1). Additionally, we compare the COSMIC data with those from the Irkutsk Digisonde (IDPS-4) (52.3N, 104.3E) (Reinisch et al., 1997) located ~100 km away from the IISR, and from the International Reference Ionosphere model (IRI) (Bilitza et al., 2014). For the IRI calculations we used two options of the topside ionosphere: IRI-Nequick and Corrected IRI 01. In the 170–600 km range both options give close results, and further we use the Corrected IRI 01 option prediction shortly denoted as IRI.
The distinctive features of this paper are the following: (1) use of the long-duration continuous IISR observations that cover various seasons and solar activity levels providing more simultaneous measurements than the previous incoherent scatter radar comparisons; and (2) special emphasis to the comparison of the topside electron density. The methods of electron density profile construction and the characteristics selected for the comparison are described in the next section.
2. Methods of electron density profile construction
Each COSMIC microsatellite has a GPS Occultation Experiment payload to operate the ionospheric radio occultation (RO) measurements. The methods of electron density profile construction from RO measurements is described by Tsai et al. (2001) and Lei et al. (2007). The assumption of spherical symmetry of electron density is the most significant source of error in the retrieval of the electron density profiles when we interpret them as actual vertical profiles (Lei et al., 2007). The geographical location of the tangent points path at the top and at the bottom of a profile may differ by several hundred kilometers, and due to the presence of latitudinal and longitudinal gradients in the ionosphere, the COSMIC profile represents spatial averaged information about electron density (Krankowski et al., 2011). All ionogram data of the IDPS-4were manually scaled using an interactive ionogram scaling software, SAO Explorer (Reinisch et al., 2004; Khmyrov et al., 2008). The electron density profiles were inverted from all suitable ionogram traces using the NHPC method (Reinisch and Huang, 1983). The electron density above hmF2 was estimated using the Reinisch and Huang (2001) technique based on the profile extrapolation by the Chapman function with the scale height determined at hmF2. The IISR radiates and receives only one linear polarization, and thus the received power profile P as function of height z is modulated by the Faraday rotation. For near vertical radar beam, the power profile P can be described by the following equations:
P z A
Nez 2 cos 2 ( z ) B, ( z ) 2 f p2 z f H zcos( z)dz , 2 ( z / cos( )) cf cos( ) 0 z
(1)
where A and B are multiplicative and additive constants, respectively; Ne(z) is the electron density profile; β is the radar beam zenith angle (10º in our case); φ is the Faraday rotation phase; fp is the electron plasma frequency ( f p2[MHz] Ne[105 cm3 ] / 0.124 ); fH is the electron gyrofrequency; θ is the angle between the radar beam and the Earth's magnetic field line (15º in our case); f is the operating frequency (155.5 MHz in our case); and c is the light velocity. From one hand, the Faraday modulation does not enable to calculate Ne(z) directly from P(z), but, on the other hand, it gives an opportunity to obtain Ne(z) without an ionosonde calibration.
In this study we use the approximation of Ne(z) by the Chapman-like function:
Ne( z ) NmF 2 exp(1 x exp( x)), where x ( z hmF 2) / H T at z hmF 2, and
(2)
x ( z hmF 2) / H B at z hmF 2,
where HT and HB are the topside and bottomside scale heights, respectively. The Ne(z) profile characteristics (NmF2, hmF2, HT , and HB) as well as A and B constants were derived by least squares fit of the model P(z) (calculated with Eq. (1) and (2)) to the received power profile. Figure 1 shows the received and model power profiles as well as the obtained electron density profile. Note, that this method does not give a high height resolution, but allows stable automatic processing of the IISR received power profiles including cases of low signal-to-noise ratio. For the comparison we selected the following electron density characteristics: the peak density (NmF2), the bottomside (170-300 km) electron content (BTEC), the topside (300-600 km) electron content (TTEC), and the electron content in 170-600 km height range that we denote as an ionospheric electron content (IEC). Initially, we determined the bottomside electron content as the electron content in the 170 km -hmF2 height range. Further, we found that the contribution of the difference in hmF2 to the difference in BTEC is much more than the contribution of the difference in the bottomside electron density. For this reason, we decided to use the constant boundary (300 km) between the bottomside and topside ionosphere. When hmF2 is less than 300 km, IDPS-4 may give errors of the profile extrapolation above hmF2. Figure 2 shows the IDPS4-IISR comparison of BTEC (scatterplot of all the data and year-to-year of the mean (ΔBTEC) and root-mean-square (σBTEC) deviations). It is seen that σBTEC is noticeably larger than ΔBTEC and varies from 0.25 to 0.45 tecu. As will be shown below, the IDPS4-IISR σBTEC is ~ 2 times smaller than that for COSMIC-IISR and COSMIC-IDPS-4 comparisons, and thus it is reasonable to consider both COSMIC-IISR and COSMIC-IDPS-4 comparisons.
3. Comparison and Discussion
Our comparison was made in terms of the mean deviation and the root-mean-square deviation between the COSMIC data and those from IISR, IDPS-4, and IRI. The mean deviation (Δ) characterizes the degree of systematic underestimation or overestimation, while the root-mean-square deviation (σ) is a measure of random differences between the data. Figure 3 shows the comparison of NmF2 by the scatterplots of COSMIC vs. IISR, COSMIC vs. IDPS-4, IRI vs. IISR, and IRI vs. IDPS-4 for low (2007-2009) and moderate (2011-2013) solar activity,
as well as, in terms of Δ and σ as functions of year for COSMIC - IISR, COSMIC - IDPS-4, IRI - IISR, and IRI - IDPS-4 deviations. The COSMIC - IISR comparison of NmF2 indicates that ΔNmF2 is mainly much less than σNmF2 that changes from 0.25-0.32·105 cm-3 at low solar activity (2007-2009) up to 0.71.0·105 cm-3 at the highest solar activity for the considered period (2011-2013) with the average σNmF2 ~ 0.5·105 cm-3. The COSMIC - IDPS-4 comparison of NmF2 shows a similar pattern. These results agree closely with those reported by Krankowski et al. (2011) (σNmF2 ~ 0.2·105 cm-3 for the European ionosondes at low solar activity); and Hu et al. (2014) (σNmF2 ~ 0.6·105 cm-3 for the Mohe ionosonde (53.5N, 123.3) at moderate solar activity); and Cherniak and Zakharenkova (2014) (σNmF2 ~ 0.3·105 cm-3 for the Kharkov incoherent scatter radar (49.6, 36.3E) at low solar activity). In all these comparisons ΔNmF2 was noticeably smaller than σNmF2. The IRI - IISR comparison give the average σNmF2 ~ 0.8·105 cm-3 and the average underestimation ~ 0.3·105 cm-3; and thus we may conclude that the IISR agrees better with COSMIC (σNmF2 ~ 0.5·105 cm-3 under small ΔNmF2) than with the IRI model (σNmF2 ~ 0.8·105 cm-3 under ΔNmF2 ~ 0.3·105 cm-3). As mentioned above, the COSMIC does not give an actual vertical profile due to the presence of latitudinal and longitudinal gradients. Assuming that NmF2 from the IISR and IDPS-4 are much closer to actual values than those from the COSMIC, we may interpret the COSMIC - IISR and COSMIC - IDPS-4 deviations as the COSMIC measurement errors without significant systematic biases and with root-meansquare value that is ~1.6 times smaller than that from the IRI prediction. The rise in σNmF2 with increasing solar activity may be explained by the corresponding rise in latitudinal and longitudinal gradients of the electron density. We can not say that our comparison of NmF2 completely agrees with all such comparisons at midlatitudes. Hu et al. (2014) and Mikhailov et al. (2014) reported that the COSMIC overestimated NmF2 obtained with ionosondes or incoherent scatter radar by the values comparable or even larger than the root-mean-square deviations. All these comparisons relate to the sites located at latitudes near 40N or lower. Hu et al. (2014), considering that the overestimation increases with lowering the latitude, assumed that overestimation may be associated with the effects of the equatorial ionization anomaly (EIA). Yue et al. (2010) simulation r showed that the COSMIC underestimated the true electron density in the region surrounding the EIA crest and overestimated it near the magnetic equator and in the north and south of the EIA crests. Figure 4 shows the same as Figure 3 but for BTEC. The results for the BTEC comparison are generally close to results for the NmF2comparison. The difference between the COSMIC - IISR and COSMIC - IDPS-4 comparisons lies in the fact that σBTEC slightly larger in the COSMIC - IDPS-4 case, and there is a small systematical difference (~0.2 tecu) for the COSMIC - IISR comparisons unlike the
