Available online at www.sciencedirect.com

www.elsevier.com/locate/asr

ScienceDirect

Advances in Space Research 61 (2018) 295–315

Bottom side profiles for two close stations at the southern crest ofthe EIA: Differences and comparison with IRI-2012 and NeQuick2

for low and high solar activity

L. Perna a,b,⇑, K. Venkatesh c, V.G. Pillat c, M. Pezzopane b, P.R. Fagundes c,R.G. Ezquer d,e,f, M.A. Cabrera d,g

aUniversita di Bologna ‘‘Alma Mater Studiorum”, Viale Berti Pichat 6/2, 40126 Bologna, Italyb Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, 00143 Roma, Italy

cUniversidade do Vale do Paraiba (UNIVAP), IP&D, Sao Jose dos Campos, SP, BrazildCIASUR, Facultad Regional Tucuman, Universidad Tecnologica Nacional, Argentina

eLaboratorio de Ionosfera, Universidad Nacional de Tucuman, ArgentinafCONICET, Buenos Aires, Argentina

gLaboratorio de Telecomunicaciones, Universidad Nacional de Tucuman, Argentina

Received 20 June 2017; received in revised form 8 September 2017; accepted 5 October 2017Available online 14 October 2017

Abstract

Bottom side electron density profiles for two stations at the southern crest of the Equatorial Ionization Anomaly (EIA), Sao Jose dosCampos (23.1�S, 314.5�E, dip latitude 19.8�S; Brazil) and Tucuman (26.9�S, 294.6�E, dip latitude 14.0�S; Argentina), located at similarlatitude and separated by only 20� in longitude, have been compared during equinoctial, winter and summer months under low (year2008, minimum of the solar cycle 23/24) and high solar activity (years 2013–2014, maximum of the solar cycle 24) conditions. An analysisof parameters describing the bottom side part of the electron density profile, namely the peak electron density NmF2, the height hmF2 atwhich it is reached, the thickness parameter B0 and the shape parameter B1, is carried out. Further, a comparison of bottom side profilesand F-layer parameters with the corresponding outputs of IRI-2012 and NeQuick2 models is also reported. The variations of NmF2 atboth stations reveal the absence of semi-annual anomaly for low solar activity (LSA), evidencing the anomalous activity of the last solarminimum, while those related to hmF2 show an uplift of the ionosphere for high solar activity (HSA). As expected, the EIA is partic-ularly visible at both stations during equinox for HSA, when its strength is at maximum in the South American sector. Despite the similarlatitude of the two stations upon the southern crest of the EIA, the anomaly effect is more pronounced at Tucuman than at Sao Jose dosCampos. The differences encountered between these very close stations suggest that in this sector relevant longitudinal-dependent vari-ations could occur, with the longitudinal gradient of the Equatorial Electrojet that plays a key role to explain such differences togetherwith the 5.8� separation in dip latitude between the two ionosondes. Furthermore at Tucuman, the daily peak value of NmF2 around21:00 LT during equinox for HSA is in temporal coincidence with an impulsive enhancement of hmF2, showing a kind of ‘‘elasticrebound” under the action of the EIA. IRI-2012 and NeQuick2 bottom side profiles show significant deviations from ionosonde obser-vations. In particular, both models provide a clear underestimation of the EIA strength at both stations, with more pronounced differ-ences for Tucuman. Large discrepancies are obtained for the parameter hmF2 for HSA during daytime at Sao Jose dos Campos, whereclear underestimations made by both models are observed. The shape parameter B0 is quite well described by the IRI-2012 model, with

https://doi.org/10.1016/j.asr.2017.10.007

0273-1177/� 2017 COSPAR. Published by Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: Istituto Nazionale di Geofisica e Vulcanologia, Via di Vigna Murata 605, 00143 Roma, Italy.E-mail addresses: gigiperna@gmail.com (L. Perna), venkatkau@gmail.com (K. Venkatesh), valdirgp@univap.br (V.G. Pillat), Michael.pezzopa-

ne@ingv.it (M. Pezzopane), fagundes@univap.br (P.R. Fagundes), rezquer@herrera.unt.edu.ar (R.G. Ezquer), mcabrera@herrera.unt.edu.ar(M.A. Cabrera).

296 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

very good agreement in particular during equinox for both stations for both LSA and HSA. On the contrary, the two models show pooragreements with ionosonde data concerning the shape parameter B1.� 2017 COSPAR. Published by Elsevier Ltd. All rights reserved.

Keywords: Bottom side electron density profiles; IRI and NeQuick2 model; B0 and B1 parameters; EIA longitudinal variability

1. Introduction

The ionospheric electron density at equatorial and lowlatitudes shows significant variations with time, season, lat-itude and altitude due to complex electrodynamic phenom-ena (e.g., Stening, 1992; Reinisch and Huang, 1996; Heelis,2004; Lee et al., 2008; Venkatesh et al., 2014a). In particu-lar, at low latitudes and along the geomagnetic equator, theionosphere shows a significant variability due to the dom-inant phenomenon of the Equatorial Ionization Anomaly(EIA) (e.g., Lyon and Thomas, 1963; MacDougall, 1969;Balan and Iyer, 1983; Batista and Abdu, 2004; Abduet al., 2008), also called Appleton anomaly (Appleton,1946), linked to the equatorial fountain effect (Martyn,1955; Duncan, 1960; MacDougall, 1969; Balan and Iyer,1983; Batista and Abdu, 2004; Abdu et al., 2008). Asdescribed by Abdu et al. (2008), during the day the devel-opment of the EIA is generated by the E-region dynamozonal electric field, linked to the Equatorial Electrojet(EEJ), which gives rise to an E � B plasma vertical driftleading to the aforementioned plasma fountain effect. Asdiscussed by Martyn (1955) and Duncan (1960), ionizationis uplifted over the magnetic equator and diffuses down thegeomagnetic lines of force, under pressure gradient andgravity forces (Stening, 1992), modifying the electron den-sity concentrations at low latitudes where the F-regionplasma increases, thus forming the electron density crestson both sides (north and south) of the magnetic equator(at about ±15–18� geomagnetic latitude) (Lyon andThomas, 1963). Consequently, the scenario of the equato-rial and low latitude ionospheric F-region is the following:F-region is lifted up at the magnetic equator, with adecrease of the peak density, while the F-region peak den-sity increases at the crests of the anomaly (Batista andAbdu, 2004). The persistence of the EIA at nighttimehours, depending on the season and solar activity, is knownto be produced by the post sunset enhancement in the east-ward electric field generated by the F-region dynamoaction (Chuo, 2012). This dynamo action, in turn, resultsfrom the eastward component of the thermospheric windblowing in the region of the decreasing dawn-to-dusk E-layer Pedersen conductivity distribution (Heelis, 2004).

Owing to these particular conditions, the equatorial andlow-latitude ionosphere show a strong spatio-temporalvariability making the modeling of the ionospheric param-eters particularly difficult.

The International Reference Ionosphere (IRI) (Bilitza,2001; Bilitza and Reinisch, 2008; Bilitza et al., 1990,2014, 2017) is an empirical model for the description ofthe ionosphere. It is an international project sponsoredby the Committee on Space Research (COSPAR) and theInternational Union of Radio Science (URSI), based onan extensive database and able to capture much of therepeatable characteristics of the ionosphere such as theelectron density, the electron content, the electron temper-ature and the ion composition, as a function of height,location, and local time for quiet and storm-time periods(Araujo-Pradere et al., 2011, 2013; Zakharenkova et al.,2013). The IRI model was first developed in 1978 (Raweret al., 1978) and thereafter several updated versions havebeen released, with IRI-2012 (Bilitza et al., 2014) andIRI-2016 (Bilitza et al., 2017) representing the most recentones.

The NeQuick2 model (Nava et al., 2008) is the secondversion of NeQuick, an empirical ionospheric model beingwidely used for the estimation of electron density profilesand related parameters. The model was developed at theInternational Centre for Theoretical Physics (ICTP) andhas been adopted by the International TelecommunicationUnion (ITU) for TEC modeling (ITU, 2003). Its outputshave been validated by several workers who reported agood agreement at mid latitudes and improved perfor-mances while assimilating measurements (Jodogne et al.,2005; Bidaine and Warnant, 2010).

The characterization of the ionospheric F-region repre-sents a central aim owing to the fact that the correspondingelectron density provides the main contribution to theTotal Electron Content (TEC) that largely affects radiowave propagations (Asmare et al., 2014). Knowledge ofthe spatial distribution of the electron density in the iono-sphere, especially the ionospheric profile N(h), is crucial forHF telecommunication, ionospheric tomography, GNSSoperations and ionospheric studies. Such representation isalso useful in practical space weather applications and formodeling various physical processes in the ionosphere(Chuo, 2012).

In particular, the study of the bottom side profile, thatrepresents the electron density distribution under the F2-layer peak, is of primary importance to improve the modelreliability and consequently the efficiency of radio commu-nications. The F2-layer bottom side electron density profilecan be well described by means of the maximum electron

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 297

density NmF2, the height hmF2 at which it is reached, thethickness parameter B0 and the shape parameter B1 (Chuo,2012). NmF2 represents the absolute maximum electrondensity in the profile and is linked to the critical frequencyof the F2-layer foF2 by the relation NmF2 = 1.24�1010(foF2)2, with NmF2 expressed in m�3 and foF2 in MHz;foF2 individuates the maximum frequency reflected bythe ionosphere for a vertical travelling wave. The bottomside thickness parameter B0 is defined as the differencebetween hmF2 and the height where the electron densityequals to 0.24 times NmF2, in the absence of the F1-layer or the F1-peak height hmF1, if the latter occurs.The shape parameter B1 describes the shape of the profilebetween the two heights from which B0 is estimated(Reinisch and Huang 1996; Bilitza et al., 1998).

The solar activity represents the main controller of theionospheric variability. The minimum of the cycle 23/24(deep low solar activity for the years 2008–2009), was thelongest and quietest since the advent of space-based mea-surements (Liu et al., 2011). During the period from Jan-uary 2008 to December 2009, 527 spotless days have beenobserved, while they were 226 and 176 for the minimum22/23 (years 1996–1997) and 21/22 (years 1986–1987)respectively; the magnetic field at the solar poles wasapproximately 40% weaker than that of cycle 22/23(Araujo-Pradere et al., 2011). Measurements by the Ulyssesspacecraft revealed a 20% drop in solar-wind pressure sincethe mid-1990s, the lowest point since the start of such mea-surements in the 1960 s (Phillips, 2009). Moreover, Chenet al. (2011, 2012) found a decrease of �15% in the EUVsolar radiation for the last solar minimum in comparisonto the previous one. This decrease explains the lower valuesof foF2 observed by ionospheric stations all around theworld (Liu et al., 2011; Chen et al., 2011; Bilitza et al.,2012). In such particular conditions, problems in modelpredictions have been discussed by several works (e.g.,Heelis et al., 2009; Coley et al., 2010; Luhr and Xiong,2010; Bilitza et al., 2012).

