Thermospheric Parameters during Ionospheric G-Conditions

: For the ﬁrst time thermospheric parameters (neutral composition, exospheric temperature and vertical plasma drift related to thermospheric winds) have been inferred for ionospheric G-conditions observed with Millstone Hill ISR on 11–13 September 2005; 13 June 2005, and 15 July 2012. The earlier developed method to extract a consistent set of thermospheric parameters from ionospheric observations has been revised to solve the problem in question. In particular CHAMP/STAR and GOCE neutral gas density observations were included into the retrieval process. It was found that G-condition days were distinguished by enhanced exospheric temperature and decreased by ~2 times of the column atomic oxygen abundance in a comparison to quiet reference days, the molecular nitrogen column abundance being practically unchanged. The inferred upward plasma drift corresponds to strong ~90 m/s equatorward thermospheric wind presumably related to strong auroral heating on G-condition days.


Introduction
Ionospheric F 2 -layer is the main layer in the Earth's ionosphere with maximum electron concentration N m F 2 , which is normally larger than in other ionospheric layers under various geophysical conditions. However, sometimes F 2 -layer disappears on the ground-based sounding ionograms due to its blanketing by underling F 1 -layer telling us that critical frequency f o F 2 becomes ≤ f o F 1 . According to the URSI handbook of ionogram interpretation and reduction [1], such situation is referred to as G-condition. G is just a descriptive letter indicating conditions of ground-based ionospheric sounding observations. During G-conditions no ionospheric information is available from the heights above F 1layer maximum normally located below 200 km. Therefore, any observations of the F 2 -layer under G-conditions are possible either with topside ionospheric sounding [2] or Incoherent Scatter Radars (ISR). In principle ionospheric radio-occultation (RO) observations provide the whole N e (h) profiles but a comparison with Millstone Hill ISR observations [3] has shown large disagreement in the bottom-side of N e (h), especially in summer when the occurrence probability of G-conditions is maximal ( [4], Figure 11). Moreover, G-conditions are mainly associated with the periods of geomagnetic disturbances ( [4], Figure 1) when large spatial gradients take place in the ionosphere, while RO method implies a spherical symmetry. Therefore, RO observations hardly can be used for analyses of G-conditions. Theoretically, rocket launching sonde observations can provide the whole N e (h) profile, but such experiments were conducted in the past in the beginning of the space era and we do not know any published results of such observations during G-conditions. Topside ionospheric sounding also has some limitations. For instance, when N m F 2 ≤ N m F 1 (such case is considered in our analysis) there is a large height range between two maxima which is invisible on the topside sounding ionograms. Moreover, satellite topside sounding observations do not allow the analysis of G-condition development in time at any particular location. Summarizing, one may conclude that ISR observation is the only source of information which may be used to analyze the G-condition formation mechanism. Along with this, a ground-based ionospheric sounding global network can be efficiently used for morphological analyses. For instance, Lobzin & Pavlov [4,5] have done a thorough and detailed morphological analysis of G-conditions using worldwide ground-based ionosonde observations over the 1957-1990 period. They have analyzed the occurrence probability (ψ G ) of G-conditions on season, local time, latitude, level of solar and geomagnetic activity. The ψ G dependence on geomagnetic activity presented by K p index is impressive-a wellpronounced linear dependence for log(ψ G ) versus K p was revealed. It was found that the occurrence probability ψ G strongly increased under large K p although G-conditions occur under low geomagnetic activity as well. The undertaken morphological analysis has shown that "the dependence of the G condition occurrence probability on K p is mainly determined by processes that control the behavior of the F 2 -layer with K p changes", while F 1 -layer only slightly reacts to geomagnetic disturbances. A weak reaction of F 1 -layer to geomagnetic activity was also stressed and analyzed from a physical point of view by Mikhailov & Schlegel [6].
Traditionally the formation of G-conditions in the ionospheric F 2 -layer is analyzed using model simulations. A good example of this approach may be found in the paper by Deminov et al. [7]. Using the MSISE00 thermospheric model [8] the authors have clearly shown how an increase of geomagnetic activity from K p = 0 to K p = 8 decreases N m F 2 leaving practically unchanged N m F 1 forming by this way G-condition in the ionosphere. Their model calculations were not related to any specific observations and give only a qualitative pattern of G-condition development.
An attempt to describe Millstone Hill ISR observations using the IZMIRAN model for quiet 23-25 June 1986 period when G-conditions took place during morning (07:32 LT) hours was undertaken by Pavlov & Buonsanto [9]. It was shown that neutral composition taken from the original MSIS-86 model [10] gives N e (h) profile which has nothing in common with the observed one. The authors arbitrarily changed model atomic oxygen [O] and molecular nitrogen [N 2 ] to get a better coincidence with the observed N e (h). However, the final coincidence is not impressive at all (Figure 3). Similar attempt was repeated later by Schlesier & Buonsanto [11] to describe G-condition observed at Millstone Hill on 11 April 1997 at 09:53 LT. They also varied model MSIS-86 ratios N 2 /O and O 2 /O in model simulations to get a better coincidence with the observed N e (h). The result is same: "none of the simulations give the correct electron density profile". These results just confirm a well-known postulate-thermospheric parameters (neutral composition, temperature, thermospheric winds producing vertical plasma drift) and ionizing solar EUV flux should constitute a self-consistent set which allows to describe the observed N e (h). Arbitrarily changing some of them is unproductive.
The aims of our analysis may be formulated as follows.