COSMIC - IDPS-4 case. Such a difference may be explained by errors of the profile extrapolation above hmF2 (when hmF2 is noticeably less than 300 km). Also note that σBTEC for the COSMIC - IISR and COSMIC - IDPS-4 comparisons is ~1.4 times smaller than for the IISR - IRI comparison, and this ratio is slightly less than that for σNmF2. Figure 5 shows the comparison of TTEC by the scatterplots of COSMIC vs. IISR and IRI vs. IISR for low (2007-2009) and moderate (2011-2013) solar activity, as well as, in terms of Δ and σ as functions of year for COSMIC – IISR and IRI - IISR deviations. The TTEC comparison differs significantly from the comparison of the bottomside characteristics (NmF2 and BTEC). The main difference lies in the fact that the COSMIC systematically overestimates the IISR TTEC by ~1 tecu on average, and this overestimation is close to σTTEC. The dependence of the COSMIC - IISR ΔTTEC on solar activity is not very pronounced. The IRI model overestimates the COSMIC TTEC (by ~0.8 tecu on average), which in turn is ~1 tecu larger than the IISR TTEC. The smallest average root-mean-square deviation is seen for the COSMIC - IISR comparisons (~1 tecu), and the largest one is seen for the IRI - IISR comparisons (~1.5 tecu) with intermediate σ for the COSMIC - IRI comparison. (~1.2 tecu). Thus in terms of σ the COSMIC and IISR agree better than each of them agrees with the IRI. None of the previous incoherent scatter radar studies have reported that the COSMIC overestimates the topside electron density with a better agreement in the bottomside. Stolle et al. (2004) did not report about any differences between the NmF2 and 150-420 km electron content statistics under the comparison of the radio occultation and EISCAT measurements. Cherniak and Zakharenkova (2014) did not provide the quantitative estimation of the deviations between the COSMIC and the Kharkov incoherent scatter radar (49.6, 36.3E) in the topside and made a qualitative conclusion that the topside profile shape showed rather good agreement between the COSMIC and the radar. Lei et al. (2007) concluded that in most cases the shapes of the Millstone Hill incoherent scatter radar (42.6N, 71.5W) profiles are captured by the COSMIC data. Only Mikhailov et al. (2014), using 35 simultaneous measurements with the COSMIC and the Millstone Hill incoherent scatter radar at low solar activity, made the separate comparison for the topside and bottom side parts. They concluded that in the majority of cases the COSMIC topside profile coincided fairly well with the radar observations, while the coincidence in the bottomside profile was not good for the 40% of the analyzed cases. Thus, they demonstrated a result that is opposite to our findings, and the reason for this difference is not clear. Possibly, the comparison results strongly depend on the ground-based facility location, at least the mentioned above ionosonde comparisons show that this may be the case. Figure 6 shows the same as Figure 5 but for IEC. The comparison of IEC = BTEC + TTEC summarizes the features of the BTEC and TTEC comparisons. The IRI model overestimates the COSMIC
IEC by ~0.6 tecu on average, and the COSMIC overestimates the IISR IEC by ~1.1 tecu on average and this overestimation does not have a pronounced dependence on solar activity. In terms of the root-meansquare deviation, the COSMIC and IISR agree better (σ ~1.3 tecu) than each of them agrees with the IRI (σ ~1.8 and 2 tecu). Below we have considered the absolute differences between the COSMIC and IISR data, whereas the percentage differences are also of interest due to their strong dependency on the background values. Figure 7 shows the COSMIC - IISR comparison in terms of the mean percentage deviation Δ(%) and the root-mean-square percentage deviation σ(%) as functions of year for NmF2, BTEC, TTEC and IEC. For the case of the bottomside characteristics (NmF2 and BTEC) the results for the percentage differences are similar to those for the absolute differences: Δ(%) is noticeably smaller than σ(%), and σ(%) rises with increasing solar activity from 8-10 to 20-25%. In the case of TTEC we see significant (from 25 to 80%) and nonmonotonic variations of σ(%) from year to year, while the absolute σTTEC values do not show such behavior. This means that that the absolute differences in TTEC depends weakly on background TTEC values, and we obtain larger percentage differences for lower background values. The case of IEC is an intermediate case between the bottomside characteristics and TTEC: the year to year variations are nonmonotonic, but the range of σ (15-40%) is not as large as in the TTEC case.
4. Conclusion
The long-duration continuous Irkutsk incoherent scatter radar observations allowed us to collect 337 electron density vertical profiles obtained almost simultaneously with the radar and the COSMIC in the radar vicinity. The comparison included 4 seasons and 2 solar activity levels (low and moderate), and the number of simultaneous cases was ~10 times more than in the previous incoherent scatter radar comparisons. The validation the COSMIC profiles through the comparison with the Irkutsk incoherent scatter radar and Irkutsk Digisonde gave the following results. In the case of the bottomside characteristics (peak density and bottomside electron content), the deviations between the COSMIC and the ground-based facilities data may be interpreted as the COSMIC measurement errors without significant systematic biases and with root-mean-square values that are ~1.41.6 times smaller those that from the IRI model prediction. The rise in the root-mean-square errors with increasing solar activity may be explained by the corresponding rise in latitudinal and longitudinal gradients of the electron density. Such a comparison pattern agree closely with the results obtained for the groundbased facilities located at latitudes near 50N, while the comparisons at latitudes near 40N or lower show
that the COSMIC overestimates NmF2 from the ground-based facilities by the values comparable or even larger than the root-mean-square deviations. In the case of the topside characteristics (topside electron content and ionospheric electron content), the IRI model overestimates the COSMIC data by 0.6-0.8 tecu on average, and the COSMIC overestimates the Irkutsk incoherent scatter radar data by 1.0-1.1 tecu on average. The percentage differences between the radar and COSMIC in the topside electron content can reach 80%. In terms of the root-mean-square deviation, the COSMIC and the radar agree better than each of them agrees with the IRI model. None of the previous incoherent scatter radar studies have reported that the COSMIC overestimates the topside electron density with a better agreement in the bottomside. Moreover, the validation the COSMIC with the Millstone Hill incoherent scatter radar demonstrates results that are opposite to our findings. Possibly, the comparison results strongly depend on the ground-based facility location, as it is seen from the comparisons of NmF2.
Acknowledgments
The reported study was funded by RFBR and NSC according to the Russian-Taiwanese joint research project 14-05-92002 HHC_a / NSC103-2923-M-006-002-MY3 and partially supported by RF President Grant of Public Support for RF Leading Scientific Schools (NSh-6894.2016.5).