The years 2013–2014, representing the maximum of thesolar cycle 24, have been characterized by a low solar activ-ity in comparison to the previous maxima. A yearly sun-spot average of 65 and 113 for the years 2013 and 2014have been registered respectively. To find a maximum withsuch a low activity it is necessary to turn back to solar cycle12 that registered respectively 90 and 102 sunspot yearlyaverages for 1906 and 1907 respectively. Hence, the mini-mum of the solar cycle 23/24 and the maximum of the cycle24 provide two very interesting natural windows to studythe ionospheric plasma response to such particular solaractivity conditions.

The characteristics of the bottom side profiles have beenlargely studied (e.g., Aggarwal et al., 1996; Sethi andPandey, 2001; Batista and Abdu, 2004; Bertoni et al.,2006; Zhang et al., 2004, 2008; Chen et al., 2006; Sethiet al., 2007, 2009; Altadill et al., 2009; de Jesus et al.,2011; Chuo, 2012; Lee and Reinisch, 2012; Venkateshet al., 2014a; Venkatesh and Fagundes, 2016), showing

variations with the season, solar activity level, and day-to-night alternation.

As mentioned, the equatorial and low-latitude iono-sphere show very particular behaviors due to the dominantphenomenon of the EIA. In particular, the EIA crests inthe South American sector, owing to the curvature of themagnetic equator, represent very complex regions thatstrongly distinguish them from the other ones. Recently,Fagundes et al. (2016) analyzed vTEC values inferred fromtwo latitudinal chains (�15–20� separated in longitude) ofGPS-TEC stations from equatorial region to low latitudesin the East and West Brazilian sectors under the geomag-netic disturbed conditions of the extreme space weatherevent of 17–18 March 2015. They found that the EIAwas very disturbed during the storm main phase, withvTEC values from the equator to beyond the EIA crestsmuch more disturbed in the West sector than in the Eastone. This difference between very close longitudinal sectorshas been observed for the first time and strongly suggeststhat relevant longitudinal-dependent EIA variations inthe South American region can occur. Hence, it is of signif-icant importance to have a clear description of the EIApattern and dynamics in South America for both quietand disturbed geomagnetic conditions.

In this work, bottom side electron density profiles andassociated parameters from two closely spaced ground-based ionosondes, namely Sao Jose dos Campos (23.1�S,314.5�E, dip latitude 19.8�S; Brazil) and Tucuman (26.9�S, 294.6�E, dip latitude 14.0�S; Argentina), located uponthe southern crest of the EIA at similar latitude and only20 degrees separated in longitude, have been comparedfor the very low solar activity year 2008 (sunspot annualaverage R = 3) and the high solar activity years 2013 and2014 (sunspot annual average R = 65 and 113 respectively),under quiet geomagnetic conditions. A comparison withIRI-2012 and NeQuick2 models is also carried out for bothbottom side profiles and the parameters that characterizethem. Moreover, a comparison with IRI-2012 gives usthe possibility to test the performances of the recent‘ABT-2009’ option available for this version of the model.

The main objectives of the study are: (1) to make a com-parison of bottom side profiles at two close stations uponthe southern crest of the EIA in the South American sector,in order to improve our knowledge about the spatial pat-tern of the EIA in this very particular region and to detectpossible relevant longitudinal-dependent variations underquiet geomagnetic conditions; (2) to detect differencesbetween low and high solar activity levels; (3) to test theperformance of IRI-2012 and NeQuick2 models.

2. Dataset and analyses

The ionosonde data from Sao Jose dos Campos (SJ) andTucuman (TU) have been analyzed during the minimum ofthe solar cycle 23/24 and the maximum of the solar cycle24. Fig. 1 shows the locations of the two stations, alongwith the constant main field inclination lines (red for posi-

Fig. 1. Map of the geomagnetic field inclination in the South American sector, with displayed the ionosonde locations of SJ and TU. Red lines indicatepositive inclination, while blue lines indicate negative inclination. Geographic latitudes and longitudes are also indicated. (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version of this article.)

298 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

tive inclination, blue for negative inclination) and the geo-graphic latitude and longitude.

In Sao Jose dos Campos a Canadian Advanced DigitalIonosonde (CADI) is operational and located at the UNI-VAP (Universidade do Vale do Paraıba) since August2000. The digital ionosonde antenna is a double deltadipole array supported by a 20 m tower, where one of thedual antennas is used for transmitting and the other oneis used for receiving (Grant et al., 1995; MacDougallet al., 1997). The ionospheric measurements at Tucumanbegan in 1957, the International Geophysical Year, when

an analog ionosonde was transferred from the Navy ofArgentina to the National University of Tucuman whichwas stopped working in 1987 (Ezquer et al., 2014). InAugust 2007, an AIS-INGV (Zuccheretti et al., 2003) wasinstalled at the Upper Atmosphere and RadiopropagationResearch Center of the Regional Faculty of Tucuman ofthe National Technological University (UTN).

Ionograms for the considered locations have been man-ually scaled to derive the parameters of interest. NmF2 val-ues have been calculated through the relationship NmF2[m�3] = 1.24�1010 (foF2 [MHz])2, using manually scaled

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 299

foF2 values that have been validated according to theInternational Union of Radio Science (URSI) standard(Wakai et al., 1987). Bottom side electron density profilesand the parameter hmF2 have been obtained applying thePOLAN (POLynomial ANalysis) true height inversionalgorithm (Titheridge, 1985) to every ionogram. This pro-cedure provides a very robust and reliable dataset for theparameter NmF2, while both bottom side profiles andhmF2 values can be interested by inaccuracy when an E-valley occurs along the profile.

The IRI electron density distribution below the F2-peakis given by the analytical function (Ramakrishnan andRawer, 1972)

NðhÞ ¼ NmF2½expð�xB1Þ�=coshðxÞ; ð1Þx ¼ ðhmF2� hÞ=B0: ð2Þ

Eq. (1) has been fitted for the bottom side electron den-sity profiles derived from ionosondes to calculate the corre-sponding bottom side thickness (B0) and shape (B1)parameters.

Bottom side profiles and modeled B0 and B1 from IRI-2012 have been obtained running the web based model(http://omniweb.gsfc.nasa.gov/vitmo/iri2012_vitmo.html)using the option ‘‘ABT-2009” (Altadill et al., 2009).Altadill et al. (2009) obtained an improvement of 40%and 20% in B0 and B1 prediction of IRI respectively; there-fore, IRI has been run selecting the ‘‘ABT-2009” option forthe bottom side part of the profile, and the ‘‘NeQuick”option for the topside part of the profile. For the F-peakmodel, the International Radio Consultative Committee(CCIR) coefficients (1967a, 1967b) were preferred to theUnion of Radio Science (URSI), since the CCIR coeffi-cients are recommended for locations on the continents(Rush et al., 1989). Furthermore, Bertoni et al. (2006) com-pared NmF2 and hmF2 from IRI-2001 with ionosonde val-ues in the South American sector and found lowerdifferences using CCIR coefficients than the URSI ones.Lee et al. (2008) also obtained similar results for the stationof Jicamarca (12�S, 77�W; Peru).

To describe the electron density of the ionosphere above90 km and up to the peak of the F2-layer, the NeQuick2model uses a modified DGR profile formulation (DiGiovanni and Radicella, 1990), which includes five semi-Epstein layers (Rawer, 1982) with modeled thicknessparameters (Radicella and Zhang, 1995). Three profileanchor points are used: the E-layer peak, the F1-layer peakand the F2-layer peak, modeled in terms of the ionosondeparameters foE, foF1, foF2 and M(3000)F2. The model

Table 1Monthly mean sunspot number for the months selected to represent the low a

Low solar activity

Month Mean sunspot number

September 2008 1June 2008 5December 2008 1

uses the expression NmA [m�3] = 1.24�1010 (foA [MHz])2

for the peak electron densities (being A = E, F1 or F2), setshmE = 120 km and hmF1 = (hmE + hmF2)/2, with hmF2calculated using the Dudeney formula (Dudeney, 1978,1983), and expresses the bottom side profile as follows(Nava et al., 2008):

NbottomsideðhÞ ¼ NEðhÞ þ NF1ðhÞ þ NF2ðhÞ; ð3Þwhere

NEðhÞ ¼ 4NmE

1þexp h�hmEBE

�nðhÞ� �� �2 exp

h�hmEBE

� nðhÞ� �

;

NF1ðhÞ ¼ 4NmF1

1þexp h�hmF1BF1

�nðhÞ� �� �2 exp

h�hmF1BF1

� nðhÞ� �

;

NF2ðhÞ ¼ 4NmF2

1þexp h�hmF2BF2

� �� �2 exph�hmF2

BF2

� �;

NmE ¼ NmE� NF1ðhmEÞ � NF2ðhmEÞ;NmF1 ¼ NmF1� NEðhmF1Þ � NF2ðhmF1Þ;

nðhÞ ¼ exp 101þ1jh�hmF2j

� �

8>>>>>>>>>>>>>>>>>>>>>><>>>>>>>>>>>>>>>>>>>>>>:

ð4Þ

with BE, BF1 and BF2 that express the thickness parameters(in km) for the layers E, F1 and F2, respectively.

The NeQuick2 profiles for the present work have beenobtained running the on-line version of the model (http://t-ict4d.ictp.it/nequick2/nequick-2-web-model). B0 and B1

have been obtained fitting the profiles with relations (1)and (2). Nevertheless, as shown by comparing Eqs. (1)and (2) with Eqs. (3) and (4), NeQuick2 uses a different for-mulation than IRI for the electron density distributionbelow the F2-peak, so the comparison for these two param-eters has to be considered only qualitative.

During both periods of low and high solar activity, afterpreliminary considerations about the availability of datafor both ionosondes, one month has been chosen as repre-sentative for every season (equinox, summer and winter) todiscuss the diurnal and seasonal characteristics of bottomside profiles (and associated parameters NmF2, hmF2, B0

and B1), and to compare them with IRI-2012 andNeQuick2 ones. The selected months with the correspond-ing monthly average sunspot number are reported inTable 1. Selecting September for LSA and March forHSA to represent the equinox season, we assume that dif-ferences/asymmetries occurring between these two monthsdo not alter the results of the analyses. Furthermore, only

nd high solar activity periods.

