The Comparison of Electron Density between CSES In Situ and Ground-Based Observations in China

Abstract

As the observation accuracy of parameters in the ionosphere cannot be directly checked, the comparison with other observations is the main way to evaluate the data quality of satellite measurements. Through the comparative analysis between the in situ electron density (Ne) observed by the China Seismo-Electromagnetic Satellite (CSES) and Ne at about 500 km altitude detected by Qujing Incoherent Scatter Radar (ISR), it was found that the pattern of CSES Ne is consistent with that of ISR Ne, and the correlation coefficient between the two sets of data is above 0.88 for different groups according to the magnitude. The value of CSES Ne is lower than that of ISR Ne, and the median value of the ratio for the difference between the conjugate data is 84.04%. Based on the comparison in the daytime between CSES Ne and ionosonde observations in China, it was found that the trend of the two datasets is mostly similar, and the correlation coefficient in some locations can reach up to 0.7. The distribution of CSES Ne and correlation coefficients at different latitudes show that the relationship is relatively better around the peak of the equatorial ionization anomaly (EIA). The differences in the value between CSES Ne and ionosonde data also exist, the relative change of which is about 80–95% in the daytime.

1. Introduction

The China Seismo-Electromagnetic Satellite (CSES), also called ZhangHeng-1 (ZH-1), was successfully launched on 2 February 2018. There are eight scientific payloads on the CSES, including a high-precision magnetometer (HPM), an electric field detector (EFD), a search coil magnetometer (SCM), a plasma analyzer package (PAP), a Langmuir probe (LAP), a high energetic particle package (HEPP) and high energetic particle detector (HEPD), a GNSS occultation receiver (GOR) and a tri-band beacon (TBB), which can provide ionospheric observations of electromagnetic fields, plasma and energetic particles [1].

As the accuracy of physical parameters in the ionosphere cannot be directly checked, the comparison with other observations or models is the main way to evaluate the quality of the satellite data. For the CSES data, Zhou et al. [2] found that the three HPM sensors have good consistency after in-orbit linear correction. The comparison of HPM data between CSES and Swarm shows that both the vector and the scalar differences are quite consistent, and the responses to a geomagnetic storm were similar for the two satellites. Diego et al. [3] reported that the observation of EFD highlights a very good agreement with the theoretical value, and can respond to both natural and artificial sources. The results of Zhao et al. [4] show that the electromagnetic field recorded by CSES EFD and SCM has a good correlation with the simulated result. The correlation coefficients of electron density (Ne) observed by the LAP payload between CSES and the Detection of Electro-Magnetic Emissions Transmitted from Earthquake Regions (DEMETER) satellite, Swarm constellation and the International Reference Ionosphere (IRI) model are all high, and the variation patterns also have similar trends between conjugate data [5,6,7]. By analyzing the data in the commissioning phase, Picozza et al. [8] attested that the particle rates of the HEPD proton and electron are consistent with radiation models, as well as their distribution along the orbit. Ambrosi et al. [9] found that not only the satellite housekeeping data, but also the scientific data of HEPD show very good stability. Wang et al. [10] found that the measurements of CSES GOR are consistent with those of ionosonde and incoherent scatter radar (ISR) in Millstone Hill, including F2 layer peak electron density (NmF2), its height (hmF2) and electron density profile.

The study of data validation for other satellite missions have also been carried out by scientists, which gives confidence for scientific research using these data. The electron density and temperature of the Swarm constellation were compared with ISR, ionosonde and COSMIC data, and the sufficiently high correlation coefficients were reported by Lomidze et al. [11]. The study of Cai et al. [12] shows that the data from Global-Scale Observations of the Limb and Disk (GOLD) match well with the GPS TEC in the morphology of equatorial ionization anomaly (EIA) and its seasonal variability. By comparing the thermospheric neutral wind observed by the Ionospheric Connection Explorer (ICON) spacecraft, and four ground-based specular meteor radars, Harding et al. [13] found the agreement between space-based and ground-based winds. Pedatella et al. [14] reported that the morphology of the total electron content (TEC) from FORMOSAT-7/COSMIC-2 and GPS TEC is similar, and the absolute TEC of MOSAT-7/COSMIC-2 shows an overall consistency with Swarm-B observations.

