CSES is the first satellite of Chinese space-based geophysical field observation system, which is aimed for the application prospects in earthquake and near-space science. It was launched into an altitude of about 507 km, with a nearly Sun-synchronous orbit of 98° inclination, and the ascending/descending node is at about 02:00/14:00 LT ( Shen et al., 2018). The Langmuir probe (LAP) provides in situ Ne covering a range from 5×10²–1×10⁷ cm⁻³, and Te from 500 k to 10,000 K. Previous studies have confirmed the reliability of CSES measured Ne and Te, that show generally consistent global distributions with earlier satellites, e.g., DEMTER and Swarm (e.g., Liu et al., 2018; Yan et al., 2020; Yan et al., 2022).

The Swarm mission, consisting of three identical satellites (A, B, C), was launched on 22 November 2013 into nearly polar, circular orbits, in altitude of approximately 490 km. In the final constellation, the Swarm A and C are flying side-by-side (longitudinal separation of 1.4°) at about 460 km, while the Swarm B is flying by about 50 km higher (Lühr et al., 2015; Xiong et al., 2016a). One key instrument of Swarm satellites is the Electric Field Instrument (EFI) (Knudsen et al., 2017), consisting of a pair of Thermal Ion Imagers (TIIs), and a pair of LPs. The LPs provide the Ne and Te with a time resolution of 2 Hz. Quite good agreements have been found between Swarm measured Ne/Te and other techniques/model predictions, suggesting a good reliability of the Swarm plasma dataset (Lomidze et al., 2018; Palin et al., 2015; Pedatella et al., 2015). As the Swarm C is at the same altitude of Swarm A and their Ne measurements are very consistent with each other (Xiong et al., 2016b), the results from Swarm C are not shown in this study. In addition, only the data with good quality are selected, with the flags are satellite potential (Vs.), Ne and Te being set to 20, and flag of LPs being set to 1.

The CHAMP satellite was launched on 15 July 2000 into a nearly-polar (inclination = 87.3°), circular orbit with an initial altitude of about 450 km (Reigber et al., 2002). When the mission ended with the reentry into the atmosphere on 19 September 2010, its altitude decayed to about 150 km. The onboard Planar Langmuir Probe (PLP) provided the in situ Ne and Te measurements for every 15 s. In-flight calibration of the PLP instrument has been performed and CHAMP has been confirmed to provide reliable plasma data. The details of the calibration and the retrieval of electron density and temperature from the PLP observations are given by McNamara et al. (2007).

Figure 1A shows the variation of altitude, local time from CSES and Swarm A/B satellites in 2019 and 2020. As there is no overlap between the CSES and CHAMP missions, we considered only the CHAMP data in 2008 and 2009 with low solar activity (Figure 1B), which is similar to the solar activity during CSES operation period. The solar flux in 2008–2009 and 2019–2020 are depicted in the first subpanel in Figures 1A, B. In order to reduce the influence of magnetic storm all data are selected under magnetically quiet conditions with Kp < 3.

For a fair comparison, we selected the simultaneous measurements from the CSES and Swarm satellites when they are close in both time and space. Three conjunction periods around 14:00 LT (marked with light-grey rectangles in the lower two panels of Figure 1A) are found between the two satellites, which are: Apr. 5–19, 2019, Aug. 22–31, 2019 and Jan. 10–23, 2020 for Swarm B and CSES, while Sep. 25- Oct. 10, 2019, May. 15- Jun. 4, 2019 and Feb. 5–22, 2020 for Swarm A and CSES, reflecting the periods at equinoxes, June solstice and December solstice (Figure 1A).

Considering CHAMP data gaps around 14:00 LT, data for the same season and local time in 2008 and 2009 were analyzed together so that sufficient data can be used for our comparison. To focus on 14:00 LT, the CHAMP data during Oct.5–21, 2008 are used for representing the situations of equinoxes; May. 28 - Jun. 18, 2008 and Jul. 1–11, 2009, as well as Feb. 25–28 & Nov. 11–14, 2009 are corresponding to June solstice and December solstice, respectively. These five conjunction periods between CSES and CHAMP have also been marked with light-grey rectangles (see Figure 1B) The CHAMP altitude in 2008 and 2009 is around 330 km (Stolle et al., 2011).