1.
Using Millstone Hill ISR observations to find cases with G-condition at various stages of its development: (a) when N m F 2 ≈ N m F 1 , (b) when F 2 -layer maximum is present but N m F 2 < N m F 1 , and (c) when F 2 -layer maximum is practically absent on the N e (h) profile.

2.
To retrieve a consistent set of the main aeronomic parameters responsible for the formation of daytime mid-latitude F-layer using the earlier developed method to extract thermospheric parameters from ionospheric observations [12]. To Include CHAllenging Minisatellite Payload (CHAMP) and Gravity field and steady state Ocean Circulation Explorer (GOCE) neutral gas density observations in the vicinity of Millstone Hill to the retrieval process to increase the reliability of the inferred thermospheric parameters.

3.
To discuss the role of neutral composition in the G-condition formation mechanism.

Observations, Method, and Results
Millstone Hill ISR observations were analyzed to find noontime cases of G-condition with different degree of its development. There are many obstacles in this. The selection is burdened by a necessity to have a reference (not a perturbed day) close in time to a selected one. This is not always possible, keeping in mind that ISR works episodically. Zenith alternating-code basic parameters observation with a 4 km height step is desirable for this type of analysis to have a sufficient number of points at F 1 -region heights, while standard zenith single-pulse observations provide N e (h) profiles with a 20 km height step, which is not a sufficient height resolution to analyze F 1 -region. Another restriction is related to satellite neutral gas density observations which should be included in the retrieval process. The selected dates of G-condition should coincide with available neutral gas density observations. Excellent CHAMP/STAR and GOCE neutral gas density (ρ) observations are used in our analysis. Therefore, the number of periods available for our consideration is limited to four cases: Sep  Our method [12] to retrieve thermospheric parameters from ionospheric observations requires observed noontime f o F 2 and plasma frequencies at 180 km height, f 180 for (10,11,12,13,14) LT; both may be taken from Millstone Hill Digisonde observations. The method is designed to work with routine ground-based ionosonde observations and it cannot be applied during G-conditions, when F 2 -layer maximum is not seen. Therefore, the method was changed to deal with the whole N e (h) profiles available from ISR observations. In addition to five f 180 values now we use observed N e at the upper boundary (normally 450-500 km) and a couple of points on the N e (h) profile controlling its shape. Median N e (h) ISR profiles calculated over a 2-h time internal around noontime are used in our analysis.
The revised method similar to the basic one [12] has two versions. The first uses only observed electron concentration as fitted parameters. The second version additionally uses observed neutral gas density as a fitted parameter. CHAMP and GOCE neutral gas density observations in the daytime American sector were reduced to the location of Millstone Hill and 12 LT using the MSISE00 thermospheric model and the following expression: During this reduction the height of ρ observation was kept unchanged not to introduce an additional uncertainty related to unknown MSISE00 neutral temperature T ex for the particular days in question. The inclusion of ρ observations into the retrieval process increases the reliability of the obtained results. In this case the retrieved neutral composition [O 2 ] concentrations) and temperature T ex along with vertical plasma drift W and total solar EUV ionizing flux not only describe the observed N e (h) but also match observed neutral gas density, which has nothing in common with retrieval process. Figure 1 gives an example of G-condition during the 11-13 September 2005 disturbed period, the quiet 7 September 2005 day being used as a reference. Normally during Gconditions the residual F 2 -layer also exists but it is blanked by underlying F 1 -layer and this is seen in Figure 1. However, cases with practically degenerated F 2 -layer maximum may also take place as on 13 June 2005 (see later). Figure 1 demonstrates that in accordance with the morphological analysis by Lobzin & Pavlov [5] G-condition is more likely to occur during the first half of a day. The first two disturbed days, September 11-12 are marked by increased daytime h m F 2 compared to the reference day 7 September 2005. On September 13 when the storm activity has decreased this difference in h m F 2 is not seen while N m F 1 remains larger than N m F 2 . This points out the disturbed neutral composition which still remains at Millstone Hill but the equatorward thermospheric wind is decreased (see later) during the recovery storm phase (see D st index in Figure 1). which still remains at Millstone Hill but the equatorward thermospheric wind is decreased (see later) during the recovery storm phase (see Dst index in Figure 1).  