The data quality is worth being researched for a long time. Eight years after Swarm launched in 2013, scientists are still comparing the Swarm data with other observations for future improvement [15,16]. Although the comparisons between LAP Ne and other observations/models have been carried out in previous studies [5,6,7], the data analysis is still needed from different aspects to meticulously and extensively evaluate the observations. CSES payloads operate in two different modes, called survey and burst modes. In most regions of the world, the satellite works in survey mode. When the satellite flies over China, the Circum-Pacific and Eurasia seismic belts, the burst mode is automatically triggered with a higher resolution. Fortunately, there is an incoherent scatter radar (ISR) station and several ionosonde stations in China, which support data to carry out the comparison between the in situ Ne of CSES with burst mode and the ground-based remote sensing of the ionosphere. In the article, Section 2 gives the description of the data. Section 3 and Section 4 illustrate the comparison between LAP Ne and ISR data according to the ionosonde observations. The discussion and conclusions are in Section 5 and Section 6, respectively.

2. Data

The CSES is a circular sun-synchronous satellite, with the descending and ascending nodes at 14:00 LT (local time) and 02:00 LT, respectively. The satellite is three-axes stable to set the altitude at about 507 km with an orbital inclination of 97.4°. As the orbit of the CSES is strictly revisited every five days, the global electromagnetic environment of the ionosphere can be obtained every five days [17].

In this study, we pay attention to the in situ electron density data observed by LAP. The LAP payload is installed on the front of the satellite with two spherical sensors and one electronics box. Sensor 1 is a larger one with a diameter of 5 cm, and the diameter of sensor 2 is 1 cm, designed as a backup. There is a long stub 50 cm in length for each sensor to reduce the interference from the satellite surface [18]. The data we used in this study were observed by sensor 1. The plasma parameters can be obtained by fitting the I–V characteristic curve of LAP based on the sweep voltage and sensor current, which is introduced in detail by Yan et al. [19]. The Ne measurement range of CSES LAP is 5 × 10²–1 × 10⁷ cm⁻³ with a relative accuracy of 10% [20]. For the LAP payload, the time resolutions of the burst and survey modes are 1.5 and 3 s, which means the latitudinal interval of each sample is about 0.1° and 0.2°, respectively. The electron density data observed by the CSES can be downloaded from the mission website (http://www.leos.ac.cn/, accessed on 10 August 2021) after registration. Currently, the CSES mission does not stamp the quality flag to LAP Ne. In our analysis, Ne samples with a value higher than 1 × 10⁷ cm⁻³ or lower than 5 × 10² cm⁻³ were omitted, as it is considered that there may be bias in the fitting process if the data exceed the measurement range.

The accuracy of the ionosonde and ISR observations is widely accepted, and these data are usually applied to validate the satellite observations [11,21,22]. In China, there is an ISR in operation at the Qujing site and some ionosondes at different latitudes, the data of which can be applied to compare with the electron density of the CSES in the China region. The Qujing ISR (QJISR) was constructed by the China Research Institute of Radio Wave Propagation (CRIRP) at the beginning of 2012 [23], and the station is situated in Yunnan province with the latitude 25.6°N and longitude 103.7°E, which is plotted in Figure 1 with a green star. There are two groups of ionosonde data to compare with the CSES Ne. One is from the Institute of Earthquake Forecasting (IEF), China Earthquake Administration, including three ionosonde stations in the southwest China, which is marked by a blue point in Figure 1. Leshan station is located in Sichuan province, with the longitude 103.8°E and latitude 30.0°N, and Yunnan province has two stations, located in Tengchong (98.5°E, 25.0°N) and Puer (101.1°E, 22.7°N) [24]. The other group is from CRIRP. Since the 1940s, CRIRP has established several ionosondes from lower (20.0°N) to middle latitudes (49.6°N) in the mainland of China [25], and the locations of the data we used are marked by red triangles in Figure 1.

3. Comparison with ISR Data

QJISR works several hours in each day, almost all during the daytime, with a time resolution of about 5 min. Corresponding to the available data of the CSES Ne, 25 events were found for QJISR from 2018 to 2021. The ISR can give an Ne profile with an altitude resolution of about 5 km. The principles of the selected comparison dataset are as follows. (1) The mean value (M) and standard deviation (std) of QJISR Ne at the altitude 450–550 km from 13:00 to 15:00 (LT) were obtained to represent the observation at the satellite altitude in the daytime for each day. There are differences for the working time of QJISR each day; thus, the time range covers 2 h to ensure sufficient samples are obtained for the comparison between the conjugate data. (2) Corresponding to the date of QJISR observations in 25 days, the M and std of the CSES Ne in the region between 93.8–113.8°E and 15.6–35.6°N were calculated for the daytime data. The purpose of the selected ±10° region is to provide a sufficient number of observations. With the burst mode of CSES, the resolution of latitude is almost 0.1°; thus, there are about 200 data points to obtain the M and std.