References
Bilitza, D., Altadill, D., Zhang, Y., Mertens, C., Truhlik, V., Richards, P., McKinnell, L.-A., Reinisch, B., 2014. The International Reference Ionosphere 2012 - a model of international collaboration. J. Space Weather Space Clim. 4 (A07), 1-12. Cherniak, Iu.V., Zakharenkova, I.E., 2014. Validation of FORMOSAT-3/COSMIC radio occultation electron density profiles by incoherent scatter radar data. Adv. Space Res. 53 (9), 1304-1312. Chu, Y.-H., Su, C.-L., Ko, H.-T., 2010. A global survey of COSMIC ionospheric peak electron density and its height: A comparison with ground-based ionosonde measurements. Adv. Space Res., 46 (4), 431– 439. Hu, L., Ning, B., Liu, L., Zhao, B., Li, G., Wu, B., Huang, Z., Hao, X., Chang, S., Wu, Z., 2014. Validation of COSMIC ionospheric peak parameters by the measurements of an ionosonde chain in China. Ann. Geophys., 32(10), 1311-1319.
Khmyrov, G.M., Galkin, I.A., Kozlov, A.V., Reinisch, B.W., McElroy, J., Dozois C., 2008. Exploring digisonde ionogram data with SAO-X and DIDBase. Radio Sounding and Plasma Physics, AIP Conf. Proc. 974, 175-185. Krankowski, A., Zakharenkova, I.E., Krypiak-Gregorczyk, A., Shagimuratov, I.I., Wielgosz, P., 2011. Ionospheric electron density observed by FORMOSAT-3/COSMIC over the European region and validated by ionosonde data. J. Geod. 85 (12), 949–964. http://dx.doi.org/10.1007/s00190-011-0481-z. Lei, J., Syndergaard, S., Burns, A.G., et al., 2007. Comparison of COSMIC ionospheric measurements with ground-based observations and model predictions: preliminary results. J. Geophys. Res. 112, A07308. http://dx.doi.org/10.1029/2006JA012240. Mikhailov, A.V., Belehaki, A., Perrone, L., Zolesi, B., Tsagouri I., 2014. On the possible use of radio occultation middle latitude electron density profiles to retrieve thermospheric parameters. J. Space Weather Space Clim. 4, A12. http://dx.doi.org/10.1051/swsc/2014009. Potekhin, A.P., Medvedev, A.V., Zavorin, A.V., Kushnarev, D.S., Lebedev, V.P., Shpynev, B.G., 2008. Development of diagnostic capabilities of the Irkutsk incoherent scattering radar. Cosmic Research. 46 (4), 347-353. Reinisch, B.W., Huang, X., 1983. Automatic Calculation of Electron Density Profiles from Digital Ionograms, 3, Processing of Bottomside Ionograms. Radio Sci. 18(3), 477-492. Reinisch, B.W., Haines, D.M., Bibl, K., Galkin, I., Huang, X., Kitrosser, D.F., Sales, G.S., Scali, J.L., 1997. Ionospheric sounding support of OTH radar. Radio Sci. 32 (4), 1681-1694. Reinisch, B.W., Huang, X., 2001. Vertical electron content from ionograms in real time. Radio Sci. 36 (2), 335-342. Reinisch, B.W., Galkin I. A., Khmyrov G., Kozlov, A., Kitrosser, D.F., 2004. Automated collection and dissemination of ionospheric data from the digisonde network. Adv. Radio Sci. 2, 241-247. Schreiner, W., Rocken C., Sokolovsky, S., Syndergaard, S., Hunt D., 2007. Estimates of the precision of GPS radio occultations from the COSMIC/FORMOSAT-3 mission. Geophys. Res. Lett., 34, L04808. http://dx.doi.org/10.1029/2006GL027557. Stolle, C., Jakowski, N., Schlegel, K., Rietveld, M., 2004. Comparison of high latitude electron density profiles obtained with the GPS radio occultation technique and EISCAT measurements. Ann. Geophys. 22 (6), 2015–2022. Tsai, L.C., Tsai, W.H., Schreiner, W.S., Berkey, F.T., Liu, J.Y., 2001. Comparisons of GPS/MET retrieved ionospheric electron density and ground based ionosonde data. Earth Planets Space 53, 193–205. Wu, X., Hu, X., Gong, X., Zhang, X., Wang, X., 2009. Analysis of inversion errors of ionospheric radio occultation. GPS Solutions. 13 (3), 231–239. http://dx.doi.org/10.1007/s10291-008-0116-x.
Yue, X., Schreiner, W. S., Lei, J., Sokolovskiy, S. V., Rocken, C., Hunt, D. C., and Kuo, Y.-H., 2010. Error analysis of Abel retrieved electron density profiles from radio occultation measurements. Ann. Geophys. 28 (1), 217-222.
Table captions Table.1 Periods of long-duration continuous Irkutsk incoherent scatter radar observations.
Figure captions
Fig.1 Received power profile of Irkutsk incoherent scatter radar (left, black), obtained electron density profile (right, grey), and model power profile calculated with Eq. (1) and (2) (left, grey). Fig.2 IDPS4-IISR comparison of BTEC (scatterplot of all the data and year-to-year of the mean (ΔBTEC) and root-mean-square (σBTEC) deviations). Fig.3 Comparison of NmF2 by the scatterplots of COSMIC vs. IISR, COSMIC vs. IDPS-4, IRI vs. IISR, and IRI vs. IDPS-4 (from left to right) for low (top panel) and moderate (middle panel) solar activity, and in terms of Δ (grey) and σ (black) as functions of year (bottom panel) for COSMIC - IISR, COSMIC IDPS-4, IRI - IISR, and IRI - IDPS-4 deviations (from left to right). Fig.4 The same as Figure 3 but for BTEC. Fig.5 Comparison of TTEC by the scatterplots of COSMIC vs. IISR and IRI vs. IISR (from left to right) for low (top panel) and moderate (middle panel) solar activity, and in terms of Δ (grey) and σ (black) as functions of year (bottom panel) for COSMIC – IISR and IRI - IISR deviations (from left to right). Fig.6 The same as Figure 5 but for IEC. Fig. 7 COSMIC - IISR comparison in terms of mean percentage deviation Δ(%) and root-mean-square percentage deviation σ(%) as functions of year for NmF2, BTEC, TTEC and IEC
650 600
600
550
550
500
500
Height (km)
Height (km)
650
Apr 20, 2011 07:12 UT (14:12 LT)
450 400 350
450 400 350
300
300
250
250
200
200
0.5 Fig. 1
0.6
0.7
0.8
0.9
Power profile (rel. un.)