High solar activity

Month Mean sunspot number

March 2014 129July – August 2014 100 – 107December 2013 124

300 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

quiet geomagnetic conditions have been taken intoaccount, selecting for each month the 10 geomagnetic qui-etest days, by virtue of the geomagnetic information fromthe GFZ German Research Centre for Geosciences. Specif-ically, the quietest days are chosen on the basis of criterialinked to the geomagnetic Kp index; a detailed descriptionof how the quietest/most disturbed days are selected andrelated data are both available at http://www.gfz-potsdam.de/en/section/earths-magnetic-field/data-products-services/kp-index/qd-days/. Finally, the comparison betweenionosonde measurements, and IRI-2012 and NeQuick2outputs, for the parameters NmF2, hmF2, B0 and B1, iscarried out calculating averages over the 10 quietest daysselected for every month.

Fig. 2. Bottom side electron density profiles from ionosonde (black), IRI-2012for low solar activity (LSA, on the left) and high solar activity (HSA, on the rigseason for the 07 LT (first column for LSA and HSA), 13 LT (central columnrefers is also indicated on the top side of every panel. The x axis for the electronfigure legend, the reader is referred to the web version of this article.)

3. Results and discussion

3.1. Bottom side electron density profiles

Figs. 2 and 3 show examples of bottom side electrondensity profiles from ionosonde, IRI-2012 and NeQuick2for SJ and TU, respectively. In these figures, the panelson the left side show the results for low solar activity(LSA) while the panels on the right side display the resultsfor high solar activity (HSA). The first row reportsequinoctial profiles (September 2008 and March 2014),the second row reports winter profiles (June 2008 andAugust 2014) and the third row reports summer profiles(December 2008 and December 2013). After choosing a

(red) and NeQuick2 (blue) for the station of Sao Jose dos Campos. Profilesht) for equinoctial (first row), winter (second row) and summer (third row)) and 19 LT (right column) are displayed. The data at which every profiledensity is logarithmic. (For interpretation of the references to colour in this

Fig. 3. Same as Fig. 2 for the station of Tucuman.

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 301

definite day for every season, three different hours, 07 LT(left column), 13 LT (central column) and 19 LT (right col-umn), have been plotted for both LSA and HSA. The samedays are displayed for both stations.

Considering preliminary inter-station differences thatwill be deeply analyzed in Section 3.2.1, the ionosonde pro-files show comparable variations with the hour of the day,season and solar activity for SJ and TU. Under the effect ofthe EIA, more specifically under the effect of the pre-reversal enhancement, at the crests of the anomaly the dailymaximum of the peak electron density is expected aroundsunset hours (19 LT) (Heelis, 2004). Venkatesh et al.(2014b) found the following characteristics for the EIA inthe Brazilian sector: (1) the strength of the anomaly crestis more pronounced for HSA than for LSA; (2) it reachesthe maximum strength during equinox for HSA; (3) theEIA is not well developed in winter; (4) it is clearlyobserved in summer for HSA and only barely visible dur-

ing equinoctial and summer seasons for LSA. At SJ sta-tion, for HSA in summer and equinox, a comparablevalue of NmF2 is reached at 13 LT and 19 LT. At TU,higher values at 19 LT than 13 LT are observed duringequinox (for both LSA and HSA) and in winter forHSA. In general, comparable electron density values areobserved for the two stations during winter and summerseasons, for both LSA and HSA. On the contrary, relevantdifferences appear for HSA during equinox, for whichmore pronounced electron density values are measured atTU, in particular at 13 and 19 LT.

A comparison of electron density profiles from iono-sonde with the ones inferred by IRI-2012 and NeQuick2shows differences between observations and models. Betteragreements with ionosonde profiles are observed for LSAthan HSA, with the best agreement in particular for LSAat 07 LT. It is appreciable how both IRI-2012 andNeQuick2 underestimate the measured values of the peak

302 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

electron density over SJ and TU at 19 LT during equinoxfor LSA. For the same season, for HSA, the NmF2 valuemeasured at both stations is slightly underestimated byIRI-2012, while NeQuick2 provides a good correspondenceat SJ and an underestimation at TU. During summer sea-son, both IRI and NeQuick2 underestimate the parameterNmF2 at 07 LT for LSA at both stations. Moreover, it isworth noting that, for both SJ and TU both models pro-vide a clear hmF2 underestimation at 13 LT for LSA andat 13 and 19 LT for HSA.

Sethi and Pandey (2001) compared the incoherent scat-ter observations of bottom side electron density profileswith the outputs of IRI-95 for midday hours (10:00–14:00LT), in summer, winter and equinox, for the solar maxi-mum year of 1981 at the low-latitude station of Arecibo(18.4�N, 66.7�W; Puerto Rico). They reported good corre-spondences in winter with an underestimation of the elec-tron density at the F1-layer height range. Furthermore,an IRI-95 overestimation is reported in summer and equi-nox at all heights. In the present study, Figs. 2 and 3 showthat the electron density is overestimated by IRI-2012around midday in winter for HSA at all heights, at bothionosonde stations. Both models show a tendency to over-estimate electron density values below heights of �250 kmat 13 LT for HSA and it is interesting to note thatNeQuick2 seems to work generally better than IRI-2012in this height range. Aggarwal et al. (1996) made a compar-ison between ionosonde bottom side profiles and IRI-90 atNew Delhi (28.6�N, 77.2�E; India), near the northern crestof the EIA. The comparison was extended to low, mid andhigh solar activity (years 1958–1959, 1964–1965 and 1968–1969), for equinox, summer and winter, considering pro-files around the local midday. Their results show an IRI-90 overestimation during equinox and summer in everycondition of solar activity; in winter, a good agreement isfound for LSA, with an underestimation for HSA. Consid-ering the profiles below �250 km at 13 LT, where the IRI-2012 model shows a tendency to the overestimation forboth SJ and TU, in particular for HSA, the results are inaccordance with those reported by Aggarwal et al. (1996).

Finally, profiles in Figs. 2 and 3 suggest that both mod-els provide better results for the very particular period oflow solar activity of the last minimum than for the highsolar activity of the recent solar maximum, with better per-formances for the SJ ionosonde. Preliminary considera-tions reported in this section about the comparisonbetween ionosonde bottom side profiles and IRI-2012and NeQuick2 outputs will be deeply examined in Sections3.2.2 and 3.3, by discussing also monthly average values ofsome parameters that characterize the bottom side electrondensity profile.

3.2. F-layer peak parameters NmF2 and hmF2

3.2.1. Pattern and inter-station comparisonIn order to study in detail the inter-station differences,

Figs. 4 and 5 display monthly averages of NmF2 and

hmF2, derived from SJ and TU ionosondes, with the corre-sponding standard deviations. The results for LSA are onthe left, while those for HSA are on the right. In everypanel, the season, the month and the year are shown. Inthe bottom side part of every panel, NmF2 and hmF2inter-station differences are displayed. The differences aredefined as DNmF2 = NmF2TU – NmF2SJ and DhmF2 =hmF2TU – hmF2SJ, being NmF2TU (or hmF2TU) andNmF2SJ (or hmF2SJ) monthly average values for TU andSJ, respectively.

As mentioned, SJ and TU are two stations located uponthe southern crest of the EIA at similar latitude and only 20degrees separated in longitude. For these reasons, it is veryinteresting to study whether the longitudinal differencesfound by Fagundes et al. (2016) under geomagnetic dis-turbed conditions are observed also under the quiet geo-magnetic conditions considered in this work. A priori,climatological patterns strongly affected by the EIA areexpected.

As expected, Fig. 4 shows higher NmF2 values for HSAthan for LSA, for both SJ and TU. This aspect is particu-larly clear during nighttime hours for equinox seasonwhere LSA/HSA excursion as high as �3.0 _s 1012 [m�3]and �2.5 _s 1012 [m�3] can be reached at TU and SJ, respec-tively. For LSA, higher and lower values at both stationsare observed in summer and winter respectively, confirmingthe findings by Ezquer et al. (2014) who reported that thesemi-annual anomaly was not observed during the lastsolar minimum at TU. Furthermore, a very similar diurnaltrend is observed at SJ and TU for the three seasons, withdifferences always within one standard deviation. On thecontrary, for HSA significant inter-station differences areobserved during equinox with no relevant differences insummer and winter. In detail, during equinox a daily min-imum at 06:00 LT and a pronounced maximum around21:00 LT are reached, with high values that persist duringpre-midnight and nighttime hours until 02:00 LT at bothstations. Nevertheless, the most important feature is thathigher NmF2 values are observed at TU than at SJ, withsignificant differences registered between 18:00–23:00 LT,for which an average value of �1.5 _s 1012 [m�3] ismeasured.

The NmF2 maximum around 21:00 LT during equinoxfor HSA, observed both at SJ and TU, represents a typicalfeature of the EIA intensification due to the pre-reversalenhancement in the vertical drift. In fact, the maximumstrength of the EIA is observed during equinox for HSAin the South American sector. The pronounced inter-station differences found during this period suggest thatnear the crests of the EIA, different patterns in the electrondensity can be observed also for very close longitudinal sec-tors. The present study shows that the effect of EIAappears much more pronounced at TU than at SJ duringequinox for HSA, when the EIA reaches its maximumstrength. According to the results reported by Fagundeset al. (2016) during disturbed conditions, the present studyconfirms that the EIA effect is stronger in the West sector

Fig. 4. Comparison between monthly averages of NmF2 at SJ (in green, with standard deviation bars) and at TU (in orange, with standard deviationbars). Left and right column report values for LSA and HSA, respectively. The month and the season are displayed in every plot. For every plot, a bottomside panel shows the calculated inter–station hourly differences (in red, with associated error bars), defined as the TU value minus the SJ value (DNmF2 =NmF2TU – NmF2SJ). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 303

than in the East one in the South American region alsoduring quiet geomagnetic conditions.

The results for hmF2 are presented in Fig. 5 and show,as previously reported for NmF2, higher values for HSA

Fig. 5. Same as Fig. 4 but for hmF2.

304 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

than for LSA. During LSA, in winter and equinox, dailytrends are quite flat with values between 200 and 350 kmfor both stations. In these seasons, significant differences

appear only during winter between 22:00–06:00 LT withhigher values at TU and an average difference of �50km. During summer, the daily hmF2 pattern at SJ and

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 305

TU is similar, characterized by two minima around sunrise(07:00 LT) and sunset (18:00 LT), an increase during day-time and a daily peak around midday; significant inter-station differences are however present between 00:00–04:00 LT, with higher values at TU. For HSA, the follow-ing relevant differences have been found: (1) in winteraround midday lower values at TU than at SJ; (2) in sum-mer around 21:00–22:00 LT, higher values at TU than atSJ; (3) during equinox between 19:00–21:00 LT, an impul-sive hmF2 increase characterizing TU is not observed at SJ,with a peak difference of �84 km reached at 20:00 LT.