Figure 2 gives the Ne time series for the CSES (red line) and QJISR (blue line) in the comparison days, which is not continuous and the detailed date is marked on the X axis. Every day, the circle and bar represent the mean value and standard deviation calculated according to the above principles. The two time series are displayed with different amplitude scales to better underline their own variations. The Ne variation pattern between the CSES and QJISR is almost similar, with a high value in 2018 and a relative low value in 2020 and 2021. Furthermore, some trends in a short time frame fit perfectly, for example, the data from 25 May to 7 June 2018.

To check the difference in the value of electron density, the relative change of CSES Ne corresponding to QJISR data was calculated according to Equation (1).

$$Rel=\frac{N{e}_{-}CSES-N{e}_{-}QJISR}{N{e}_{-}QJISR}\times 100$$

(1)

where Ne_CSES represents the data of CSES Ne, and Ne_QJISR is the Ne at satellite altitude calculated based on the above principle. A histogram of the relative change value is shown in Figure 3. It can be seen that all of the ratios are negative with a median value −84.04%, which means the average value of CSES Ne is lower than that of QJISR Ne at the satellite altitude.

To further analyze the correlation between the two datasets, panel a of Figure 4 gives the Ne scatters of CSES and QJISR with the detailed date for each point. It seems that these data can be separated into two groups, which are marked by the red oval. It can be seen that except four points in 2018, the time range of the group 1 data is in 2020 and 2021. The data of group 2 belong to 2018, except one point in 2020. Combined with the curves in Figure 2, the magnitude of the group 2 data is mostly higher than that of the group 1 data. Since the available dataset for the comparison of CSES and QJISR Ne is limited, the characteristics of ionospheric variation cannot be completely considered, and the magnitude of observation may play a leading role in the correlation analysis. Based on the two groups, the data were separately fitted in panel b of Figure 4, with the blue and black points denoting the data of group 1 and 2, respectively. The relationship of the Ne data for CSES and QJISR is better represented by a linear correlation in each group with the empirical formula in Equation (2).

Y = 19.94 × X − 2.45 × 10¹¹, Y = 11.49 × X − 4.3 × 10¹¹

(2)

Furthermore, the correlation coefficient between the two datasets was calculated according to Equation (3).

$$R\left(X,Y\right)=\frac{Cov\left(X,Y\right)}{\sqrt{Var\left[X\right]Var\left[Y\right]}}$$

(3)

where X and Y represent the Ne values of the CSES and QJISR. Cov means covariance between the two datasets, and Var is the variance of each dataset. The correlation coefficients are 0.9250 and 0.8884 for the conjugate data of groups 1 and 2, respectively, which means the Ne of the CSES and QJISR exhibits a strong correlation.

4. Comparing with Ionosonde Data

Based on the two groups of ionosonde data, the continuous-time observations of three stations from IEF were used to carry out time series analysis, and the data of 10 ionosonde stations from CRIRP were applied to analyze the correlation at different latitudes.

In the ionosphere, the electron density above the F2 region is determined by chemical production and loss, transport due to neutral wind, E × B drift, ambipolar diffusion and so on. In the daytime, the production, loss and transportation of electrons are mainly affected by the sun. The ionospheric electron density is mainly produced by the photoionization of solar EUV and X-ray radiation, which is represented by the fact that NmF2, TEC and Ne at the topside of the ionosphere linearly change with the solar proxies at low and medium solar activity levels [26,27,28]. Solar activity and photochemical equilibrium play a leading role in the variability of Ne in the daytime, and the electron density at different altitudes may synchronously variate with the solar effect. In the nighttime, the compound and transport actions induced by chemical reaction, dynamic and electrodynamic processes are the main effect factors to make the ionosphere express local characteristics at the latitude, longitude and altitude [29,30]. The linear dependence between electron density and altitude may be broken down, which means some biases will be brought when the NmF2 are inferred to Ne at satellite altitude based on the empirical model. Therefore, in the Ne comparison of the CSES and ionosonde, we just paid attention to the data analysis in the daytime.