1
0
2
4
6
8
Electron density profile
10
12
(105cm-3)
2
BTEC, BTEC (tecu)
IDPS-4 BTEC (tecu)
12 10 8 6 4 2 0
0
2
4
6
8
IISR BTEC (tecu) Fig. 2
10
12
1.5 1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
Low Solar activity
2
4
IISR NmF2
6
14 12 10 8 6 4 2 4
6
8
10 12 14 16
IISR NmF2 (105cm-3)
NmF2 (105cm-3), NmF2
NmF2 (105cm-3), NmF2
1 0.5 0 -0.5 2010
Year
Fig. 3
2012
2
0
8
0
2014
2
4
IISR NmF2
6
(105cm-3)
4
2
0
8
Moderate Solar activity
16
14
14
12 10 8 6 4 2 0
2
4
6
8
12 10 8 6 4 2 0
10 12 14 16
IDPS-4 NmF2 (105cm-3)
1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
4
6
8
10 12 14 16
IISR NmF2 (105cm-3)
1.5 1 0.5 0 -0.5 2008
2010
Year
2012
6
8
8 6 4 2 0
2
4
6
8
10 12 14 16
IDPS-4 NmF2 (105cm-3)
IRI - IDPS-4
2
-1 2006
4
10
IRI - IISR
2 1.5
2
2
IDPS-4 NmF2 (105cm-3)
12
0 0
0
Moderate Solar activity
14
Cosmic - IDPS-4
1.5
2008
6
(105cm-3)
4
6
16
Cosmic - IISR 2
-1 2006
4
6
16
0 2
2
Moderate Solar activity COSMIC NmF2 (105cm-3)
COSMIC NmF2 (105cm-3)
16
0
0
IDPS-4 NmF2
Moderate Solar activity
0
0
8
(105cm-3)
IRI NmF2 (105cm-3)
0
2
NmF2 (105cm-3), NmF2
0
4
IRI NmF2 (105cm-3)
2
6
IRI NmF2 (105cm-3)
4
8
8
NmF2 (105cm-3), NmF2
6
Low Solar activity
Low Solar activity
8
IRI NmF2 (105cm-3)
COSMIC NmF2 (105cm-3)
COSMIC NmF2 (105cm-3)
Low Solar activity 8
2014
2 1.5 1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
Low Solar activity
4
6
0
8
IISR BTEC (tecu)
10 8 6 4 2 4
6
8
10
12
14
IISR BTEC (tecu)
1 0.5 0 -0.5 2010
Year
2012
2
2014
4
6
4
2
0
8
0
14
12
12
12
8 6 4
10 8 6 4
2
2
0
0 0
2
4
6
8
10 12 14 16
1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
6
8
10 8 6 4 2
0
2
4
6
8
10
12
0
14
IISR BTEC (tecu)
0
4
6
8
10
12
14
IRI - IDPS-4
IRI - IISR 2
1.5 1 0.5 0 -0.5 -1 2006
2
IDPS-4 BTEC (tecu)
2
1.5
4
Moderate Solar activity
14
10
2
IDPS-4 BTEC (tecu)
Moderate Solar activity
Cosmic - IDPS-4
1.5
2008
0
IISR BTEC (tecu)
2
BTEC, BTEC (tecu)
BTEC, BTEC (tecu)
8
6
14
Cosmic - IISR
Fig. 4
6
IDPS-4 BTEC (tecu)
2
-1 2006
4
IRI BTEC (tecu)
COSMIC BTEC (tecu)
COSMIC BTEC (tecu)
12
2
2
Moderate Solar activity
14
0
2
IDPS-4 BTEC (tecu)
Moderate Solar activity
0
4
0 0
IRI BTEC (tecu)
2
6
IRI BTEC (tecu)
2
4
BTEC, BTEC (tecu)
0
6
8
BTEC, BTEC (tecu)
2
IRI BTEC (tecu)
COSMIC BTEC (tecu)
COSMIC BTEC (tecu)
4
Low Solar activity
8
8
6
0
Low Solar activity
Low Solar activity
8
2008
2010
Year
2012
2014
1.5 1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
Low Solar activity
Low Solar activity 8
IRI TTEC (tecu)
COSMIC TTEC (tecu)
8
6
4
2
0
0
2
4
6
6
4
2
0
8
0
2
IISR TTEC (tecu)
16
16
14
14
12 10 8 6 4 2
10 8 6 4 2
0
2
4
6
8
0
10 12 14 16
0
4
6
8
10 12 14 16
IRI - IISR 3
TTEC, TTEC (tecu)
3 2.5 2 1.5 1 0.5 0 -0.5 2006
2
IISR TTEC (tecu)
Cosmic - IISR TTEC, TTEC (tecu)
8
12
IISR TTEC (tecu)
2008
2010
Year Fig. 5
6
Moderate Solar activity IRI TTEC (tecu)
COSMIC TTEC (tecu)
Moderate Solar activity
0
4
IISR TTEC (tecu)
2012
2014
2.5 2 1.5 1 0.5 0 -0.5 2006
2008
2010
Year
2012
2014
Low Solar activity
12
12
10
10
IRI IEC (tecu)
COSMIC IEC (tecu)
Low Solar activity
8 6 4
6 4 2
2 0
8
0
2
4
6
8
10
0
12
0
2
28
24
24
20 16 12 8 4
10
12
16 12 8 4
0
4
8
12
16
20
24
0
28
0
8
12
16
20
24
28
IRI - IISR 4
IEC, IEC (tecu)
4 3 2 1 0 2006
4
IISR IEC (tecu)
Cosmic - IISR IEC, IEC (tecu)
8
20
IISR IEC (tecu)
3 2 1 0
2008
2010
Year Fig. 6
6
Moderate Solar activity
28
IRI IEC (tecu)
COSMIC IEC (tecu)
Moderate Solar activity
0
4
IISR IEC (tecu)
IISR IEC (tecu)
2012
2014
2006
2008
2010
Year
2012
2014
BTEC
NmF2
40 20 0 2006
2008
2010
Year
Fig. 7
2012
2014
80 60 40 20 0 2006
2008
2010
Year
2012
2014
IEC (%), IEC (%)
60
100
100
TTEC (%), TTEC (%)
BTEC (%), BTEC (%)
NmF2 (%), NmF2 (%)
80
IEC
TTEC
100
100
80 60 40 20 0 2006
2008
2010
Year
2012
2014
80 60 40 20 0 2006
2008
2010
Year
2012
2014
Observation periods JUN 05 - JUN 19, 2007 AUG 25 – SEP 28, 2008 APR 01 – APR 18, 2009 JAN 12 - FEB 28, 2010 JAN 16 - FEB 16, 2011 APR 12 - APR 21, 2011 JAN 18 - FEB 05, 2012 APR 06 – APR 22, 2012 JAN 01 – JAN 21, 2013 JUN 21 – JUN 30, 2013 DEC 25 - DEC 31, 2013 In total Table 1
Number of Days 15 35 16 48 32 10 8 12 19 9 4 208
Number of Simultaneous cases 46 44 32 62 28 6 13 23 49 24 10 337
Yearly mean F10.7 (s.f.u.) 73 69 71 80 114 120 123
S0273-1177(16)30748-7 http://dx.doi.org/10.1016/j.asr.2016.12.026 JASR 13025
To appear in:
Advances in Space Research
Received Date: Revised Date: Accepted Date:
13 May 2016 13 December 2016 17 December 2016
Please cite this article as: Ratovsky, K.G., Dmitriev, A.V., Suvorova, A.V., Shcherbakov, A.A., Alsatkin, S.S., Oinats, A.V., Comparative study of COSMIC/FORMOSAT-3, Irkutsk incoherent scatter radar, Irkutsk Digisonde and IRI model electron density vertical profiles, Advances in Space Research (2016), doi: http://dx.doi.org/10.1016/ j.asr.2016.12.026
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Comparative study of COSMIC/FORMOSAT-3, Irkutsk incoherent scatter radar, Irkutsk Digisonde and IRI model electron density vertical profiles
K.G. Ratovskya, A.V. Dmitrievb, A.V. Suvorovac, A.A. Shcherbakovd, S.S. Alsatkine, A.V. Oinatsf a
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, tel: +7 3952 564539, fax: +7 3952 425557, [email protected] b
National Central University, Jhongli District, Taoyuan City 32001, Taiwan; Scobeltsyn Institute of
Nuclear Physics, Moscow State University, Moscow, 119991, Russia, [email protected] c
National Central University, Jhongli District, Taoyuan City 32001, Taiwan; Scobeltsyn Institute of
Nuclear Physics, Moscow State University, Moscow, 119991, Russia, [email protected] d
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, [email protected] e
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, [email protected] f
Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, Irkutsk,
664033, Russia, [email protected]
Abstract
The long-duration continuous Irkutsk incoherent scatter radar observations allowed us to collect 337 electron density vertical profiles obtained almost simultaneously with the radar and the COSMIC in the radar vicinity. The COSMIC electron density profiles were compared with those from the radar, Digisonde, and the IRI model. The comparison included 4 seasons and 2 solar activity levels (low and moderate).The number of simultaneous cases was ~10 times more than in the previous incoherent scatter radar comparisons. In the case of the bottomside characteristics (peak density and bottomside electron content), the deviations between the COSMIC and the ground-based facilities data may be interpreted as the COSMIC measurement errors without significant systematic biases and with root-mean-square values that are ~1.4-1.6 times smaller those that from the IRI model prediction. In the case of the topside characteristics (topside electron content and ionospheric electron content), the IRI model overestimates
the COSMIC data by 0.6-0.8 tecu on average, and the COSMIC overestimates the Irkutsk incoherent scatter radar data by 1.0-1.1 tecu on average. The percentage differences between the radar and COSMIC in the topside electron content can reach 80%. In terms of the root-mean-square deviation, the COSMIC and the radar agree better than each of them agrees with the IRI model.