To sum up, we researched the correlation of Ne and Te at around LT 14:00 using satellites with different altitude (∼330km, ∼480 and ∼500 km separately). And at around 500 km, we can get the data at closest time and distance from two satellites (CSES and Swarm B) for mutual authentication.

Figure 2 presents the Ne/Te MLat profiles of CSES, Swarm B, Swarm A and CHAMP from top to bottom. Data have been separated into equinoxes, June solstice, and December solstice, as shown in three columns from left to right. For each satellite, the Ne/Te profiles during conjunction period are presented as black lines, and the red lines represent the average profile of each season. A common feature between all four satellites is the Ne peaks at the magnetic equator, corresponding to the well-known equatorial ionization anomaly (EIA), while the Te shows a clear valley above the magnetic equator, suggesting a negative relation between the Ne and Te.

Slight differences are also seen between the four satellites. For example, the double crests of EIA are more prominent from the Ne profiles of Swarm A and CHAMP, as they fly at a relative lower altitude; while for Swarm B and CSES, which fly at a similar altitude, the double crests are more evidenced from Swarm B. Such difference of Ne between CSES and Swarm B has been earlier reported by Yan et al. (2020), which might be caused by the lower absolute Ne values measured by CSES and different data inversion method. The EIA crests show similar seasonal variations from Swarm B, Swarm A and CHAMP, with the same magnitude in the two hemispheres during equinoxes but with stronger crest in the local summer hemisphere during two solstices. Such a seasonal variation of the inter-hemispheric asymmetry of EIA crests is caused by the trans-equatorial plasma flue along magnetic fluxtube, which has been well discussed by Xiong et al. (2013). Another interesting feature seen in Figure 2 is that the Te profiles from Swarm A and B seem to form two branches at the equatorial and low latitudes regardless of seasons, while such a feature cannot be seen from the CSES and CHAMP Te profiles.

Figure 3 repeats the average profiles of Ne/Te from the four satellites by comparing variations along MLat between CSES with another satellite. To better show the comparison, the average profiles of CSES are repeated as black lines in each subpanels, while the profiles from the other three satellites are shown as blue lines. They are shown in dual coordinate axis so that we better compare their trend along MLat although the different absolute values of Ne/Te. Similar to Figure 2, the subpanels from left to right depict the results in three different seasons. Very similar variations of Ne/Te are seen between CSES and the other three satellites, except the single or double crests of EIA that being attributed to satellite probing altitude. In addition, a feature seen from CHAMP is that the Te exhibits two minima at latitudes corresponding well to the EIA crests.

Figure 4 show the scatter distribution of Ne and Te around 14:00 LT from the four satellites (from top to bottom, CSES, Swarm B, Swarm A and CHAMP). The subpanels from left to right depict the results in equinoxes, June solstice, and December solstice. We have limited our analysis to the low and middle latitudes (±50° MLat) to avoid the much more complicate Ne/Te relation at auroral latitudes. The colors from blue to red represent the measurements at different MLat range. A general negative relation is found between the Ne and Te at 14:00 LT from all four satellites. However, the slope of the negative relation reduces and even becomes positive when the Ne becomes larger.

A feature for the two Swarm satellites is that the relation between Ne and Te is quite scatter when the Ne is low (less than 1×10¹¹ m⁻³). For Ne at a fixed level, the Te can vary from 2000 K to over 4000 K. Similar wild variation of Te is also seen from the CSES observation, with less data points which are confined to equinoxes and June solstice seasons. However, such wild variation of Te can’t be seen from the CHAMP observations. In addition, when the Ne is larger than 1×10¹¹ m⁻³, the relation between Ne and Te forms two branches for Swarm A and B irrespective of seasons, which should be caused by the two groups of Swarm Te profiles at the equatorial and low latitudes as shown in Figure 2.