The detailed data process of the selected comparison dataset is as follows. On the one hand, the mean value (M) and standard deviation (std) of the CSES Ne falling in the region of ±10° around the ionosonde station were calculated to represent the Ne data above the ionosonde station. As the CSES is a circular sun-synchronous satellite, the longitudinal distance of each orbit in a day is about 24°. The reason for selecting ±10° as the range is to provide a sufficient number of observations and temporally continuous data to make the comparison statistically meaningful, and to avoid the ionospheric variation on a large scale. On the other hand, the IRI model was used to infer the Ne data at satellite altitude based on the ionosonde observations. The topside density profile can be obtained by fitting data to a Chapman-α layer [31], while the selection of ionospheric scale height (H) can also affect the fitting result [32,33]. As the electron density is assumed to be mostly linear-dependent at altitudes above the hmF2 [34,35], the ratio outputted by the IRI model was applied to obtain the Ne at satellite orbit in this study. The IRI model is an empirical standard model of the ionosphere, which includes the main ionospheric functions and available data sources [36], and is updated yearly according to the analysis of the IRI working group (http://irimodel.org/, accessed on 1 September 2021). There are three steps to obtaining the comparison dataset for ionosonde data. Firstly, the ratio of Ne at 507 km and NmF2 at the ionosonde location was outputted with a 15-min time resolution based on the IRI 2016 model. Secondly, the NmF2 observed by the ionosonde was multiplied by the above ratio to obtain the Ne data at the satellite altitude. Finally, the mean value and standard deviation of the calculated Ne from 13:00 to 15:00 (LT) were acquired to represent the ionosonde data in the daytime.

4.1. Time Series Analysis

As the first three months after the CSES launch is within the commissioning phase, the data provided by the satellite mission are only taken from 8 May 2018. Since the date of the available data is different, the time range of the comparative dataset is different for each ionosonde—Leshan station from 8 May 2018 to 13 October 2020, Tengchong station from 1 January 2019 to 31 July 2021, and Puer station from 8 May 2018 to 31 July 2021. Figure 5 shows the time series of the CSES Ne and Ne at the satellite altitude inferred by ionosonde observations and the IRI model, which is displayed with the log scale. The patterns of two curves are almost same, and the annual and semiannual variations can both be found for the two datasets, with high values in the summer and low values in the winter. Furthermore, the correlation coefficient between the two datasets was calculated according to Equation (3), where X and Y represent the Ne of the CSES and ionosonde. The scatters of the two datasets and linear fitting curves are plotted in Figure 6. The calculated values for the Leshan, Tengchong, and Puer stations are 0.4858, 0.5973 and 0.4944, respectively, which means the Ne of the CSES and ionosonde represents a medium correlation and can be well linearly fitted. In Figure 5, it can be seen that there is a difference in the Ne absolute value between the CSES and the ionosonde. The relative change between the CSES and ionosonde Ne was calculated according to Equation (1), where Ne_QJISR was changed to Ne_ionosonde. Figure 7 shows a histogram of the relative change between the two datasets for the Leshan Figure 7a), Tengchong (Figure 7b) and Puer (Figure 7c) stations. All of the ratios are negative, which means the electron density of the CSES is lower than the Ne at the satellite altitude inferred by ionosonde observations and the IRI model. Based on the median values of the ratios in each station, the value of CSES Ne is lower by 80–90% in the daytime, corresponding to the ionosonde Ne.

4.2. Latitudinal Analysis

According to the above data process illustrated at the beginning of Section 4, the comparison dataset between the Ne of the CSES and ionosonde data from CRIRP was constructed in March, June, September and December of 2019. Taking Lhasa station as an example, Figure 8 gives the time series of the two data, and the median value of relative change (calculated by Equation (1)) and correlation coefficient (obtained by Equation (3)) between the CSES and ionosonde Ne are also marked in each panel for different months. The variation patterns of the two datasets are almost consistent, and the peaks or troughs coincide with each other at times, for example, the similar trend seen in mid-September and the single peak in December. The correlation coefficients all exceed 0.4, and the value exceeds 0.7 in June. The value of the CSES Ne is lower than that of the ionosonde Ne with a ratio of 80–90%.