As previously mentioned in the results for NmF2, theEIA is particularly strong for HSA during equinox andmore pronounced at TU than at SJ. It is interesting to notethat during equinox the high values of NmF2 in the pre-midnight hours for HSA (Fig. 4) are linked with differenthmF2 patterns observed at the two stations: at SJ a slightlydecrease of hmF2 from daytime values is observed, with aconstant value during nighttime (hmF2 � 300 km); atTU, where the NmF2 increase is very pronounced, hmF2shows an impulsive increase that starts around 17:00 LT(hmF2 � 300 km), reaches a peak value (hmF2 � 430 km)in correspondence of the NmF2 peak at 21:00 LT, and thenstarts decreasing till around midnight. Therefore, at TU theEIA produces an increase of the local electron density asso-ciated with a kind of ‘‘elastic rebound” between 17:00–00:00 LT. This feature is not observed at SJ, where theeffect of the EIA is less pronounced.

Venkatesh et al. (2014a) found that at low latitudes andclose to the southern crest of EIA, for HSA, hmF2 showsmuch lower values in winter than during equinox and sum-mer. This feature is observed at both SJ and TU stations.Chuo (2012) studied the characteristics of NmF2, hmF2and B0 at Chung–Li (24.9�N, 121.1�E; Taiwan), close tothe northern crest of EIA, for the HSA year of 1999. It isvery interesting to note that hmF2 trends at Chung–Lishow a three-peaks behavior for all seasons, with relativemaxima occurring at pre-sunrise, local noon, and in theevening hours, with the first two being more pronouncedthan the third one. Similarly, Batista and Abdu (2004)observed that, for HSA, at Cachoeira Paulista (22.5�S,45�W; Brazil), close to the southern crest of EIA, hmF2shows three peaks at sun-rise (around 07:00 LT), pre-noon (around 11:00 LT) and sunset (around 20:00 LT),with the first one visible during all the year and probablycaused by the equatorward neutral wind. Bertoni et al.(2006) at SJ also find an hmF2 increase around sunrisefor the years 2003–2004. Comparing these results withthe trend shown in Fig. 5 for HSA, it is observed thatthe three-peak behavior is barely visible during equinoxand summer at TU, while it is not observed at SJ.

The observed inter-station differences, in particularthose observed during equinox for HSA, might be due toa joint action of the longitudinal gradients characterizingthe EEJ and the 5.8� separation in dip latitude betweenthe two ionosondes. The analysis of the longitudinal varia-tions of the EEJ has been the focus of a large number of

studies (e.g. Rastogi, 1962; Sastri, 1996; Luhr et al., 2004;England et al. 2006; Rastogi et al., 2007, 2013; Alken andMaus, 2007; Shume et al., 2010; Yizengaw et al., 2012,2014; Alken et al., 2013; Chandrasekhar et al., 2014;Moro et al., 2016; Yamazaki and Maute, 2016; Zhouet al., 2016), that have reported clear differences with lon-gitude, for both long- and short-distance separated sectors.In particular, the EEJ longitudinal variations in the com-plex South American sector have shown very particularbehaviors, as reported by Kane and Trivedi (1982, 1985),Shume et al. (2010), Yizengaw et al. (2014) and Moroet al. (2016). Kane and Trivedi (1982, 1985) comparedEEJ measures for two stations located in the South Amer-ican region, namely Huancayo (12�S, 75�W; Peru) andEusebio (4�S, 39�W; East Brazil), from October 1978 toSeptember 1979, for quiet geomagnetic conditions, andreported that the electrojet results to be stronger at Huan-cayo than at Eusebio. Shume et al. (2010) studied the char-acteristics of the EEJ at Jicamarca (12�S, 77�W; Peru) andSao Luıs (2.3�S, 44�W; Brazil) during a solar maximum(years 2001–2002) and a solar minimum (years 2006–2007) showing that: (1) the EEJ varies longitudinally, beingstronger on the West coast (Jicamarca) than in the Eastcoast (Sao Luıs), for both low and high solar activity con-ditions; (2) the EEJ has a maximum during equinoxes inJicamarca, but it has a prominent maximum in solstice sea-son in Sao Luıs; (3) the magnitude of the EEJ is more vari-able with season and solar cycle in Sao Luıs than inJicamarca. A similar study has been recently carried outby Moro et al. (2016) who studied the variations of theEY and EZ components of the Equatorial Electric Field(EEF) at Jicamarca and Sao Luıs for data collected duringthe period 2001–2010, under quiet geomagnetic conditions(Kp � 3+). The results revealed that EY ranges from 0.21to 0.35 mV/m at Sao Luıs and from 0.23 to 0.45 mV/mat Jicamarca, while the EZ component ranges from 7.09mV/m to 8.80 mV/m at Sao Luıs and from 9.00 mV/m to11.18 mV/m at Jicamarca. Therefore, Moro et al. (2016)show that the EEF varies longitudinally, being the EY

and EZ components less intense in the Brazilian sector thanin the Peruvian one. Finally, using couples of ground-basedmagnetometers and data from the Ion Velocity Meter(IVM) instrument onboard the Communication/Naviga-tion Outage Forecasting System (C/NOFS) satellite,Yizengaw et al. (2014) studied the behaviors of the EEJand E � B vertical plasma drift, for the period 2009–2013, for five longitudinal sectors: 76.9�W and 69.2�W(West South America), 56.1�W (Center–East South Amer-ica), 7.6�E (West African coast), 38.8�E (East Africancoast). The results are very interesting and show that: (1)the EEJ, and thus the vertical plasma drift, decreases inmagnitude from the American to African sector; (2) thelongitudinal EEJ and E � B drift distribution providehigher values in the West American sector and startdecreasing moving toward east; (3) the EEJ shows clearseasonal variations with higher magnitudes during equi-noxes and lower magnitudes in the June solstice. Further-

306 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

more, the study of Yizengaw et al. (2014) shows that, com-paring the E � B drift measured at 76.9�W and 56.1�W(only 20.8� separated in longitude in the South Americansector), clear differences during March equinox are visiblein the range 17:00–21:00 LT. In particular, the impulsiveenhancement observed at 76.9�W (peak value of �30 m/saround 19:00 LT) is much more pronounced comparedto the one observed at 56.1�W (peak value of �10 m/saround 18:30 LT). This is a key result that confirms howrelevant differences can be observed also for close longitu-dinal sectors in the EEJ strength, and consequently in theEIA, in the area of South America under quiet geomag-netic conditions.

By virtue of the aforementioned results, the strongerEIA effect observed at TU station for HSA during Marchequinox, can be in part attributed to a more pronouncedEEJ strength at TU, located 20� West of SJ, which mighthave led to a more pronounced E � B vertical plasma drift.Indeed, it is important to mention that, as shown in Fig. 1,TU and SJ are located at different geomagnetic field incli-nation isolines with a 5.8� separation in dip latitude, withisolines at SJ longitudes that are closer than those at TU.These different configurations can considerably affect thegeometry of the EIA crest over the 20� of longitudinalrange, giving rise to the diverse patterns obtained at SJand TU, in particular for HSA during equinox. Finally,it is worth noting that the different kind of ionosondes,CADI and AIS-INGV respectively working at SJ andTU, can lead by itself to differences linked to the accuracyof the two instruments. Furthermore, as mentioned, someloss of accuracy could interest the parameter hmF2 dueto the occurrence of E-valley that can influence the resultsof the inversion algorythm.

3.2.2. Comparison with IRI-2012 and NeQuick2

Hourly monthly averages of NmF2 and hmF2 fromionosonde and models are presented in Figs. 6 and 7,respectively. Ionosonde data are shown in black with thecorresponding standard deviation as vertical bars. Redand blue lines represent monthly averages from IRI-2012and NeQuick2 respectively. For both Figs. 6 and 7, the lefthand side columns refer to SJ, while right columns refer toTU. Similarly to Figs. 2 and 3, the first row reportsequinoctial values, the central row winter values and thebottom row summer values. In Fig. 6, different scales forLSA and HSA are presented to have clear visibility ofthe electron density variations. The differences betweenionosonde values and model outputs are displayed, forevery plot, in a bottom side panel. The differences aredefined as the IRI-2012 value minus the ionosonde value(in red, DNmF2IRI = NmF2IRI � NmF2Ionosonde) and asthe NeQuick2 value minus the ionosonde value (in blue,DNmF2NeQuick2 = NmF2NeQuick2 � NmF2Ionosonde). There-fore, a positive/negative difference underlines a model overestimation/underestimation.

A comparison between ionosonde and model valuesshows that, for both SJ and TU, a good agreement in

NmF2 predictions is observed in winter for both LSAand HSA. During summer season, the good resultsobtained in winter are confirmed at SJ for both LSA andHSA and at TU for HSA. Nevertheless, at TU for LSA,both models show an anticipation of two hours in thedescription of the daily NmF2 peak that leads to a daytimeoverestimation, particularly pronounced in the range08:00–15:00 LT. During equinox, relevant differencesbetween ionosonde and model values are found. In detail,both models (1) overestimate ionosonde values for LSAduring daytime, between 14:00–17:00 LT at SJ and between08:00–17:00 at TU; (2) for HSA, an underestimation isobserved around 22:00–23:00 LT at SJ station; (3) at TUa relevant underestimation occurs for HSA, being particu-larly pronounced between 14:00–02:00 LT and much morepronounced than that observed at SJ. This last point repre-sents the most striking result because it points out how themodeling of the EIA is far from being well accomplished byboth models. It is also worth noting that the model under-estimation is larger at TU, where the ionosonde measures amore pronounced EIA than at SJ.

Batista and Abdu (2004) made a comparison betweenionosonde measurements and IRI-2001 (using URSI coef-ficients), considering foF2 values recorded at CachoeiraPaulista, for both LSA (from March 1996 to February1997) and HSA (from March 2000 to February 2001). Onlya partial agreement with the results obtained here can befound for HSA, in fact on the one hand it is confirmedthe underestimation made by IRI in equinox during night-time, but on the other hand the overestimation made byboth models reported by Batista and Abdu (2004) for win-ter season around 06:00 LT is not confirmed. De Jesuset al. (2011) made a comparison between IRI-2007 andionosonde foF2 values recorded at SJ, for the year 2003,finding good agreements during all seasons, in particularfor the spring months of September and October (equinox).The results we obtain are partially in accordance with thosereported by de Jesus et al. (2011), being characterized bydifferences in particular during equinox for both LSAand HSA. At the northern crest of EIA, Chuo (2012) foundsignificant underestimations made by IRI-2007 between12:00–22:00 LT during equinox and winter for HSA. Thisbehavior is partially observed during equinox at TU.Ezquer et al. (2014) made a comparison between foF2 fromionosonde and IRI-2012 outputs for the very low solaractivity years of the last minimum 2008–2009 at TU. Theyfound an evident overestimation given by the model from08:00 LT to 15:00 LT in summer, which is in agreementwith the results reported in this study.