For CHAMP, we see similar Ne/Te relation as that observed by CSES. Slight difference is that instead of the positive relation between Ne and Te, a saturation of Te is observed (or Te keeps at a same level) when Ne continues increases from 1×10^11.5 m⁻³. Such a saturation of Te is also evidenced when Ne is very low, especially during June solstice.

To check if there is a latitudinal and seasonal dependence of the Ne/Te relation, the data in different MLat ranges are presented separately, and the results from the four satellites are presented in Figures 5–8. For each satellite, the observations are further divided into three seasons.

It is obvious that the correlation of Ne and Te at middle MLat (30°N -50°N and 30°S-50°S) are mainly negative, while at low and equatorial latitudes (from 30°S to 50°N), both negative and positive relation between Ne and Te are found. The reduced negative or even positive relation between Ne and Te is only found when Ne is larger. Around 14:00 LT, the Ne at middle latitudes is usually lower than the values at low and equatorial latitudes, especially when the EIA structure is well developed.

In addition, seasonal influence can also be seen. In equinoxes the correlation of Ne and Te are symmetric with respect to the geomagnetic equator. In solstice seasons, the correlation of Ne and Te vary with seasons and has similar characteristics when in same season hemisphere. For example, at MLat range 10°S-30°S, the Ne and Te measured from CSES show positive correlation only during December solstice hemisphere. This is because the southern hemisphere is in summer during December solstice, and the Ne is larger in southern hemisphere due to large solar illumination. Similar trend between Ne and Te is found in the northern hemisphere (10°N-30°N), though with no clear positive relation observed, the slope of the negative Ne/Te relation becomes spread during local summer (June solstice). From this perspective, whether the Ne is negatively and positively correlated with Te, depends largely on the absolute value of Ne.

The relation between Ne and Te from Swarm satellites (Figures 6, 7) shows very similar latitudinal and seasonal dependences as that observed by CSES. However, additional two features need to be mentioned. First, the scatter of Ne/Te relation mainly happens at the low and middle latitudes (|MLat|>30°), as well as at MLat range 10°N-30°N in December solstice and 10°S-30°S in June solstice. Second, the two branches of Ne/Te relation happens mainly at equatorial and low latitudes (|MLat|<30°) when Ne is larger than 1×10¹¹ m⁻³. From the latitudinal and seasonal distributions, we see the scatter of Ne/Te relation is dominated by MLat, while forming two branches of Ne/Te relation is dominated by the magnitude of Te.

For CHAMP (Figure 8), we see similar seasonal and MLat dependences of Ne/Te relation as that observed by CSES. In addition, a saturation of Te is observed (or Te keeps at a same level) at equatorial and low latitudes (|MLat|<30°), and also in the southern middle latitude (30°S -50°S) during June solstice.

The intensity of Ne and Te shows clear seasonal and magnetic latitudinal dependences from the measurements of all four satellites. For example, the Ne is symmetric about the magnetic equator in the equinoxes, while with larger values shifted towards the summer hemisphere in solstice seasons (seen Figures 2, 3). Such seasonal and latitudinal variation of topside Ne is also consistent with pervious findings observed by other low earth orbiting satellites, e.g., the Gravity Recovery and Climate Experiment (GRACE) and DEMETER (Xiong et al., 2013; Zhang et al., 2014), as well as the total electron content (TEC) derived from the ground receivers of global navigation satellite system (GNSS) (e.g., Yeh et al., 2001). In addition, considering the altitude dependences, the Ne shows larger values and the two EIA crests are much more prominent from the satellites fly at lower altitude. This is well to be understood, as the EIA crests have stronger intensity at the F2 peak, and the four satellites fly above the F2 peak during almost all the time.