Figure 2 gives observed near noontime and calculated Ne(h) profiles for some of the selected G-condition and reference days. The retrieval method is designed to work above 150 km so the lower boundary for calculated Ne(h) is located at this height. All G-condition cases in Figure 2 manifest a strong decrease of electron concentration above F1-layer with a degraded F2-layer which may be well-pronounced like on 12 September 2005, but in other cases it looks like a hint of the F2-layer maximum. It is interesting to note that a very strong decrease of electron concentration at F2-layer heights took place on 13 June 2005, under a moderate level of geomagnetic activity (Ap = 32/54 nT) with a quiet previous period.  Figure 2 gives observed near noontime and calculated N e (h) profiles for some of the selected G-condition and reference days. The retrieval method is designed to work above 150 km so the lower boundary for calculated N e (h) is located at this height. All G-condition cases in Figure 2 manifest a strong decrease of electron concentration above F 1 -layer with a degraded F 2 -layer which may be well-pronounced like on 12 September 2005, but in other cases it looks like a hint of the F 2 -layer maximum. It is interesting to note that a very strong decrease of electron concentration at F 2 -layer heights took place on 13 June 2005, under a moderate level of geomagnetic activity (A p = 32/54 nT) with a quiet previous period.
This example demonstrates that G-conditions do associate to geomagnetic activity but the dependence is not that straight, as shown by Lobzin & Pavlov [4] and Deminov et al. [7]. Along with this, in accordance with previous analyses F 1 -layer manifests much less variations of electron concentration.
Results of the retrieval process for the days in question along with some other observations are given in Table 1. Firstly, it is seen that observed G-condition cases are not related to day-to-day changes in solar EUV radiation. Due to Solar Radiation and Climate Experiment (SORCE) mission and the Thermosphere, Mesosphere, Ionosphere, Energetic, and Dynamics (TIMED) mission daily EUV (100-1200) Å observations are available since 2002 [13]. These data were used to control day-to-day EUV variations for the periods in question. Both retrieved and observed EUV manifest very small variations if the disturbed and reference days are compared. This example demonstrates that G-conditions do associate to geomagnetic activity but the dependence is not that straight, as shown by Lobzin & Pavlov [4] and Deminov et al. [7]. Along with this, in accordance with previous analyses F1-layer manifests much less variations of electron concentration.
Results of the retrieval process for the days in question along with some other observations are given in Table 1. Firstly, it is seen that observed G-condition cases are not related to day-to-day changes in solar EUV radiation. Due to Solar Radiation and Climate Experiment (SORCE) mission and the Thermosphere, Mesosphere, Ionosphere, Energetic, and Dynamics (TIMED) mission daily EUV (100-1200) Å observations are available since 2002 [13]. These data were used to control day-to-day EUV variations for the periods in question. Both retrieved and observed EUV manifest very small variations if the disturbed and reference days are compared. All considered G-condition events occurred under magnetically disturbed conditions (see A p indices in Figure 2). According to present-day F 2 -layer storm concept [14][15][16][17][18][19][20][21][22] atomic oxygen concentration and exospheric temperature strongly change in the course of a storm and the mechanism of these changes is well-established. In accordance with this storm mechanism exospheric temperature should be larger and the atomic oxygen abundance should be less for disturbed days compared to reference ones. Atomic oxygen is totally produced and lost in the upper atmosphere [23] therefore we used column atomic density calculated above the level with the N 2 column density of 10 17 cm −2 [24]. For the analyzed cases this level is located at 140-145 km altitude. Column density is a convenient quantity which does not depend on the temperature height profile. Table 1 manifests features of the F 2 -layer storm mechanism: T ex is systematically larger and column atomic density is less by~2 times for G-condition days compared to reference ones. On the other hand, due to the difference in T ex , neutral gas density at a fixed height is systematically larger for disturbed days. The same occurs for the atomic oxygen concentration if we take large heights (355 km) where [O] provides the main contribution to ρ but the situation is inversed at lower height (261 km) where the [N 2 ] contribution to ρ is dominating. In accordance with the F 2 -layer storm concept, normal solar driven daytime northward thermospheric wind V nx should be inversed due to enhanced auroral heating (see AE index in Figure 1) during disturbed periods. Table 1 indicates strong upward plasma drift W = V nx sinIcosI (where I-magnetic inclination) on G-condition days and negative W which corresponds to northward V nx for quiet reference days.