A comparison between hmF2 ionosonde data and modeloutputs, shown in Fig. 7, evidences that, according toBatista and Abdu (2004), the models work better forLSA than for HSA, both at SJ and TU. Focusing on theresults for SJ, the models underestimate the uplift observedduring daytime hours for HSA. This feature is visible forall seasons and particularly pronounced in summerbetween 09:00–18:00 LT. These results are in accordance

Fig. 6. NmF2 monthly averages from ionosonde (black, with standard deviation bars), IRI–2012 (red) and NeQuick2 (blue), for the station of SJ (on theleft) and TU (on the right). For every station, values for LSA and HSA at equinox (row on the top), in winter (central row) and in summer (row at thebottom) are displayed. For every panel the month is highlighted. Note that the electron density scale is different for LSA and HSA. For every plot, abottom side panel shows the calculated hourly differences between ionosonde and model values, defined as the IRI-2012 value minus the ionosonde value(red, DNmF2IRI = NmF2IRI – NmF2Ionosonde) and as the NeQuick2 value minus the ionosonde value (blue, DNmF2NeQuick2 = NmF2NeQuick2 –NmF2Ionosonde). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 307

with those of de Jesus et al. (2011), who compared theionosonde measured hpF2 values with IRI-2001 and IRI-2007 hmF2 outputs at SJ for the high solar activity year2003, and reported a particularly pronounced underestima-tion made by both versions of IRI during daytime hours insummer. For LSA at SJ, the models provide a good agree-ment with ionosonde data for both equinox and winter. Insummer, it is observed an underestimation made by bothmodels between 11:00–17:00 LT. At TU, both models pro-vide very good performances, in particular during winterseason (for both LSA and HSA). Nevertheless, at TU bothmodels provide underestimations in some particular cases.As previously pointed out, at TU for HSA in equinoctialseason, an impulsive hmF2 increase/decrease is observedbetween 17:00 LT and 00:00 LT. Both models do not fol-low this enhancement, providing underestimations between19:00–21:00 LT.

Table 2 summarizes the differences between ionosondevalues and model outputs for NmF2, hmF2, B0 and B1.The model reliability is emphasized through a point-to-point relative percentage ‘‘distance” defined as (|AModel �Aionosonde|/Aionosonde)�100, with A representing one of thefour aforementioned parameters. The table reports LSA,HSA and ‘Overall’ percentage average distances, obtainedaveraging over the 24 h and over September, June andDecember 2008 for LSA, over March 2014, July/August2014 and December 2013 for HSA, and over all the sixmonths for the ‘Overall’ value. Of course, a good modelreliability is indicated by a low percentage value.

Comparing the results given by models for the peak elec-tron density, slightly better estimations are made by IRI-2012 than NeQuick2, for both SJ and TU. Moreover, aver-age overall percentages of 33% (IRI) and 56% (NeQuick2)have been obtained for SJ, and 42% (IRI) and 58%

Fig. 7. Same as Fig. 6 for hmF2. In this case the unit of the y axis is the same for all panels.

Table 2Relative percentage average distances between ionosonde and model values. The single percentage distance is calculated as(|AModel � Aionosonde|/Aionosonde)�100, where A can be one of the parameters NmF2, hmF2, B0 and B1. LSA (or HSA) averages have been obtainedaveraging over the 24 h and over the three considered months (September, June and December 2008 for LSA; March 2014, July/August 2014 andDecember 2013 for HSA). Overall averages have been calculated in the same manner but averaging over all the six considered months.

Station Sao Jose dos Campos Tucuman

Parameter/Model IRI–2012 NeQuick2 IRI–2012 NeQuick2

LSA HSA Overall LSA HSA Overall LSA HSA Overall LSA HSA Overall

NmF2 29% 37% 33% 60% 53% 56% 44% 40% 42% 87% 30% 58%

hmF2 11% 14% 13% 9% 12% 10% 7% 13% 11% 7% 7% 7%

B0 17% 21% 19% 26% 11% 19% 17% 18% 17% 25% 16% 20%

B1 33% 23% 28% 34% 46% 40% 44% 20% 32% 31% 37% 34%

308 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

(NeQuick2) for TU, confirming the better performancesobtained at SJ by using IRI. For the parameter hmF2,Table 2 points out that: (1) NeQuick2 works slightly betterthan IRI for both stations and for both LSA and HSA; (2)better agreements are obtained for LSA than for HSA; (3)contrary to what obtained for NmF2, better performanceshave been obtained at TU.

In general, our results for the peak parameters suggestthat the IRI-2012 description of the bottom side profile(Eqs. (1) and (2)) appears more accurate than theNeQuick2 one (Eqs. (3) and (4)), with the peak electrondensity NmF2 representation that needs to be improvedin NeQuick2. Differences between IRI-2012 and NeQuick2in terms of hmF2 are not particularly relevant, the models

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 309

provide similar results. However, further comparisons atboth low- and mid-latitude sites are recommended in orderto have an improved knowledge of the F2-layer peakparameters description made by the two models.

3.3. Bottom side thickness (B0) and shape (B1) parameters

The comparison of average diurnal variations of thebottom side profile thickness and shape parameters, B0

and B1, are reported in Fig. 8 and Fig. 9 respectively. Bothfigures are structured in the same way of Figs. 5 and 6, withresults for SJ on the left and for TU on the right. Values forequinox (first row), winter (second row) and summer (thirdrow), for LSA (left column) and HSA (right column), aredisplayed for every station. A bottom side panel displaysthe measured differences between ionosonde and modelvalues for every plot.

It is seen from Fig. 8 that B0 daily variations derivedfrom ionosonde show a similar pattern at SJ and TU. A

Fig. 8. Same as F

pre-noon maximum around 10:00 LT is observed duringequinox and summer, both for LSA and HSA. The noon-time maximum of B0 is attributed to the upward plasmaflow at the equator (Chuo, 2012), and appears more pro-nounced in summer. A flattened trend characterizes thewinter season for HSA at SJ and for both LSA and HSAat TU. Sethi et al. (2009) compared the parameters B0

and B1 measured at New Delhi (28.6�N, 77.2�E; India) withthe outputs of IRI-2007 for LSA (January 2004–August2006). They found a clear increase (20% to 80%) of themeasured B0 from LSA to HSA during winter and equinox,while in summer the increase was within 30%. In the pre-sent study, a tendency to increase is observed during equi-nox and summer at TU, while at SJ, the B0 values increasefrom LSA to HSA only during nighttime hours with areverse pattern during daytime. Moreover, accordinglywith Sethi et al. (2009), a slight dependence of B0 valueson season is observed, with higher values in summer andlower in winter.

ig. 7 for B0.

Fig. 9. Same as Fig. 7 for B1.

310 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

Interesting analyses about a possible correlationbetween B0 and hmF2 variations have been carried outby Sethi et al. (2007), Chuo (2012) and Zhang et al.(2008). Sethi et al. (2007) made a comparison between bot-tom side profile parameters recorded by the ionosonde ofNew Delhi and IRI-2001 for the HSA years 2001–2002.They found a good correlation (linear correlation coeffi-cient of 0.7) between hmF2 and B0 in winter and equinoxduring daytime hours (06:00–18:00 LT). Chuo (2012) alsofound a correspondence between hmF2 and B0 for all sea-sons at the northern crest of the EIA. Zhang et al. (2008)found a strong positive linear correlation between hmF2and B0 during daytime hours at Hainan station (19.4�N,109.0�E; China) for mid/high solar activity. The impor-tance of this analysis can be found in the implications thata high correlation between hmF2 and B0 can have, provid-ing the possibility to obtain synthetic B0 values directlyfrom hmF2, which is easier to obtain from experimentaldata (Zhang et al., 2008).

The B0 pattern observed for HSA is in good agreementwith the results of Chuo (2012) and a correlation can beperceived between hmF2 and B0 at SJ and TU for bothLSA and HSA. Nevertheless, unlike those reported bySethi et al. (2007), between 06:00–18:00 LT, the presentstudy reveals a good linear correlation at SJ for all seasonsduring HSA (0.87 at equinox, 0.57 in winter and 0.51 insummer), while at TU good linear correlations are foundfor HSA during equinox (0.69) and summer (0.85).Fig. 10 shows the linear correlation between B0 andhmF2 for both stations, for all seasons during HSA,between 06:00–18:00 LT. The possible reason behind therelationship between hmF2 and B0 could be find in a‘‘cross-action” of equatorward neutral winds and EIA.Southward/equatorward neutral winds can induce anupward plasma drift and, under the action of the upwardvelocity, a simultaneous F-layer uplift (increasing ofhmF2) and B0 increase can occur. Fejer et al. (1995), study-ing the global plasma drift during daytime hours at equato-

Fig. 10. B0 monthly averages as a function of hmF2 at SJ (left column) and at TU (right column) for all seasons during HSA between 06:00–18:00 LT. Redlines highlight the linear fit. In every panel the month, the year and the linear correlation coefficient (LCC) are displayed. The y scale is the same for allpanels; the x scale is different for equinox, winter and summer season. (For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 311

rial latitudes, found that an upward plasma flow can occurduring 06:00–09:00 LT in equinox and summer, with alonger duration (until 13:00 LT) in winter. This resultcan explain the B0 pattern observed by Chuo (2012) during06:00–13:00 LT, indicating that the vertical plasma driftscan play an important role concerning the B0 features inthe EIA region. On the other hand, the EIA and in partic-ular the EEJ should be taken into account. Obrou et al.(2003) found that both hmF2 and B0 have a strong linearpositive correlation with the EEJ strength around middayhours in the equatorial West African sector. This resultconfirms what was found by Abdu et al. (1990) in the SouthAmerican sector, proofing how the electric field E thatdrives the EEJ plays a major role in the variations of boththe thickness parameter B0 and the peak electron densityheight hmF2. However, further studies about the relation-ship between hmF2 and B0, in particular for low-latitudeionosondes, are necessary to have a clear description ofthe phenomenon and its driving physical mechanisms.