Keywords: topside ionosphere, COSMIC/FORMOSAT-3, incoherent scatter radar, Digisonde.
1. Introduction
The Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC or FORMOSAT-3) (Schreiner et al., 2007) provides vertical electron density profiles globally and thus represents a unique system for studying the ionospheric electron density on a global scale and using these data for different application and empirical model assimilation. The validation of the COSMIC profiles through comparison with ground-based radio-sounding facilities data has been performed in many papers (e.g. Lei et al., 2007; Stolle et al., 2004; Wu et al., 2009; Chu et al., 2010; Krankowski et al., 2011; Cherniak and Zakharenkova, 2014; Hu et al., 2014; Mikhailov et al., 2014). The radio facilities considered in the above-cited papers are ionosondes (Wu et al., 2009; Chu et al., 2010; Krankowski et al., 2011; Hu et al., 2014) and/or incoherent scatter radars (Lei et al., 2007; Stolle et al., 2004; Cherniak and Zakharenkova, 2014; Mikhailov et al., 2014). In case of ionosondes, the comparisons were made for the bottomside characteristics (F2 peak density (NmF2) and the peak height (hmF2)), while the comparisons with incoherent scatter radars gave an additional opportunity to validate of the topside electron density. The main objective of this paper is a comparative analysis of the electron density vertical profiles obtained with the COSMIC satellite and the Irkutsk incoherent scatter radar (IISR) (52.9N, 103.3E) (Potekhin et al., 2008) during the periods of long-duration continuous radar observations presented in Table 1. For these periods we collected 337 COSMIC measurements that have tangent points at hmF2 within ±5º by latitude and longitude of the IISR location and the time within ±10 min of the IISR measurements (Table 1). Additionally, we compare the COSMIC data with those from the Irkutsk Digisonde (IDPS-4) (52.3N, 104.3E) (Reinisch et al., 1997) located ~100 km away from the IISR, and from the International Reference Ionosphere model (IRI) (Bilitza et al., 2014). For the IRI calculations we used two options of the topside ionosphere: IRI-Nequick and Corrected IRI 01. In the 170–600 km range both options give close results, and further we use the Corrected IRI 01 option prediction shortly denoted as IRI.
The distinctive features of this paper are the following: (1) use of the long-duration continuous IISR observations that cover various seasons and solar activity levels providing more simultaneous measurements than the previous incoherent scatter radar comparisons; and (2) special emphasis to the comparison of the topside electron density. The methods of electron density profile construction and the characteristics selected for the comparison are described in the next section.
2. Methods of electron density profile construction
Each COSMIC microsatellite has a GPS Occultation Experiment payload to operate the ionospheric radio occultation (RO) measurements. The methods of electron density profile construction from RO measurements is described by Tsai et al. (2001) and Lei et al. (2007). The assumption of spherical symmetry of electron density is the most significant source of error in the retrieval of the electron density profiles when we interpret them as actual vertical profiles (Lei et al., 2007). The geographical location of the tangent points path at the top and at the bottom of a profile may differ by several hundred kilometers, and due to the presence of latitudinal and longitudinal gradients in the ionosphere, the COSMIC profile represents spatial averaged information about electron density (Krankowski et al., 2011). All ionogram data of the IDPS-4were manually scaled using an interactive ionogram scaling software, SAO Explorer (Reinisch et al., 2004; Khmyrov et al., 2008). The electron density profiles were inverted from all suitable ionogram traces using the NHPC method (Reinisch and Huang, 1983). The electron density above hmF2 was estimated using the Reinisch and Huang (2001) technique based on the profile extrapolation by the Chapman function with the scale height determined at hmF2. The IISR radiates and receives only one linear polarization, and thus the received power profile P as function of height z is modulated by the Faraday rotation. For near vertical radar beam, the power profile P can be described by the following equations:
P z A
Nez 2 cos 2 ( z ) B, ( z ) 2 f p2 z f H zcos( z)dz , 2 ( z / cos( )) cf cos( ) 0 z
(1)
where A and B are multiplicative and additive constants, respectively; Ne(z) is the electron density profile; β is the radar beam zenith angle (10º in our case); φ is the Faraday rotation phase; fp is the electron plasma frequency ( f p2[MHz] Ne[105 cm3 ] / 0.124 ); fH is the electron gyrofrequency; θ is the angle between the radar beam and the Earth's magnetic field line (15º in our case); f is the operating frequency (155.5 MHz in our case); and c is the light velocity. From one hand, the Faraday modulation does not enable to calculate Ne(z) directly from P(z), but, on the other hand, it gives an opportunity to obtain Ne(z) without an ionosonde calibration.
In this study we use the approximation of Ne(z) by the Chapman-like function:
Ne( z ) NmF 2 exp(1 x exp( x)), where x ( z hmF 2) / H T at z hmF 2, and
(2)
x ( z hmF 2) / H B at z hmF 2,
where HT and HB are the topside and bottomside scale heights, respectively. The Ne(z) profile characteristics (NmF2, hmF2, HT , and HB) as well as A and B constants were derived by least squares fit of the model P(z) (calculated with Eq. (1) and (2)) to the received power profile. Figure 1 shows the received and model power profiles as well as the obtained electron density profile. Note, that this method does not give a high height resolution, but allows stable automatic processing of the IISR received power profiles including cases of low signal-to-noise ratio. For the comparison we selected the following electron density characteristics: the peak density (NmF2), the bottomside (170-300 km) electron content (BTEC), the topside (300-600 km) electron content (TTEC), and the electron content in 170-600 km height range that we denote as an ionospheric electron content (IEC). Initially, we determined the bottomside electron content as the electron content in the 170 km -hmF2 height range. Further, we found that the contribution of the difference in hmF2 to the difference in BTEC is much more than the contribution of the difference in the bottomside electron density. For this reason, we decided to use the constant boundary (300 km) between the bottomside and topside ionosphere. When hmF2 is less than 300 km, IDPS-4 may give errors of the profile extrapolation above hmF2. Figure 2 shows the IDPS4-IISR comparison of BTEC (scatterplot of all the data and year-to-year of the mean (ΔBTEC) and root-mean-square (σBTEC) deviations). It is seen that σBTEC is noticeably larger than ΔBTEC and varies from 0.25 to 0.45 tecu. As will be shown below, the IDPS4-IISR σBTEC is ~ 2 times smaller than that for COSMIC-IISR and COSMIC-IDPS-4 comparisons, and thus it is reasonable to consider both COSMIC-IISR and COSMIC-IDPS-4 comparisons.