In the same way as for the Lhasa station, the correlation coefficients between the CSES and ionosonde Ne for each station from CRIRP were calculated according to Equation (3) for March, June, September and December of 2019, which was summarized in

Table 1. Correlation coefficient between CSES and ionosonde Ne at different ionosonde stations.

Station	March	June	September	December
Manzhouli (MZL)	0.5331	0.2192	0.1460	−0.2932
Changchun (CC)	0.2410	0.5243	0.4167	−0.0542
Urumuchi (UC)	0.3017	0.2121	−0.0263	−0.0101
Beijing (BJ)	0.4895	0.2193	0.5864	0.1550
Lanzhou (LZ)	0.4018	0.6445	0.5585	0.4069
Xi’an (XA)	0.3986	0.5119	0.6665	0.0989
Lhasa (LS)	0.4476	0.7124	0.5153	0.6356
Chongqing (CQ)	0.5444	0.7533	0.5366	0.3109
Kunming (KM)	0.3939	0.5786	0.7311	0.3447
Haikou (HK)	0.3552	0.5524	0.5228	0.6494

. For a more intuitive representation, the differential correlation coefficients by 1° × 1° were obtained based on all of the data in Table 1; the contour lines with colors are shown in Figure 9, where the name of each station is used the abbreviation. It can be seen that the values at the low to mid-low latitudes are mostly higher than those at the middle latitude, which is more obvious in June, September and December. In general, the correlation between the two datasets in June and September is relatively better than that in other months, with the highest value in June for the CC, LZ, LS and CQ stations, and in September for the BJ, XA and KM stations.

The median value of the relative change of the CSES Ne corresponding to the ionosonde observations (calculated by Equation (1)) for each station is summarized in

Table 2. Median value for relative change of CSES Ne corresponding to ionosonde Ne at different stations.

Station	March	June	September	December
Manzhouli (MZL)	−93.51%	−90.35%	−94.48%	−94.22%
Changchun (CC)	−85.52%	−80.47%	−84.62%	−91.08%
Urumuchi (UC)	−93.81%	−88.35%	−93.46%	−94.29%
Beijing (BJ)	−93.13%	−87.11%	−93.50%	−94.10%
Lanzhou (LZ)	−90.48%	−85.00%	−89.99%	−93.45%
Xi’an (XA)	−89.72%	−84.45%	−90.10%	−92.86%
Lhasa (LS)	−87.09%	−82.18%	−86.02%	−91.17%
Chongqing (CQ)	−85.52%	−80.47%	−84.62%	−91.08%
Kunming (KM)	−88.72%	−84.31%	−89.52%	−91.91%
Haikou (HK)	−85.29%	−81.85%	−87.13%	−93.21%

. All of the data are negative, which means that the value of the CSES Ne is lower than that of the Ne at the satellite altitude inferred by ionosonde observations and the IRI model at different latitudes. The values are from −80% to −95%. In general, the difference between the two datasets is a little larger in December compared to other months, when the correlation coefficient is relatively low.

5. Discussion

Figure 5 shows that the day-to-day variation can be found in the observations of both the CSES and ionosonde Ne, and the scaled ionograms of the ionosonde can also be affected by the phenomena of spread F, the stratification of the F2 layer, sporadic E layer (Es) and so on [37,38,39], which can induce differences between the CSES and ionosonde Ne. Meanwhile, the disturbances of the running mean with 15 days (shown by red and blue lines in Figure 5) are not frequent. The correlation coefficients of the running mean values between the two datasets were calculated according to Equation (3), and are listed in

Table 3. Correlation coefficient between CSES and ionosonde Ne for Leshan, Tengchong and Puer stations in different conditions.

Station	Raw Data	Running Mean	Geomagnetically Quiet Day
Leshan	0.4858	0.7557	0.5785
Tengchong	0.5973	0.7506	0.6334
Puer	0.4944	0.6425	0.5431

for these three stations. The coefficients increase compared to the former analysis, with values up to 0.75, especially for the Leshan and Tengchong stations. Furthermore, the ionospheric disturbances can also be induced by solar activity, magnetic storms, and other space events [30]. In order to analyze the correlation on a geomagnetically quiet day, the coefficients between the CSES and ionosonde Ne were only calculated for the data in the days with 10.7 cm solar radio flux (F_10.7) < 160, Kp ≤ 3, |Dst| < 30 nT and AE < 500 nT, as shown in Table 3. Compared to the coefficient calculated by all data, the correlation between the CSES and ionosonde Ne is better on geomagnetically quiet days, which indicates that the consistency of the different observations for the ionosphere can be affected by geomagnetic events. It can be seen that the day-to-day variation and disturbance events could result in differences in the correlation analysis.