A comparison of experimental B0 with the outputs ofIRI-2012 and NeQuick2 shows that, for both stations, agood agreement with IRI is obtained, in particular duringequinox. In winter, an IRI overestimation is visible forHSA during daytime hours (09:00–17:00 LT) at both sta-tions, with NeQuick2 that provides better agreements thanIRI. These results are in accord with those reported by

Venkatesh and Fagundes (2016), while they differ fromthe ones reported by Sethi et al. (2007) and Chuo (2012)who compared ionosonde values with both B0 ‘‘standard”option and Gulyaeva method (Gulyaeva, 1987) for the bot-tom side profile given as output by the IRI model. In sum-mer, weak correspondences with models are found. At SJ,both models underestimate the ionosonde values aroundthe daily peak for LSA; for HSA, NeQuick2 provides betteragreements with ionosonde data than IRI which signifi-cantly overestimate observed values between 11:00–17:00LT. For the same season at TU, both models stronglyunderestimate the daily peak for HSA occurring at around10:00 LT, while better correspondences are obtained duringnighttime hours. Using data from Cachoeira Paulista,Batista and Abdu (2004) found that the IRI amplitude ofthe B0 daily variation is much smaller than those estimatedfrom observations, with larger discrepancies during summermonths of HSA, which results in a tendency to underesti-mate the daily maximum around midday or overestimatethe nighttime values. Our results confirm this pattern atboth stations only in summer for LSA. For HSA the max-imum daily peak is described very well at SJ and just slightlyunderestimated at TU. In accord with Sethi et al. (2007),good results are obtained during nighttime hours.

In general, the qualitative values provided by theNeQuick2 model underestimate the ionosonde values at

312 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

SJ and TU for LSA, while better correspondences than IRIcan be found at SJ for HSA. Instead, at TU the IRI modelprovides very good results for LSA and for HSA duringequinox and summer months, respectively. The overall per-centage average distances between ionosonde and modeledvalues reported in Table 2 reveal that IRI and NeQuick2provide the same results (�19%) at SJ, while at TU IRIprovides better results than NeQuick2. However, it is inter-esting to observe that IRI provides better results for LSAthan for HSA, while the contrary is observed for theNeQuick2 model.

The diurnal variations of the shape parameter B1 aredisplayed in Fig. 9. Comparing the values of B1 obtainedfrom ionosonde profiles recorded at SJ and TU, significantinter-station differences are obtained for HSA, while trendsare quite similar for LSA. In detail, at SJ, B1 values showsmall differences between nighttime and daytime values inequinox and winter for HSA; in summer, a decreasebetween 09:00–12:00 LT modifies the flat trend observedfor the remaining hours of the day. At TU the results aredifferent and show a similar pattern for all seasons charac-terized by high values of B1 (between 3 and 4) during night-time hours, with a decrease and consequent low values(between 1 and 2) during daytime hours (08:00–17:00 LT).

In contrast to the results of Sethi et al. (2007, 2009), thecomparison between measured values and IRI-2012 andNeQuick2 outputs shows that both models do not catchcorrectly the observed ionosonde trends of B1. The qualita-tive values of B1 obtained by fitting NeQuick2 profiles withEq. (1) generally underestimate the ionosonde values, inparticular at SJ during HSA where differences larger than2 can be reached during nighttime hours. ConcerningIRI-2012, good correspondences with ionosonde valuesare obtained only in winter for HSA at TU. On the con-trary, the most relevant differences between ionosonde val-ues and IRI outputs are registered: (1) during equinox andwinter seasons for LSA at both stations, where IRI pro-vides relevant overestimations; (2) for HSA in summer,where a clear underestimation is observed during all dayat SJ and between 19:00–07:00 LT at TU.

As shown in Table 2, the overall percentage distancesfor the parameter B1, 28% (IRI) and 40% (NeQuick2) atSJ, and 32% (IRI) and 34% (NeQuick2) at TU, revealhow IRI-2012 provides better results than NeQuick2 forB1. However, despite the qualitative B1 values providedby NeQuick2, it is interesting to underline that, during day-time hours of LSA at TU and during all day of LSA in win-ter season for both SJ and TU, correspondences betweenNeQuick2 and ionosonde values are better than thoseobserved for IRI.

4. Conclusions

Bottom side electron density profiles at two closely sep-arated stations (at similar latitude and 20� separated in lon-gitude) located at the southern crest of the EIA in theSouth American sector, namely Sao Jose dos Campos

(23.1�S, 314.5�E, dip latitude 19.8�S; Brazil) and Tucuman(26.9�S, 294.6�E, dip latitude 14.0�S; Argentina), have beencompared among them and with the outputs of the IRI-2012 and NeQuick2 models. The analysis was carried outselecting a representative month for equinoctial, winterand summer seasons during the very low solar activity year2008 (sunspot annual average R = 3) and the high solaractivity years 2013–2014 (R = 65 and R = 113 respec-tively), under quiet geomagnetic conditions.

A careful investigation of the diurnal and seasonal vari-ations of the F-layer peak parameters NmF2 and hmF2 atboth sites reveals corresponding higher values for HSAthan for LSA. Concerning NmF2, results show that: (1)the EIA is at its maximum strength during equinox seasonfor HSA; (2) pronounced inter-station differences arefound during this period, with the EIA effect that is morepronounced at TU than at SJ. Furthermore, at TU thenighttime high values of NmF2 (peak around 21:00 LT)during equinox for HSA are linked to an impulsive con-temporary increase of hmF2, showing a kind of ‘‘elasticrebound” of the ionosphere.

The significant inter-station difference found in the EIAeffect suggests that relevant longitudinal-dependent varia-tions in the EIA can occur. These differences can be prop-erly explained considering that: (1) the EEJ in the SouthAmerican sector is characterized by a significant longitudi-nal gradient, with pronounced variations also for very closelongitudinal sectors; (2) concerning the dip latitude, there isa difference of 5.8� between the two ionosondes, which areconsequently characterized by different geomagnetic fieldconfigurations. It is important to underline that relevantdifferences in the EIA effect are found under quiet geomag-netic conditions for two so close ground-based ionosondesin South America for the first time. Therefore, further stud-ies in this sector on ionospheric characteristics recorded atclose longitude locations nearby the southern crest of theEIA gain significant importance.

IRI-2012 and NeQuick2 bottom side electron densityprofiles show significant deviations from the ionosondeobservations. The best agreements are found around sun-rise (07 LT), with larger differences around noon (13 LT)and sunset (19 LT). A better correspondence has beenobserved at SJ than at TU, and better results are obtainedfor LSA than for HSA. In general, the IRI-2012 modelshows better agreements with the observations than theNeQuick2 model. Nevertheless, NeQuick2 provides betterresults in the height range around 250 km at 13 LT, theregion at which both models overestimate the ionosondeelectron density.

Concerning the comparison between ionosonde andmodels for NmF2, relevant differences are observed duringequinox when the following pattern is observed for bothstations: (1) for LSA both models overestimate daytimevalues; (2) for HSA both models tend to underestimatenighttime values, with very pronounced differences atTU. Therefore, large discrepancies between ionosonde val-ues and model outputs are observed when the EIA is at its

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 313

maximum strength. Moreover, the IRI-2012 model pro-vides more reliable results than NeQuick2 for both sta-tions, with better correspondences with ionosonde datafor LSA than for HSA. The hmF2 parameter is describedquite well for LSA. For HSA larger differences with modelpredictions have been observed in particular at SJ, whereboth models underestimate the uplift observed during day-time hours. Moreover, both models do not follow theimpulsive increase of hmF2 observed at TU for HSA dur-ing equinox around post-sunset hours, providing conse-quently a clear underestimation. Accordingly to theresults obtained for NmF2, both models provide betterresults for LSA, but in this case, the NeQuick2 model pro-vides more reliable results than IRI for both stations.

SJ and TU show a similar diurnal pattern for the param-eter B0: (1) a pre-noon peak value around 10:00 LTobserved during equinox and summer, for both LSA andHSA; (2) a peak around noon that is more pronouncedin summer; (3) a moderate seasonal dependence withhigher values in summer and lower values in winter. Con-cerning B1, significant inter-station differences are obtainedfor HSA, while for LSA the trends are similar.

Comparing measured B0 values with those given bymodels, it is observed a good agreement with IRI-2012,in particular during the equinoctial season (for both sta-tions and for both LSA and HSA). Further, both IRI-2012 and NeQuick2 do not provide good results for theshape parameter B1, with IRI that can be considered morereliable than NeQuick2, as for NmF2 and B0, for bothstations.

Differences in model performances have to be ascribedto the corresponding different description used to modelthe bottom side electron density profile. In general, theIRI-2012 description, for ionosondes located around thesouth crest of the EIA in the South American sector,appears more appropriate than the semi-Epstein layerdescription used by NeQuick2.

Acknowledgments

P.R. Fagundes expresses his sincere thanks to the Fun-dacao de Amparo a Pesquisa do Estado de Sao Paulo(FAPESP), for providing financial support through theprocess № 2012/08445-9 and № 2013/17380-0, CNPq grant№ 302927/2013-1, and FINEP № 01.100661-00 for the par-tial financial support. K. Venkatesh expresses his sincerethanks to FAPESP, for providing financial supportthrough the process № 2013/17380-0.

Authors express their sincere thanks to the UNIVAP forproviding ionograms recorded at SJC, and to the Labora-torio de Ionosfera of the Universidad Nacional de Tucu-man and INGV for providing ionograms recorded at TU.Authors express their acknowledgements to GSFC, NASAfor providing online version of the IRI-2012 model (http://omniweb.gsfc.nasa.gov/vitmo/iri2012_vitmo.html), and tothe Telecommunications for Development (T/ICT4D)Laboratory of the Abdus Salam International Centre for

Theoretical Physics, Trieste, Italy, for providing theNeQuick2 model (http://t-ict4d.ictp.it/nequick2/nequick-2-web-model).

References

Abdu, M.A., Walker, G.O., Reddy, B.M., Sobral, J.H.A., Fejer, B.G.,Kikuchi, T., Trivedi, N.B., Szuszczewicz, E.P., 1990. Electric fieldversus neutral wind control of the equatorial anomaly under quiet anddisturbed condition: a global perspective from SUNDIAL 86. Ann.Geophys. 8 (6), 419–430.

Abdu, M.A., Brum, C.G.M., Batista, I.S., Sobral, J.H.A., de Paula, E.R.,Souza, J.R., 2008. Solar flux effects on equatorial ionization anomalyand total electron content over Brazil: observational results versus IRIrepresentations. Adv. Space Res. 42, 617–625. https://doi.org/10.1016/j.asr.2007.09.043.

Aggarwal, S., Venkatachari, R., Sachdeva, V.P., Jain, V.C., 1996.Comparison of electron density profiles for Delhi with correspondingprofiles obtained from IRI-90. Adv. Space Res. 18 (6), 39–40. https://doi.org/10.1016/0273-1177(95), 00896-9.

Alken, P., Maus, S., 2007. Spatio-temporal characterization of theequatorial electrojet from CHAMP, Ørsted, and SAC-C satellitemagnetic measurements. J. Geophys. Res. 112, A09305. https://doi.org/10.1029/2007JA012524.

Alken, P., Chulliat, A., Maus, S., 2013. Longitudinal and seasonalstructure of the ionospheric equatorial electric field. J. Geophys. Res.Space Phys. 118, 1298–1305. https://doi.org/10.1029/2012JA018314.