Interpretation
G-condition effect in the ionospheric F 2 -region should be considered and can be explained in the framework of the present day F 2 -layer storm concept (see earlier cited references) considering its part related to negative storm phase. Disturbed neutral composition with a decreased atomic oxygen concentration formed in the auroral zone-due to heating of the thermosphere by magnetospheric electric fields and particle precipitations-is moved to middle latitudes by thermospheric winds resulted from the competition between solar-driven (background) and storm-induced circulations. The bulge of disturbed neutral composition with a decreased atomic oxygen concentration is pushed around by winds and may move back and forth in latitude [18]. Such effect was confirmed by the storm simulations [17] as well as by ESRO-4 data analysis [16]. The pattern strongly depends on season. In summer the [O]/[N 2 ] disturbance zone may extend all the way from the polar to the low latitudes while in winter it is only restricted to high latitudes [25,26]. This seasonal difference is due to different pattern of global thermospheric circulation during two seasons. The solar-driven circulation is equatorward at middle latitudes during nighttime, i.e., it coincides with the storm-induced one while it is poleward during daytime hours. Therefore, disturbed neutral composition flows in middle latitudes during nighttime and is pushed back during daytime. This may explain the morphological fact "that G-condition is more likely to occur during the first half of a day" [5].
G-condition cases considered in our analysis (Table 1) manifest strong equatorward V nx~9 0 m/s around noontime hours. This tells us about strong auroral heating during the selected days. However even such strong equatorward V nx could not overpower the decrease in electron concentration at F 2 -layer heights resulted from neutral composition changes. It should be stressed that namely the decrease in atomic oxygen abundance rather than in [O]/[N 2 ] distinguishes the G-condition days. This follows from ] column density give practically unchanged value~1.03 × 10 17 cm −2 for all dates in question. This is not surprising as molecular nitrogen is chemically inactive thermospheric species whose height distribution is close to a barometric one. Moreover, its distribution is not practically affected by eddy diffusion since molecular weight of N 2 is about the same as the average molecular weight of the mixed atmosphere (see e.g., [27]).
The effect of neutral composition changes is also seen at F 1 -region heights (Figure 2). Electron concentration for G-condition days is slightly less compared to reference ones. Using a simplified scheme of photo-chemical processes it is possible to write down an expression for electron concentration at F 1 -region heights [6]: where q(O + )-ion O + production rate, M + -average-weighted dissociative recombination rate coefficient. The first term in the right side of Equation (2) represents the O + ion concentration, while the second term-the concentration of M + ions. The main contribution to N e at F 1 -region heights provide M + ions especially during disturbed periods when atomic oxygen and correspondingly the first term in expression (2) is decreased. This explains a weak reaction of F 1 -layer to geomagnetic disturbances usually mentioned in the literature. On the contrary F 2 -layer is mainly composed of O + ions whose concentration is proportional to the atomic oxygen ion production rate q(O + ), i.e., to the concentration of atomic oxygen [O]. Table 1 indicates that column atomic oxygen abundance in the upper atmosphere is reduced by~2 times for G-condition days compared to reference ones. However, this [O] decrease is seen only at lower altitudes, such as 261 km on 15 July 2012, but not at larger altitudes, such as 355 km on 12 September 2005, due to larger T ex on disturbed days. Neutral gas density depending both on atomic oxygen and molecular nitrogen is always larger on G-condition days: due to [O] at large altitudes and due to [N 2 ] at lower heights.