In terms of time, the correlation coefficient in June and September is generally better than that in March and December. In Figure 5, the values of the CSES Ne in June and September of 2019 are all higher than those in March and December for three ionosonde stations. It seems that the CSES Ne in the month with a relative high value may exhibit good correlation with the ionosonde observations. In general, the signal-to-noise ratio (SNR) of the payload will be better under the conditions of the observation parameters with a relatively high value. In terms of latitude, the correlation between the CSES and ionosonde Ne is better at the low latitude than that at the middle latitude during the daytime. The spatial distributions of the monthly mean for the CSES Ne in March, June, September and December of 2019 are exhibited in Figure 10. The patterns of the CSES Ne in these four months all show the high values at low latitude and low values at middle latitude. The region of the Ne peak around EIA occupies a wider area with stronger strength in March, June and September, and the values at mid to low latitude are relatively high in June. Compared to the pattern of CSES Ne distribution, it can be seen that the high value of the correlation coefficient may have a relationship with the EIA phenomenon. Affected by the fountain effect during the daytime, the phenomenon of EIA can be found at the low latitude for both the CSES and ionosonde; thus, the correlation coefficient is high around the peak of EIA. Compared to other locations, Lomidze et al. [11] also found that the initial Ne observed by Swarm agrees relatively well with COSMIC data near the magnetic equator in the daytime. Yan et al. [6] reported that the phenomena of EIA and longitudinal wave number around the magnetic equator can be observed by both the CSES and the Swarm B satellite. In the comparison between the CSES Ne and the IRI model, Liu et al. [7] detected that the correlation coefficient in the daytime is higher around the peak of EIA than that in other regions.

Compared with the QJISR data and ionosonde observations, the average values of the CSES Ne are relatively lower, with the relative change about 80–95% in the daytime corresponding to the ground-based data. Although the IRI model can exactly exhibit the ionospheric characteristics, some researchers found that the value of Ne at the topside ionosphere are overestimated by the IRI model [32,40]. The systematic biases are also found in the comparison of electron density between the CSES and Swarm constellation [5,6], DEMETER satellite and the IRI model [7], which may be caused by the different designs of the payload, different data processing and the overestimation of the IRI model at the topside ionosphere. Furthermore, similar to the above illustrations, there may be some biases in the interpretation of ionograms when the phenomena of spread F, stratification of F2 layer and Es occur in the ionosphere, which may induce the discrepancies of Ne between the CSES and ionosonde observations. Furthermore, the reliability of ISR data is widely accepted, although it is somewhat model-dependent. Different from ground-based observations for the topside ionosphere, the in situ datum of the satellite is the direct measurement of ionospheric parameters. The payload of LAP operates as an absolute instrument on the platform of the CSES, and is able to identify the plasma potential, thus retrieving the actual electron density, which is a valuable measurement for ionospheric in situ observations. Through the comparison, scientists found that the accuracy of LAP installed on a satellite could be different from mission to mission [21,41]. Although there are differences in the Ne values between ground-based remote sensing and satellite in situ measurements, similar variations and relatively good correlation between the two datasets can certify the authenticity of in situ Ne observed by the CSES.

6. Conclusions

Based on the comparison between the CSES Ne and ground-based observations in China, some conclusions were summarized from the variation trend, correlation and values.

• The pattern of the CSES Ne is consistent with that of the ISR Ne at the satellite orbit. The two datasets can be linearly fitted, seemly according to the magnitude, and the correlation coefficient is above 0.88. However, more data are needed to more deeply understand the relationship between the two sets of data.
• The trend of the CSES Ne is mostly similar to the ionosonde observations no matter whether the data are continuous in time or in a certain month; some correlation coefficients between the two datasets exceeded 0.7. On the China mainland, the correlation coefficient is higher around the peak of EIA compared to other locations.
• The value of the CSES Ne is low at 80–95% in the daytime, corresponding to the ISR and ionosonde observations in China.