Altadill, D., Torta, J., Blanch, E., 2009. Proposal of new models of thebottom-side B0 and B1 parameters for IRI. Adv. Space Res. 43, 1825–1834. https://doi.org/10.1016/j.asr.2008.08.014.

Appleton, E.V., 1946. Two anomalies in the ionosphere. Nature 157, 691.Araujo-Pradere, E.A., Redmon, R., Fedrizzi, M., Viereck, R., Fuller-

Rowell, T.J., 2011. Some characteristics of the ionospheric behaviorduring the solar cycle 23–24 minimum. Sol. Phys. 274, 439–456.https://doi.org/10.1007/s11207-011-9728-3.

Araujo-Pradere, E.A., Buresova, D., Fuller-Rowell, D.J., Fuller-Rowell,T.J., 2013. Initial results of the evaluation of IRI hmF2 performancefor minima 22–23 and 23–24. Adv. Space Res. 51 (4), 630–638. https://doi.org/10.1016/j.asr.2012.02.010.

Asmare, Y., Kassa, T., Nigussie, M., 2014. Validation of IRI-2012 TECmodel over Ethiopia during solar minimum (2009) and solar maximum(2013) phases. Adv. Space Res. 53, 1582–1594. https://doi.org/10.1016/j.asr.2014.02.017.

Balan, N., Iyer, K.N., 1983. Equatorial anomaly in ionospheric electroncontent and its relation to dynamo currents. J. Geophys. Res. 88(A12), 10259–10262.

Batista, I.S., Abdu, M.A., 2004. Ionospheric variability at Brazilian lowand equatorial latitudes: comparison between observations and IRImodel. Adv. Space Res. 34 (9), 1894–1900. https://doi.org/10.1016/j.asr.2004.04.012.

Bertoni, F., Sahai, Y., Lima, W.L.C., Fagundes, P.R., Pillat, V.G.,Becker-Guedes, F., Abalde, J.R., 2006. IRI-2001 model predictionscompared with ionospheric data observed at Brazilian low latitudestations. Ann. Geophys. 24 (8), 2191–2200.

Bidaine, B., Warnant, R., 2010. Assessment of the NeQuick model at mid–latitudes using GNSS TEC and ionosonde data. Adv. Space Res. 45(9), 1122–1128. https://doi.org/10.1016/j.asr.2009.10.010.

Bilitza, D., Rawer, K., Bossy, L., Kutiev, I., Oyama, K.-I., Leitinger, R.,Kazimirovsky, E., 1990. International Reference Ionosphere 1990.NSSDC 90-22, Greenbelt, Maryl., 53, 160, https://doi.org/10.1017/CBO9781107415324.004.

Bilitza, D., Hernandez-Pajares, M., Juan, J.M., Sanz, J., 1998. Compar-ison between IRI and GPS–IGS derived electron content during 1991–97. Phys. Chem. Earth 24 (4), 311–319. https://doi.org/10.1016/S1464-1917(99), 00004-5.

Bilitza, D., 2001. International Reference Ionosphere 2000. Radio Sci. 36,261–275. https://doi.org/10.1029/2000RS002432.

314 L. Perna et al. / Advances in Space Research 61 (2018) 295–315

Bilitza, D., Reinisch, B.W., 2008. International Reference Ionosphere2007: improvements and new parameters. Adv. Space Res. 42, 599–609. https://doi.org/10.1016/j.asr.2007.07.048.

Bilitza, D., Brown, S.A., Wang, M.Y., Souza, J.R., Roddy, P., 2012.Measurements and IRI model predictions during the recent solarminimum. J. Atmos. Sol. Terr. Phys. 86, 99–106. https://doi.org/10.1016/j.jastp.2012.06.010.

Bilitza, D., Altadill, D., Zhang, Y., Mertens, C., Truhlik, V., Richards, P.,McKinnell, L.-A., Reinisch, B., 2014. The International ReferenceIonosphere 2012 – a model of international collaboration. J. Sp.Weather Sp. Clim. 4, A07. https://doi.org/10.1051/swsc/2014004.

Bilitza, D., Altadill, D., Truhlik, V., Shubin, V., Galkin, I., Reinisch, B.,Huang, X., 2017. International Reference Ionosphere 2016: fromionospheric climate to real-time weather predictions. Space. Weath. 15(2), 418–429. https://doi.org/10.1002/2016sw001593.

Chandrasekhar, N.P., Arora, K., Nagarajan, N., 2014. Evidence of shortspatial variability of the equatorial electrojet at close longitudinalseparation. Earth Planets Space 66 (1), 1 http://www.earth-planets-space.com/content/66/1/110.

Chen, H., Liu, L., Wan, W., Ning, B., Lei, J., 2006. A comparative studyof the bottomside profile parameters over Wuhan with IRI-2001 for1999–2004. Earth Planets Space 58, 601–605.

Chen, Y., Liu, L., Wan, W., 2011. Does the F10.7 index correctly describesolar EUV flux during the deep solar minimum of 2007–2009? J.Geophys. Res. Sp. Phys. 116, 1–6. https://doi.org/10.1029/2010JA016301.

Chen, Y., Liu, L., Wan, W., 2012. The discrepancy in solar EUV-proxycorrelations on solar cycle and solar rotation timescales and itsmanifestation in the ionosphere. J. Geophys. Res. Sp. Phys. 117, 2–13.https://doi.org/10.1029/2011JA017224.

Chuo, Y.J., 2012. Variations of ionospheric profile parameters duringsolar maximum and comparison with IRI-2007 over Chung-Li,Taiwan. Ann. Geophys. 30, 1249–1257. https://doi.org/10.5194/an-geo-30-1249-2012.

Coley, W.R., Heelis, R.A., Hairston, M.R., Earle, G.D., Perdue, M.D.,Power, R.A., Harmon, L.L., Holt, B.J., Lippincott, C.R., 2010. Iontemperature and density relationships measured by CINDI from theC/NOFS spacecraft during solar minimum. J. Geophys. Res. 115,A02313. https://doi.org/10.1029/2009JA014665.

de Jesus, R., Sahai, Y., Guarnieri, F.L., Fagundes, P.R., de Abreu, A.J.,Pillat, V.G., Lima, W.L.C., 2011. F-region ionospheric parametersobserved in the equatorial and low latitude regions during mediumsolar activity in the Brazilian sector and comparison with the IRI-2007model results. Adv. Space Res. 47, 718–728. https://doi.org/10.1016/j.asr.2010.10.020.

Di Giovanni, G., Radicella, S.M., 1990. An analytical model of theelectron density profile in the ionosphere. Adv. Sp. Res. 10 (11), 27–30.

Dudeney, J.R., 1978. An improved model of the variation of electronconcentration with height in the ionosphere. J. Atm. Terr. Phys. 40,195–203.

Dudeney, J.R., 1983. The accuracy of simple methods for determiningthe height of the maximum electron concentration of the F2-layerfrom scaled ionospheric characteristics. J. Atm. Terr. Phys. 45, 629–640.

Duncan, R.A., 1960. The equatorial F-region of the ionosphere. J. Atmos.Terr. Phys. 18, 89–100.

England, S.L., Maus, S., Immel, T.J., Mende, S.B., 2006. Longitudinalvariation of the E-region electric fields caused by atmospheric tides.Geophys. Res. Lett. 33, L21105. https://doi.org/10.1029/2006GL027465.

Ezquer, R.G., Lopez, J.L., Scida, L.A., Cabrera, M.A., Zolesi, B.,Bianchi, C., Pezzopane, M., Zuccheretti, E., Mosert, M., 2014.Behaviour of ionospheric magnitudes of F2 region over Tucumanduring a deep solar minimum and comparison with the IRI 2012 modelpredictions. J. Atmos. Terr. Phys. 107, 89–98.

Fagundes, P.R., Cardoso, F.A., Fejer, B.G., Venkatesh, K., Ribeiro, B.A.G., Pillat, V.G., 2016. Positive and negative GPS-TEC ionosphericstorm effects during the extreme space weather event of March 2015

over the Brazilian sector. J. Geophys. Res. Space Phys. 121, 5613–5625. https://doi.org/10.1002/2015JA022214.

Fejer, B.G., de Paula, E.R., Heelis, R.A., Hanson, W.B., 1995. Globalequatorial ionosphere vertical plasma drifts measured by the AE-Esatellite. J. Geophys. Res. 100, 5769–5776.

Grant, I.F., MacDougall, J.W., Ruohoniemi, J.M., Bristow, W.A., Sofko,G.J., Koehler, J.A., Danskin, D., Andre, D., 1995. Comparison ofplasma flow velocities determined by the ionosonde Doppler drifttechnique, SuperDARN radars, and patch motion. Radio Sci. 30,1537–1549. https://doi.org/10.1029/95RS00831.

Gulyaeva, T.L., 1987. Progress in ionospheric informatic based on electrondensity profile analysis of ionograms. Adv. Space Res. 7 (2), 51–60.

Heelis, R.A., 2004. Electrodynamics in the low and middle latitudeionosphere: a tutorial. J. Atmos. Sol. Terr. Phys. 66, 825–838. https://doi.org/10.1016/j.jastp.2004.01.034.

Heelis, R.A., Coley, W.R., Burrell, A.G., Hairston, M.R., Earle, G.D.,Perdue, M.D., Power, R.A., Harmon, L.L., Holt, B.J., Lippincott, C.R., 2009. Behavior of the O+–H+ transition height during the extremesolar minimum of 2008. Geophys. Res. Lett. 36, L00C03. https://doi.org/10.1029/2009GL038652.

International Radio Consultative Committee (CCIR), 1967a. Atlas ofIonospheric Characteristics. Report 340, International Telecommuni-cation Union, Geneva.

International Radio Consultative Committee (CCIR), 1967b. Atlas ofIonospheric Characteristics. Report 340–2 (and later suppl.), Interna-tional Telecommunication Union, Geneva.

ITU, 2003. Ionospheric Propagation Data and Prediction MethodsRequired for the Design of Satellite Services and Systems. Recom-mendation, Geneva, pp. 531–537.

Jodogne, J.C., Nebdi, H., Warnant, R., 2005. GPS TEC and ITEC fromdigisonde data compared with NEQUICK model. Adv. Radio Sci. 2(11), 269–273. https://doi.org/10.5194/ars-2-269-2004.

Kane, R.P., Trivedi, N.B., 1982. Comparison of equatorial electrojetcharacteristics at Huancayo and Eusebio (Fortaleza) in the southAmerican region. J. Atmos. Terr. Phys. 44, 785–792.