The Te in the low-latitude topside ionosphere is mainly controlled by the photoelectron heating and cooling through Coulomb collisions with ions that are both related to the integrated Ne. As shown in Figures 5–8, the Te in the same Ne range increases with increasing MLat. On the dayside around 14:00 LT, a clear negative correlation between Te and Ne is seen at low and middle latitudes; especially for the lower altitudes where Swarm A and CHAMP fly, two Te minima are observed right at the latitudes of EIA crests (Figure 2). As shown in Figure 4, no scatter distribution between Ne and Te is observed from the CHAMP satellite. Even though, for a fixed Ne value, the Te can vary with a range over 500 K. One possible reason could be that the Ne considered here is the in situ measurement along the satellite track. As pointed out by Kakinami et al. (2011a), Te is more closely related to the integrated Ne along magnetic fluxtube, and the in situ Ne does not completely correlate with the integrated Ne, for example, the meridional wind can transport the plasma from one hemisphere to the other (Kelley, 2009).

In addition, due to influences from the lower atmospheric tidal and waves forcing, the topside ionospheric exhibit different longitudinal patterns in different seasons. The most prominent patterns are the longitudinal wave-4 and wave-3 structures during August and December solstice, which is considered to be dominated by the non-migrating tidal component of DE3 and DE2, respectively (e.g., Immel et al., 2006; Wan et al., 2010; Lühr et al., 2012). These results suggest that even at a fixed local time, the Ne can vary with quite a large range at different longitudes during different seasons, so it will further affect the seasonal dependence of Ne and Te relation as shown in Figure 4.

One interesting feature seen from CHAMP is that the Te keeps almost at a constant level when Ne is very low (in the southern middle latitude, 30°S -50°S, during June solstice). Kakinami et al. (2011a) reported an “inverted U shape” of the relation between Ne and Te during 06:00–08:00 LT from the Hinotori satellite observations The inverted U shape shown in their Figure 1 arises from the positive correlation for lower Ne and a negative correlation for higher Ne, which looks very similar to the CHAMP results as shown in our Figure 8. And for both satellites, such low Ne values appear at middle latitudes.

Two prominent features of the Ne/Te relation observed by Swarm satellites are: 1) when Ne is larger than 1×10¹¹ m⁻³, Te are grouped into two branches at equatorial and low latitudes; 2) when Ne is lower than 1×10¹¹ m⁻³, Te sometimes becomes very scatter at the low and middle latitudes.

For the first feature, we repeated the Ne/Te relation from Swarm B, shown in Figure 9. In Figure 9A the observations during equinoxes, June solstice and December solstice are marked with green, red and blue. Te has been grouped into two branches when Ne is larger than 1×10¹¹ m⁻³, but no clear seasonal dependence of the two-branch feature of Te dependencies observed. For a fixed Ne level, the difference between the two Te branches is about 600 K. In addition, we checked further if the two branches of Swarm Te have geographical dependence, e.g. on longitude. As shown in Figure 9B, data have been divided into four geographic longitude bins: 180°– -90°, -90° 0°, 0° 90°, 90° 180° marked separately with red, green, blue and yellow. However, no clear longitude dependence is found of the two-branch feature.

From Figures 4, 9, the Ne value of 1×10¹¹ m⁻³ seems to be the threshold when the two branches of Te occur. However, as seen from Figures 4, 8, the Ne observed by CHAMP are sometimes also greater than 1×10¹¹ m⁻³, but the two branches of Te were not observed in the CHAMP data. So, it seems that the two branches of Swarm Te data are not directly related to the absolute value of Ne.