3. Comparison and Discussion
Our comparison was made in terms of the mean deviation and the root-mean-square deviation between the COSMIC data and those from IISR, IDPS-4, and IRI. The mean deviation (Δ) characterizes the degree of systematic underestimation or overestimation, while the root-mean-square deviation (σ) is a measure of random differences between the data. Figure 3 shows the comparison of NmF2 by the scatterplots of COSMIC vs. IISR, COSMIC vs. IDPS-4, IRI vs. IISR, and IRI vs. IDPS-4 for low (2007-2009) and moderate (2011-2013) solar activity,
as well as, in terms of Δ and σ as functions of year for COSMIC - IISR, COSMIC - IDPS-4, IRI - IISR, and IRI - IDPS-4 deviations. The COSMIC - IISR comparison of NmF2 indicates that ΔNmF2 is mainly much less than σNmF2 that changes from 0.25-0.32·105 cm-3 at low solar activity (2007-2009) up to 0.71.0·105 cm-3 at the highest solar activity for the considered period (2011-2013) with the average σNmF2 ~ 0.5·105 cm-3. The COSMIC - IDPS-4 comparison of NmF2 shows a similar pattern. These results agree closely with those reported by Krankowski et al. (2011) (σNmF2 ~ 0.2·105 cm-3 for the European ionosondes at low solar activity); and Hu et al. (2014) (σNmF2 ~ 0.6·105 cm-3 for the Mohe ionosonde (53.5N, 123.3) at moderate solar activity); and Cherniak and Zakharenkova (2014) (σNmF2 ~ 0.3·105 cm-3 for the Kharkov incoherent scatter radar (49.6, 36.3E) at low solar activity). In all these comparisons ΔNmF2 was noticeably smaller than σNmF2. The IRI - IISR comparison give the average σNmF2 ~ 0.8·105 cm-3 and the average underestimation ~ 0.3·105 cm-3; and thus we may conclude that the IISR agrees better with COSMIC (σNmF2 ~ 0.5·105 cm-3 under small ΔNmF2) than with the IRI model (σNmF2 ~ 0.8·105 cm-3 under ΔNmF2 ~ 0.3·105 cm-3). As mentioned above, the COSMIC does not give an actual vertical profile due to the presence of latitudinal and longitudinal gradients. Assuming that NmF2 from the IISR and IDPS-4 are much closer to actual values than those from the COSMIC, we may interpret the COSMIC - IISR and COSMIC - IDPS-4 deviations as the COSMIC measurement errors without significant systematic biases and with root-meansquare value that is ~1.6 times smaller than that from the IRI prediction. The rise in σNmF2 with increasing solar activity may be explained by the corresponding rise in latitudinal and longitudinal gradients of the electron density. We can not say that our comparison of NmF2 completely agrees with all such comparisons at midlatitudes. Hu et al. (2014) and Mikhailov et al. (2014) reported that the COSMIC overestimated NmF2 obtained with ionosondes or incoherent scatter radar by the values comparable or even larger than the root-mean-square deviations. All these comparisons relate to the sites located at latitudes near 40N or lower. Hu et al. (2014), considering that the overestimation increases with lowering the latitude, assumed that overestimation may be associated with the effects of the equatorial ionization anomaly (EIA). Yue et al. (2010) simulation r showed that the COSMIC underestimated the true electron density in the region surrounding the EIA crest and overestimated it near the magnetic equator and in the north and south of the EIA crests. Figure 4 shows the same as Figure 3 but for BTEC. The results for the BTEC comparison are generally close to results for the NmF2comparison. The difference between the COSMIC - IISR and COSMIC - IDPS-4 comparisons lies in the fact that σBTEC slightly larger in the COSMIC - IDPS-4 case, and there is a small systematical difference (~0.2 tecu) for the COSMIC - IISR comparisons unlike the
COSMIC - IDPS-4 case. Such a difference may be explained by errors of the profile extrapolation above hmF2 (when hmF2 is noticeably less than 300 km). Also note that σBTEC for the COSMIC - IISR and COSMIC - IDPS-4 comparisons is ~1.4 times smaller than for the IISR - IRI comparison, and this ratio is slightly less than that for σNmF2. Figure 5 shows the comparison of TTEC by the scatterplots of COSMIC vs. IISR and IRI vs. IISR for low (2007-2009) and moderate (2011-2013) solar activity, as well as, in terms of Δ and σ as functions of year for COSMIC – IISR and IRI - IISR deviations. The TTEC comparison differs significantly from the comparison of the bottomside characteristics (NmF2 and BTEC). The main difference lies in the fact that the COSMIC systematically overestimates the IISR TTEC by ~1 tecu on average, and this overestimation is close to σTTEC. The dependence of the COSMIC - IISR ΔTTEC on solar activity is not very pronounced. The IRI model overestimates the COSMIC TTEC (by ~0.8 tecu on average), which in turn is ~1 tecu larger than the IISR TTEC. The smallest average root-mean-square deviation is seen for the COSMIC - IISR comparisons (~1 tecu), and the largest one is seen for the IRI - IISR comparisons (~1.5 tecu) with intermediate σ for the COSMIC - IRI comparison. (~1.2 tecu). Thus in terms of σ the COSMIC and IISR agree better than each of them agrees with the IRI. None of the previous incoherent scatter radar studies have reported that the COSMIC overestimates the topside electron density with a better agreement in the bottomside. Stolle et al. (2004) did not report about any differences between the NmF2 and 150-420 km electron content statistics under the comparison of the radio occultation and EISCAT measurements. Cherniak and Zakharenkova (2014) did not provide the quantitative estimation of the deviations between the COSMIC and the Kharkov incoherent scatter radar (49.6, 36.3E) in the topside and made a qualitative conclusion that the topside profile shape showed rather good agreement between the COSMIC and the radar. Lei et al. (2007) concluded that in most cases the shapes of the Millstone Hill incoherent scatter radar (42.6N, 71.5W) profiles are captured by the COSMIC data. Only Mikhailov et al. (2014), using 35 simultaneous measurements with the COSMIC and the Millstone Hill incoherent scatter radar at low solar activity, made the separate comparison for the topside and bottom side parts. They concluded that in the majority of cases the COSMIC topside profile coincided fairly well with the radar observations, while the coincidence in the bottomside profile was not good for the 40% of the analyzed cases. Thus, they demonstrated a result that is opposite to our findings, and the reason for this difference is not clear. Possibly, the comparison results strongly depend on the ground-based facility location, at least the mentioned above ionosonde comparisons show that this may be the case. Figure 6 shows the same as Figure 5 but for IEC. The comparison of IEC = BTEC + TTEC summarizes the features of the BTEC and TTEC comparisons. The IRI model overestimates the COSMIC
IEC by ~0.6 tecu on average, and the COSMIC overestimates the IISR IEC by ~1.1 tecu on average and this overestimation does not have a pronounced dependence on solar activity. In terms of the root-meansquare deviation, the COSMIC and IISR agree better (σ ~1.3 tecu) than each of them agrees with the IRI (σ ~1.8 and 2 tecu). Below we have considered the absolute differences between the COSMIC and IISR data, whereas the percentage differences are also of interest due to their strong dependency on the background values. Figure 7 shows the COSMIC - IISR comparison in terms of the mean percentage deviation Δ(%) and the root-mean-square percentage deviation σ(%) as functions of year for NmF2, BTEC, TTEC and IEC. For the case of the bottomside characteristics (NmF2 and BTEC) the results for the percentage differences are similar to those for the absolute differences: Δ(%) is noticeably smaller than σ(%), and σ(%) rises with increasing solar activity from 8-10 to 20-25%. In the case of TTEC we see significant (from 25 to 80%) and nonmonotonic variations of σ(%) from year to year, while the absolute σTTEC values do not show such behavior. This means that that the absolute differences in TTEC depends weakly on background TTEC values, and we obtain larger percentage differences for lower background values. The case of IEC is an intermediate case between the bottomside characteristics and TTEC: the year to year variations are nonmonotonic, but the range of σ (15-40%) is not as large as in the TTEC case.