Discussion
For the first time, aeronomic parameters responsible for the formation of G-conditions in the mid-latitude ionospheric F 2 -layer have been inferred from Millstone Hill ISR observations. It should be stressed that only ISR observations are able to provide the whole N e (h) profile which is necessary for such analysis. Our standard method [12] to retrieve a consistent set of aeronomic parameters from routine ground-based ionosonde observations has been changed specially to analyze G-conditions. The required F 2 -layer maximum may be absent during G-conditions, therefore three points with observed electron concentration, i.e., the upper boundary and two points controlling the N e (h) profile, were used in the method. Satellite CHAMP and GOCE neutral gas density observations were also introduced to the method. This is a very important component of the method. Mid-latitude daytime F 2 -layer electron density profile crucially depends on neutral composition (O, O 2 , and N 2 concentrations), which in their turn depend on neutral temperature T ex . The same parameters specify neutral gas density ρ at a given height. Simultaneous fitting observed N e (h) and ρ leaves not much room for an accidental selection of the thermospheric parameters. An additional 'nail' hammered in the method confirming its validity is a comparison of the retrieved to observed solar EUV ( Table 1). The correlation coefficient between two variations is 0.974 ± 0.06 which is significant at the 99.9% confidence level according to t-criterion. It should be stressed that the retrieved EUV flux has nothing in common with EUV observations [13]. Therefore, we may conclude that the retrieved thermospheric parameters manifest reality for the analyzed periods of G-conditions.
The results of our analysis tell us that G-conditions present a manifestation of strong F 2 -layer negative storm phase. We have revealed a strong decrease (by~2 times) of atomic oxygen abundance, presumably related to the transfer of disturbed neutral composition from the auroral zone. The presence of such transfer is confirmed by strong equatorward thermospheric wind obtained in our calculations for the days of G-conditions (Table 1). Normal northward solar-driven V nx (negative vertical plasma drift W) on reference days was inversed for positive one. Empirical models of disturbed thermospheric winds like [28] do not reproduce such strong (~90 m/s) equatorward noontime meridional wind.
It is commonly accepted that F 2 -layer negative storm phase is related to a decrease in the [O]/[N 2 ] ratio at F 2 -layer heights and this was confirmed by direct observations (e.g., [26]). However, the mechanism of this decrease should be specified. Molecular nitrogen, as discussed earlier, just follows neutral temperature variations being close to a barometric distribution. Neutral temperature increases during storm periods so [N 2 ] concentration also increases at a fixed height in the upper atmosphere but the column [N 2 ] density remains practically unchanged~1.03 × 10 17 cm −2 for the dates in Table 1. Contrary to this atomic oxygen manifests a real strong decrease of its column abundance during disturbed days in a comparison to reference ones (see Table 1). Therefore G-condition (as a manifestation of the F 2 -layer negative storm phase) is due to a decrease in the total atomic oxygen abundance in the upper atmosphere and an increase of neutral temperature during disturbed periods.
Of course, it would be interesting to compare our retrieved thermospheric parameters to MSISE00 model values as this or MSIS86 thermospheric models were earlier used to describe G-conditions in the F 2 -region [7,9,11]. Such a comparison is given in Table 2 for the days in question. The comparison is done at the height of a satellite where neutral gas density was observed. The main difference (~200 K) takes place in T ex for disturbed days while this difference is much less for quite reference days. This results in larger retrieved [N 2 ] and [O 2 ] concentrations mainly follow neutral temperature variations. However, inferred atomic oxygen manifests quite different variations. Despite larger T ex retrieved [O] are systematically (by 58% on average) less than MSISE00 model concentrations. This issue was discussed earlier when we mentioned the general decrease of the atomic oxygen abundance in the upper atmosphere during G-condition periods. For this reason, it is not a surprise that neither Pavlov & Buonsanto [9] nor Schlesier & Buonsanto [11] succeeded in trying to describe G-condition N e (h) profiles observed with Millstone Hill ISR using MSIS model.

Conclusions
The main results of our analysis may be summarized as follows.

1.
For the first time thermospheric parameters (neutral composition, exospheric temperature, and vertical plasma drift related to thermospheric winds) have been retrieved for ionospheric G-conditions observed with Millstone Hill ISR on 11-13 September 2005; 13 June 2005; and 15 July 2012.

2.
A revised method by Perrone & Mikhailov [12] has been used to retrieve a consistent set of the main aeronomic parameters responsible for the formation of daytime midlatitude F-region. Observed with CHAMP and GOCE satellites neutral gas density is used as a fitted parameter in this method. 3.
The analyzed G-condition cases are distinguished by enhanced exospheric temperature and decreased by~2 times of the column atomic oxygen abundance in a comparison to quiet reference days. Along with this the column abundance of molecular nitrogen remains practically unchanged. Therefore, the O/N 2 ratio decrease at F 2layer heights is totally due to an increase in T ex and to a decrease in the atomic oxygen abundance.

4.
The inferred upward plasma drift~30 m/s corresponds to strong~90 m/s equatorward thermospheric wind presumably related to strong auroral heating on Gcondition days. 5.
The inferred set of thermospheric parameters for G-condition days strongly differs from empirical model values therefore the observed N e (h) can hardly be described on the basis of such models.