To have a further look at the two Te branches, we selected two orbits data both from Swarm B and CSES as examples, illustrated in Figure 10. Figures 10A, C show two latitudinal profile of Ne (first panel) and Te (second panel) from Swarm B marked as blue, while the Ne and Te measurements from CSES during this conjunction orbit are shown as black lines. The values of Ne/Te from Swarm B and CSES are depicted by the left and right y-axis, respectively. The CSES orbits are quite close to Swarm, and the near-simultaneous footprints are presented also as black lines and blue lines (Figures 10B, D). Among them, in the first example (Figures 10A, B), Ne and Te profiles from CSES and Swarm B agree quite well with each other, except the difference of absolute value of Ne. The Te measurements from both satellites are basically at the same range, and the Te keeps at a lower value of about 2000 K at the low and equatorial latitudes, corresponding to the lower Te branch in Figure 9. From the second example (Figures 10C, D), we can see another characteristic, the Te from Swarm B keeps almost a constant value of about 2500 K at low and equatorial latitudes, representing the upper branch in Figure 9. For both examples, we see clearly that the Te from CSES keeps a value of about 1700 K at the low and equatorial latitudes. We checked also the quality flag of Te from Swarm B (third panel in Figures 10A, C), and the parameter “Flag-Te” is always marked as 20 for both orbits, suggesting no obvious disturbances during the two orbits. Such a direct comparison suggests that the upper branch of Swarm Te might be biased by a value of 600 K, though we don’t know what causes the bias of the Swarm Te. The two-branch feature of Te seems exists in both hemispheres during all seasons, therefore, it might not be simply attributed to the illumination of the sunlight.

For the second feature, we see that the Te from Swarm B satellite shows clear fluctuations at the southern low and middle latitudes (15°S - 38°S), shown in Figures 10A, C. We also checked the flags of Swarm LP product, and the parameter “Flag-Te” is marked with a value of 20 during the Te fluctuations, indicating no abnormal behavior of Te data. However, we think such kind of Te fluctuations should not be related to natural geophysical structures, as the Te profile from CSES is quite smooth, especially considering their orbits are really close. Figure 9B shows also that there is no geophysical longitude dependence. In addition, we have further checked the Swarm orbits when such kind of Te fluctuations are observed and the parameter “Flag-Te” are also marked with a value of 20. When looking at the seasonal and latitudinal distributions of Ne/Te relation in Figures 6, 7, for both Swarm B and A, the scatter of Te mainly happens at low and middle latitudes during equinoxes and local winter, while they are almost absent during local summer.

As the scatter of Te mainly happens at the upper-middle latitudes, where is close to the subauroral ionosphere. At this region, the Te varies often in a quite wide range, as it can be fed by the heat conduction from the outer plasmasphere (Fok et al., 1991). Though we only considered the data during geomagnetically quiet condition (Kp < 3), shorter time-scale processes or disturbances in the solar wind and magnetosphere can trigger heat flow to the ionosphere. Another possible reason to alter Te variations could be the Subauroral Ion Drifts (SAID) driven by large polarization electric fields from the auroral oval towards lower latitude (Moffett et al., 1998). Further dedicated study is encouraged for resolving this feature in details.

Ne in the ionosphere is mainly produced by the solar EUV radiation, since the solar photons have significant energy to ionize the neutral atmosphere. Photoelectrons produce heat for the local electrons as well as remote electrons along the magnetic field line. Te is governed by the heating of photoelectron, cooling caused by collisions among ions and neutral species, and heat conduction in the ionosphere (e.g., Watanabe et al., 1995). In the topside ionosphere (above the F2 peak), the heat conduction becomes a dominant factor because of the reduced collision between charged particles and neutrals. Therefore, Te is not controlled by the local variation of Ne but rather by the integrated Ne along the magnetic field line below the F2 peak (Kakinami et al., 2011b). Above the magnetic equator, the plasma is transported upward by the E×B drift at the magnetic equator, so that the plasma for the EIA when diffusing poleward along the magnetic field line (e.g. Appleton, 1946). In addition, thermospheric winds also transport the plasma in the topside ionosphere upward/downward along the magnetic field line (Watanabe and Oyama, 1996; Kakinami et al., 2011b). Therefore, the positive dependence of Ne/Te at the EIA region is controlled mainly by the plasma dynamics. When the plasma is uplifted from below to the altitudes of interest, Ne increases, but Te is remains lower. When plasma is transported from higher altitudes, Te is higher while Ne is lower. In both cases the plasma transport results in negative Te/Ne dependence. The only case when we could expect a positive tendency is the uplifting of neutral gas along with the plasma. Then the photoionization will produce fresh photoelectrons which will heat the electron gas and increase Te.