4. Conclusion
The long-duration continuous Irkutsk incoherent scatter radar observations allowed us to collect 337 electron density vertical profiles obtained almost simultaneously with the radar and the COSMIC in the radar vicinity. The comparison included 4 seasons and 2 solar activity levels (low and moderate), and the number of simultaneous cases was ~10 times more than in the previous incoherent scatter radar comparisons. The validation the COSMIC profiles through the comparison with the Irkutsk incoherent scatter radar and Irkutsk Digisonde gave the following results. In the case of the bottomside characteristics (peak density and bottomside electron content), the deviations between the COSMIC and the ground-based facilities data may be interpreted as the COSMIC measurement errors without significant systematic biases and with root-mean-square values that are ~1.41.6 times smaller those that from the IRI model prediction. The rise in the root-mean-square errors with increasing solar activity may be explained by the corresponding rise in latitudinal and longitudinal gradients of the electron density. Such a comparison pattern agree closely with the results obtained for the groundbased facilities located at latitudes near 50N, while the comparisons at latitudes near 40N or lower show
that the COSMIC overestimates NmF2 from the ground-based facilities by the values comparable or even larger than the root-mean-square deviations. In the case of the topside characteristics (topside electron content and ionospheric electron content), the IRI model overestimates the COSMIC data by 0.6-0.8 tecu on average, and the COSMIC overestimates the Irkutsk incoherent scatter radar data by 1.0-1.1 tecu on average. The percentage differences between the radar and COSMIC in the topside electron content can reach 80%. In terms of the root-mean-square deviation, the COSMIC and the radar agree better than each of them agrees with the IRI model. None of the previous incoherent scatter radar studies have reported that the COSMIC overestimates the topside electron density with a better agreement in the bottomside. Moreover, the validation the COSMIC with the Millstone Hill incoherent scatter radar demonstrates results that are opposite to our findings. Possibly, the comparison results strongly depend on the ground-based facility location, as it is seen from the comparisons of NmF2.
Acknowledgments
The reported study was funded by RFBR and NSC according to the Russian-Taiwanese joint research project 14-05-92002 HHC_a / NSC103-2923-M-006-002-MY3 and partially supported by RF President Grant of Public Support for RF Leading Scientific Schools (NSh-6894.2016.5).
References
Bilitza, D., Altadill, D., Zhang, Y., Mertens, C., Truhlik, V., Richards, P., McKinnell, L.-A., Reinisch, B., 2014. The International Reference Ionosphere 2012 - a model of international collaboration. J. Space Weather Space Clim. 4 (A07), 1-12. Cherniak, Iu.V., Zakharenkova, I.E., 2014. Validation of FORMOSAT-3/COSMIC radio occultation electron density profiles by incoherent scatter radar data. Adv. Space Res. 53 (9), 1304-1312. Chu, Y.-H., Su, C.-L., Ko, H.-T., 2010. A global survey of COSMIC ionospheric peak electron density and its height: A comparison with ground-based ionosonde measurements. Adv. Space Res., 46 (4), 431– 439. Hu, L., Ning, B., Liu, L., Zhao, B., Li, G., Wu, B., Huang, Z., Hao, X., Chang, S., Wu, Z., 2014. Validation of COSMIC ionospheric peak parameters by the measurements of an ionosonde chain in China. Ann. Geophys., 32(10), 1311-1319.
Khmyrov, G.M., Galkin, I.A., Kozlov, A.V., Reinisch, B.W., McElroy, J., Dozois C., 2008. Exploring digisonde ionogram data with SAO-X and DIDBase. Radio Sounding and Plasma Physics, AIP Conf. Proc. 974, 175-185. Krankowski, A., Zakharenkova, I.E., Krypiak-Gregorczyk, A., Shagimuratov, I.I., Wielgosz, P., 2011. Ionospheric electron density observed by FORMOSAT-3/COSMIC over the European region and validated by ionosonde data. J. Geod. 85 (12), 949–964. http://dx.doi.org/10.1007/s00190-011-0481-z. Lei, J., Syndergaard, S., Burns, A.G., et al., 2007. Comparison of COSMIC ionospheric measurements with ground-based observations and model predictions: preliminary results. J. Geophys. Res. 112, A07308. http://dx.doi.org/10.1029/2006JA012240. Mikhailov, A.V., Belehaki, A., Perrone, L., Zolesi, B., Tsagouri I., 2014. On the possible use of radio occultation middle latitude electron density profiles to retrieve thermospheric parameters. J. Space Weather Space Clim. 4, A12. http://dx.doi.org/10.1051/swsc/2014009. Potekhin, A.P., Medvedev, A.V., Zavorin, A.V., Kushnarev, D.S., Lebedev, V.P., Shpynev, B.G., 2008. Development of diagnostic capabilities of the Irkutsk incoherent scattering radar. Cosmic Research. 46 (4), 347-353. Reinisch, B.W., Huang, X., 1983. Automatic Calculation of Electron Density Profiles from Digital Ionograms, 3, Processing of Bottomside Ionograms. Radio Sci. 18(3), 477-492. Reinisch, B.W., Haines, D.M., Bibl, K., Galkin, I., Huang, X., Kitrosser, D.F., Sales, G.S., Scali, J.L., 1997. Ionospheric sounding support of OTH radar. Radio Sci. 32 (4), 1681-1694. Reinisch, B.W., Huang, X., 2001. Vertical electron content from ionograms in real time. Radio Sci. 36 (2), 335-342. Reinisch, B.W., Galkin I. A., Khmyrov G., Kozlov, A., Kitrosser, D.F., 2004. Automated collection and dissemination of ionospheric data from the digisonde network. Adv. Radio Sci. 2, 241-247. Schreiner, W., Rocken C., Sokolovsky, S., Syndergaard, S., Hunt D., 2007. Estimates of the precision of GPS radio occultations from the COSMIC/FORMOSAT-3 mission. Geophys. Res. Lett., 34, L04808. http://dx.doi.org/10.1029/2006GL027557. Stolle, C., Jakowski, N., Schlegel, K., Rietveld, M., 2004. Comparison of high latitude electron density profiles obtained with the GPS radio occultation technique and EISCAT measurements. Ann. Geophys. 22 (6), 2015–2022. Tsai, L.C., Tsai, W.H., Schreiner, W.S., Berkey, F.T., Liu, J.Y., 2001. Comparisons of GPS/MET retrieved ionospheric electron density and ground based ionosonde data. Earth Planets Space 53, 193–205. Wu, X., Hu, X., Gong, X., Zhang, X., Wang, X., 2009. Analysis of inversion errors of ionospheric radio occultation. GPS Solutions. 13 (3), 231–239. http://dx.doi.org/10.1007/s10291-008-0116-x.
Yue, X., Schreiner, W. S., Lei, J., Sokolovskiy, S. V., Rocken, C., Hunt, D. C., and Kuo, Y.-H., 2010. Error analysis of Abel retrieved electron density profiles from radio occultation measurements. Ann. Geophys. 28 (1), 217-222.
Table captions Table.1 Periods of long-duration continuous Irkutsk incoherent scatter radar observations.
Figure captions
Fig.1 Received power profile of Irkutsk incoherent scatter radar (left, black), obtained electron density profile (right, grey), and model power profile calculated with Eq. (1) and (2) (left, grey). Fig.2 IDPS4-IISR comparison of BTEC (scatterplot of all the data and year-to-year of the mean (ΔBTEC) and root-mean-square (σBTEC) deviations). Fig.3 Comparison of NmF2 by the scatterplots of COSMIC vs. IISR, COSMIC vs. IDPS-4, IRI vs. IISR, and IRI vs. IDPS-4 (from left to right) for low (top panel) and moderate (middle panel) solar activity, and in terms of Δ (grey) and σ (black) as functions of year (bottom panel) for COSMIC - IISR, COSMIC IDPS-4, IRI - IISR, and IRI - IDPS-4 deviations (from left to right). Fig.4 The same as Figure 3 but for BTEC. Fig.5 Comparison of TTEC by the scatterplots of COSMIC vs. IISR and IRI vs. IISR (from left to right) for low (top panel) and moderate (middle panel) solar activity, and in terms of Δ (grey) and σ (black) as functions of year (bottom panel) for COSMIC – IISR and IRI - IISR deviations (from left to right). Fig.6 The same as Figure 5 but for IEC. Fig. 7 COSMIC - IISR comparison in terms of mean percentage deviation Δ(%) and root-mean-square percentage deviation σ(%) as functions of year for NmF2, BTEC, TTEC and IEC
650 600
600
550
550
500
500
Height (km)
Height (km)
650
Apr 20, 2011 07:12 UT (14:12 LT)
450 400 350
450 400 350
300
300
250
250
200
200
0.5 Fig. 1
0.6
0.7
0.8
0.9
Power profile (rel. un.)