Kane, R.P., Trivedi, N.B., 1985. Equatorial electrojet movements atHuancayo and Eusebio (Fortaleza) on selected quiet days. J. Geomag.Geoelect. 37, 1–9. https://doi.org/10.5636/jgg.37.1.

Lee, C.C., Reinisch, B.W., Su, S.Y., Chen, W.S., 2008. Quiet–timevariations of F2-layer parameters at Jicamarca and comparison withIRI–2001 during solar minimum. J. Atmos. Sol. Terr. Phys. 70 (1),184–192. https://doi.org/10.1016/j.jastp.2007.10.008.

Lee, C.C., Reinisch, B.W., 2012. Variations in equatorial F2-layerparameters and comparison with IRI-2007 during a deep solarminimum. J. Atmos. Sol. Terr. Phys. 74, 217–223. https://doi.org/10.1016/j.jastp.2011.11.002.

Liu, L., Chen, Y., Le, H., Kurkin, V.I., Polekh, N.M., Lee, C.-C., 2011.The ionosphere under extremely prolonged low solar activity. J.Geophys. Res. 116, A04320. https://doi.org/10.1029/2010JA016296.

Luhr, H., Maus, S., Rother, M., 2004. Noon-time equatorial electrojet: Itsspatial features as determined by the CHAMP satellite. J. Geophys.Res. 109, A01306. https://doi.org/10.1029/2002JA009656.

Luhr, H., Xiong, C., 2010. The IRI2007 model overestimates electrondensity during the 23/24 solar minimum. Geophys. Res. Lett. 37,L23101. https://doi.org/10.1029/2010GL045430.

Lyon, A.J., Thomas, L., 1963. The F2-region equatorial anomaly in theAfrican, American and East Asian sectors during sunspot maximum. J.Atmos. Terr. Phys. 25, 373–386.

MacDougall, J.W., 1969. The equatorial ionospheric anomaly and theequatorial electrojet. Rad. Sci. 4 (9), 805–810.

MacDougall, J.W., Halland, G.E., Hayashi, K., 1997. F-region gravitywaves in the central polar cap, J. Geophys. Res. 102, 14513–14530.

Martyn, D.F., 1955. The Physics of the Ionosphere. Physical Society,London, p. 254.

Moro, J., Denardini, C.M., Resende, L.C.A., Chen, S.S., Schuch, N.J.,2016. Equatorial E region electric fields at the dip equator: 1.Variabilities in eastern Brazil and Peru. J. Geophys. Res. Space Phys.121. https://doi.org/10.1002/2016JA022751.

L. Perna et al. / Advances in Space Research 61 (2018) 295–315 315

Nava, B., Coısson, P., Radicella, S.M., 2008. A new version of theNeQuick ionosphere electron density model. J. Atmos. Sol. Terr. Phys.70 (15), 1856–1862. https://doi.org/10.1016/j.jastp.2008.01.015.

Obrou, O.K., Radicella, S.M., Adeniyi, J.O., 2003. The equatorialelectrojet and the profile parameters B0 and B1 around midday. J.Atmos. Sol. Terr. Phys. 65 (3), 299–304. https://doi.org/10.1016/S1364-6826(02)00336-X.

Phillips, T., 2009. Deep Solar Minimum [WWW Document]. NASA Sci.http://science.nasa.gov/science-news/science-at-nasa/2009/01apr_deepsolarminimum.

Radicella, S.M., Zhang, M.L., 1995. The improved DGR analytical modelof electron density height profile and total electron content in theionosphere. Ann. Geofis. XXXVIII 1, 35–41.

Ramakrishnan, S., Rawer, K., 1972. Model electron density profilesobtained by empirical procedures, in Space Research XII. In:Proceedings of the Fourteen Plenary Meeting, Seattle, Wash, GermanDemocratic Republic, Akademie, Berlin, 18 June–2 July 1971, pp.1253–1259.

Rastogi, R.G., 1962. Longitudinal variation in the equatorial electrojet. J.Atmos. Terr. Phys. 24, 1031–1040.

Rastogi, R.G., Chandra, H., Chakrabarty, D., Kitamura, K., Yumoto,K., 2007. Day-to-day correlation of equatorial electrojet at twostations separated by 2000 km. Ann. Geophys. 25 (4), 875–880.

Rastogi, R.G., Chandra, H., Shah, R., Trivedi, N.B., Fontes, S.L., 2013.A comparison of equatorial electrojet in Peru and east Brazil. OpenAtmos. Sci. J. 7, 29–36.

Rawer, K., Bilitza, D., Ramakrishnan, S., 1978. International referenceionosphere 78, Special Report, International Union of Radio Science(URSI), Brussels, Belgium.

Rawer, K., 1982. Replacement of the present sub-peak plasma densityprofile by a unique expression. Adv. Sp. Res. 2 (10), 183–190.

Reinisch, B.W., Huang, X., 1996. Low latitude digisonde measurementsand comparison with IRI. Adv. Space Res. 18 (6), 5–12. https://doi.org/10.1016/0273-1177(95), 00892-6.

Rush, C.M., Fox, M., Bilitza, D., Davies, K., McNamara, L., Stewart, F.G., PoKempner, M., 1989. Ionospheric mapping–an update of foF2coefficients. Telecommun. J. 56, 179–182.

Sastri, J.H., 1996. Longitudinal dependence of equatorial F region verticalplasma drifts in the dusk sector. J. Geophys. Res. 101 (A2), 2445–2452.

Sethi, N.K., Pandey, V.K., 2001. Comparative study of electron densityfrom incoherent scatter measurements at Arecibo with the IRI-95model during solar maximum. Ann. Geophys. 18, 1630–1634. https://doi.org/10.1007/s00585-001-1630-0.

Sethi, N., Dabas, R.S., Das, R.D., 2007. Diurnal and seasonal variation ofB0, B1 parameters during high solar activity period at low mid-latitudeand their comparisons with IRI-2001 model. J. Atmos. Sol. Terr. Phys.69, 767–774. https://doi.org/10.1016/j.jastp.2007.01.008.

Sethi, N., Dabas, R.S., Upadhayaya, A., 2009. Midday bottomsideelectron density profiles during moderate solar activity and compar-ison with IRI-2001. Adv. Space Res. 43, 973–983. https://doi.org/10.1016/j.asr.2008.08.007.

Shume, E.B., Denardini, C.M., De Paula, E.R., Trivedi, N.B., 2010.Variabilities of the equatorial electrojet in Brazil and Peru. J. Geophys.Res. Space Phys. 115, A06306. https://doi.org/10.1029/2009JA014984.

Stening, R.J., 1992. Modeling the low-latitude F-region, a review. J.Atmos. Solar Terr. Phys. 54, 1387–1412. https://doi.org/10.1016/0021-9169(92)90147-D.

Titheridge, J.E., 1985. Ionogram analysis with generalized programPOLAN, Rep. UAG-93, World Data Cent. for Sol.–Terr. Phys.,Boulder, Colorado.

Venkatesh, K., Fagundes, P.R., de Jesus, R., De Abreu, A.J., Pillat, V.G.,Sumod, S.G., 2014a. Assessment of IRI-2012 profile parameters bycomparison with the ones inferred using NeQuick2, ionosonde andFORMOSAT-1 data during the high solar activity over Brazilianequatorial and low latitude Sector. J. Atmos. Solar Terr. Phys. 121,10–23. https://doi.org/10.1016/j.jastp.2014.09.014.

Venkatesh, K., Fagundes, P.R., Seemala, G.K., Jesus, R.D., Abreu, A.J.,Pillat, V.G., 2014b. On the performance of the IRI-2012 andNeQuick2 models during the increasing phase of the unusual 24thsolar cycle in the Brazilian equatorial and low-latitude sectors. J.Geophys. Res. Sp. Phys. 119 (6), 5087–5105. https://doi.org/10.1002/2014JA019960.

Venkatesh, K., Fagundes, P.R., 2016. Bottom-side profile parameters (B0,B1) characteristics over the Brazilian equatorial and low latitudes andtheir comparison with different options in the IRI-2012 model duringthe 24th solar minimum (2010–2011). J. Atmos. Sol. Terr. Phys. 146,16–27. https://doi.org/10.1016/j.jastp.2016.05.004.

Wakai, N., Ohyama, H., Koizumi, T., 1987. Manual of Ionogram Scaling.Radio Research Laboratory Ministry of Posts and Telecommunica-tions, Japan.

Yamazaki, Y., Maute, A., 2016. Sq and EEJ-a review on the dailyvariation of the geomagnetic field caused by ionospheric dynamocurrents. Space Sci. Rev. https://doi.org/10.1007/s11214-016-0282-z.

Yizengaw, E., Zesta, E., Moldwin, M.B., Damtie, B., Mebrahtu, A.,Valladares, C.E., Pfaff, R.F., 2012. Longitudinal differences ofionospheric vertical density distribution and equatorial electrodynam-ics. J. Geophys. Res. Space Phys. 117, A7312. https://doi.org/10.1029/2011JA017454.

Yizengaw, E., Moldwin, M.B., Zesta, E., Biouele, C.M., Damtie, B.,Mebrahtu, A., Rablu, B., Valladares, C.F., Stoneback, R.A., 2014.The longitudinal variability of equatorial electrojet and vertical driftvelocity in the African and American sectors. Ann. Geophys. 32, 231–238. https://doi.org/10.5194/angeo-32-231-2014.

Zakharenkova, I., Krankowski, A., Bilitza, D., Cherniak, Y., Shagimu-ratov, I., Sieradzki, R., 2013. Comparative study of foF2 measure-ments with IRI-+2007 model predictions during extended solarminimum. Adv. Space Res. 51 (1), 620–629. https://doi.org/10.1016/j.asr.2011.11.015.

Zhang, M.L., Shi, J.K., Wang, X., Wu, S.Z., Zhang, S.R., 2004.Comparative study of ionospheric characteristic parameters obtainedby DPS-4 digisonde with IRI2000 for low latitude station in China.Adv. Space Res. 33, 869–873. https://doi.org/10.1016/j.asr.2003.07.013.

Zhang, M.L., Wan, W., Liu, L., Shi, J.K., 2008. Variability of thebehavior of the bottomside (B0, B1) parameters obtained from theground-based ionograms at China’s low latitude station. Adv. SpaceRes. 42, 695–702. https://doi.org/10.1016/j.asr.2007.07.022.

Zhou, Y.L., Luhr, H., Alken, P., Xiong, C., 2016. New perspectives onequatorial electrojet tidal characteristics derived from the Swarmconstellation. J. Geophys. Res. Space Phys. 121, 7226–7237. https://doi.org/10.1002/2016JA022713.

Zuccheretti, E., Tutone, G., Sciacca, U., Bianchi, C., Arokiasamy, B.J.,2003. The new AIS-INGV digital ionosonde. Ann. Geophys, 46.