1
0
2
4
6
8
Electron density profile
10
12
(105cm-3)
2
BTEC, BTEC (tecu)
IDPS-4 BTEC (tecu)
12 10 8 6 4 2 0
0
2
4
6
8
IISR BTEC (tecu) Fig. 2
10
12
1.5 1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
Low Solar activity
2
4
IISR NmF2
6
14 12 10 8 6 4 2 4
6
8
10 12 14 16
IISR NmF2 (105cm-3)
NmF2 (105cm-3), NmF2
NmF2 (105cm-3), NmF2
1 0.5 0 -0.5 2010
Year
Fig. 3
2012
2
0
8
0
2014
2
4
IISR NmF2
6
(105cm-3)
4
2
0
8
Moderate Solar activity
16
14
14
12 10 8 6 4 2 0
2
4
6
8
12 10 8 6 4 2 0
10 12 14 16
IDPS-4 NmF2 (105cm-3)
1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
4
6
8
10 12 14 16
IISR NmF2 (105cm-3)
1.5 1 0.5 0 -0.5 2008
2010
Year
2012
6
8
8 6 4 2 0
2
4
6
8
10 12 14 16
IDPS-4 NmF2 (105cm-3)
IRI - IDPS-4
2
-1 2006
4
10
IRI - IISR
2 1.5
2
2
IDPS-4 NmF2 (105cm-3)
12
0 0
0
Moderate Solar activity
14
Cosmic - IDPS-4
1.5
2008
6
(105cm-3)
4
6
16
Cosmic - IISR 2
-1 2006
4
6
16
0 2
2
Moderate Solar activity COSMIC NmF2 (105cm-3)
COSMIC NmF2 (105cm-3)
16
0
0
IDPS-4 NmF2
Moderate Solar activity
0
0
8
(105cm-3)
IRI NmF2 (105cm-3)
0
2
NmF2 (105cm-3), NmF2
0
4
IRI NmF2 (105cm-3)
2
6
IRI NmF2 (105cm-3)
4
8
8
NmF2 (105cm-3), NmF2
6
Low Solar activity
Low Solar activity
8
IRI NmF2 (105cm-3)
COSMIC NmF2 (105cm-3)
COSMIC NmF2 (105cm-3)
Low Solar activity 8
2014
2 1.5 1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
Low Solar activity
4
6
0
8
IISR BTEC (tecu)
10 8 6 4 2 4
6
8
10
12
14
IISR BTEC (tecu)
1 0.5 0 -0.5 2010
Year
2012
2
2014
4
6
4
2
0
8
0
14
12
12
12
8 6 4
10 8 6 4
2
2
0
0 0
2
4
6
8
10 12 14 16
1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
6
8
10 8 6 4 2
0
2
4
6
8
10
12
0
14
IISR BTEC (tecu)
0
4
6
8
10
12
14
IRI - IDPS-4
IRI - IISR 2
1.5 1 0.5 0 -0.5 -1 2006
2
IDPS-4 BTEC (tecu)
2
1.5
4
Moderate Solar activity
14
10
2
IDPS-4 BTEC (tecu)
Moderate Solar activity
Cosmic - IDPS-4
1.5
2008
0
IISR BTEC (tecu)
2
BTEC, BTEC (tecu)
BTEC, BTEC (tecu)
8
6
14
Cosmic - IISR
Fig. 4
6
IDPS-4 BTEC (tecu)
2
-1 2006
4
IRI BTEC (tecu)
COSMIC BTEC (tecu)
COSMIC BTEC (tecu)
12
2
2
Moderate Solar activity
14
0
2
IDPS-4 BTEC (tecu)
Moderate Solar activity
0
4
0 0
IRI BTEC (tecu)
2
6
IRI BTEC (tecu)
2
4
BTEC, BTEC (tecu)
0
6
8
BTEC, BTEC (tecu)
2
IRI BTEC (tecu)
COSMIC BTEC (tecu)
COSMIC BTEC (tecu)
4
Low Solar activity
8
8
6
0
Low Solar activity
Low Solar activity
8
2008
2010
Year
2012
2014
1.5 1 0.5 0 -0.5 -1 2006
2008
2010
Year
2012
2014
Low Solar activity
Low Solar activity 8
IRI TTEC (tecu)
COSMIC TTEC (tecu)
8
6
4
2
0
0
2
4
6
6
4
2
0
8
0
2
IISR TTEC (tecu)
16
16
14
14
12 10 8 6 4 2
10 8 6 4 2
0
2
4
6
8
0
10 12 14 16
0
4
6
8
10 12 14 16
IRI - IISR 3
TTEC, TTEC (tecu)
3 2.5 2 1.5 1 0.5 0 -0.5 2006
2
IISR TTEC (tecu)
Cosmic - IISR TTEC, TTEC (tecu)
8
12
IISR TTEC (tecu)
2008
2010
Year Fig. 5
6
Moderate Solar activity IRI TTEC (tecu)
COSMIC TTEC (tecu)
Moderate Solar activity
0
4
IISR TTEC (tecu)
2012
2014
2.5 2 1.5 1 0.5 0 -0.5 2006
2008
2010
Year
2012
2014
Low Solar activity
12
12
10
10
IRI IEC (tecu)
COSMIC IEC (tecu)
Low Solar activity
8 6 4
6 4 2
2 0
8
0
2
4
6
8
10
0
12
0
2
28
24
24
20 16 12 8 4
10
12
16 12 8 4
0
4
8
12
16
20
24
0
28
0
8
12
16
20
24
28
IRI - IISR 4
IEC, IEC (tecu)
4 3 2 1 0 2006
4
IISR IEC (tecu)
Cosmic - IISR IEC, IEC (tecu)
8
20
IISR IEC (tecu)
3 2 1 0
2008
2010
Year Fig. 6
6
Moderate Solar activity
28
IRI IEC (tecu)
COSMIC IEC (tecu)
Moderate Solar activity
0
4
IISR IEC (tecu)
IISR IEC (tecu)
2012
2014
2006
2008
2010
Year
2012
2014
BTEC
NmF2
40 20 0 2006
2008
2010
Year
Fig. 7
2012
2014
80 60 40 20 0 2006
2008
2010
Year
2012
2014
IEC (%), IEC (%)
60
100
100
TTEC (%), TTEC (%)
BTEC (%), BTEC (%)
NmF2 (%), NmF2 (%)
80
IEC
TTEC
100
100
80 60 40 20 0 2006
2008
2010
Year
2012
2014
80 60 40 20 0 2006
2008
2010
Year
2012
2014
Observation periods JUN 05 - JUN 19, 2007 AUG 25 – SEP 28, 2008 APR 01 – APR 18, 2009 JAN 12 - FEB 28, 2010 JAN 16 - FEB 16, 2011 APR 12 - APR 21, 2011 JAN 18 - FEB 05, 2012 APR 06 – APR 22, 2012 JAN 01 – JAN 21, 2013 JUN 21 – JUN 30, 2013 DEC 25 - DEC 31, 2013 In total Table 1
Number of Days 15 35 16 48 32 10 8 12 19 9 4 208
Number of Simultaneous cases 46 44 32 62 28 6 13 23 49 24 10 337
Yearly mean F10.7 (s.f.u.) 73 69 71 80 114 120 123