Observation of Postsunset OI 135.6 nm Radiance Enhancement Over South America by the GOLD Mission

The Global-scale Observation of Limb and Disk (GOLD) mission, for the first time, provides synoptic two-dimensional (2D) maps of OI 135.6 nm observations. These maps describe the unambiguous and dynamic evolution of nighttime ionospheric F 2 -peak electron densities ( N m F 2 ) as the 135.6 nm airglow emission radiance correlates well with N m F 2 at night. On November 19, 2018, the 135.6 nm radiance measured by GOLD, N m F 2 measured by a digisonde, and GPS total electron content (TEC) measurements at Cachoeira Paulista (CP) all showed a postsunset enhancement, with an increase near 22:30 universal time. The 135.6 nm radiance map showed that this enhancement was due to the southward movement of the southern equatorial ionization anomaly (EIA) crest. Therefore, the GOLD observation showed the linkage between postsunset enhancement of N m F 2 and EIA movement. Furthermore, unlike the southward movement of the southern crest, the corresponding EIA northern crest, however, did not show northward motion. This is the

scatter radars (ISRs) (Mikhailov et al., 2000;Pavlov & Pavlova, 2005), total electron content (TEC) maps (Balan & Bailey, 1992;Balan & Rao, 1987;Balan et al., 1994;Davies et al., 1979;Essex & Klobuchar, 1980;Rajesh et al., 2016;Trivedi et al., 2013;Zhao et al., 2008), and Low Earth Orbiting (LEO) satellites (Y. Chen et al., 2015;He et al., 2009;Luan et al., 2008;Zhong et al., 2019). Many numerical simulation studies have also focused on the reproduction of the nighttime electron density enhancement (C. H. Chen et al., 2013Chen et al., , 2011Le et al., 2014;Nicolls, et al., 2006;Ren et al., 2012;Thampi et al., 2011). Based on these numerous previous studies, it has been found that the nighttime N m F 2 enhancement can be categorized into low latitude and mid-latitude ones from the perspective of geographic latitude. Furthermore, nighttime N m F 2 enhancement can also be classified into postsunset and postmidnight enhancement. Farelo et al. (2002) gave a detailed description of the latitudinal and seasonal variations of nighttime N m F 2 enhancement in mid-latitudes. Zhong et al. (2019) have provided a detailed explanation on the nighttime N m F 2 enhancement in mid-latitudes for both ionosphere F 2 region and topside. They argued that the upward plasma motion due to neutral winds was the primary source of the nighttime enhancement in the ionosphere F 2 region.
Compared with many studies on the mid-latitude nighttime N m F 2 enhancement, there are relative fewer studies focusing on low-latitude nighttime N m F 2 enhancement. Anderson and Klobuchar (1983) suggested that the E × B drifts near the geomagnetic equator were responsible for the postsunset N m F 2 enhancement between 19:00 and 23:00 local time (LT) over Ascension Island (7.95°S, 15°W; 15°S dip latitude), which is located equatorward of the southern crest of the equatorial ionization anomaly (EIA). Zhao et al. (2008) reported a postsunset N m F 2 enhancement event in the Asia-Australian sector (June 28, 2002) and attributed it to the strong prereversal enhancement (PRE) of vertical plasma drifts near the geomagnetic equator based on the data from TEC maps, ionosondes and two LEO satellites. L. Liu et al. (2013) investigated a case of postmidnight N m F 2 enhancement using data from a digisonde at Sanya (18.3°N, 109.6°E, dip latitude 8. 44°N) and argued that the westward electric field played a major role in the formation of this enhancement. This case was further analyzed by Le et al. (2014) with a numerical model. They proposed that both neutral wind and westward electric field contributed to the reported postmidnight N m F 2 enhancement in L. Liu et al. (2013). Y. Zhang et al. (2015) utilized four ionosondes in Japan to study the nighttime N m F 2 enhancement and found that the postsunset enhancement was more frequent in summer months. Jiang et al. (2016) utilized four ionosondes in South China and Southeast Asia to study the latitudinal variation of low-latitude postmidnight N m F 2 enhancement. They found that the LT of the occurrence of the postmidnight N m F 2 enhancement was earlier in the geomagnetic southern hemisphere than the geomagnetic northern hemisphere.
Most of these previous studies of the low-latitude nighttime N m F 2 enhancements were carried out using local observations such as ionosondes with only temporal variations. They rarely reported the ionosphere structure over a large area such as the corresponding nearby EIA structure. In reality, all these reported low-latitude nighttime N m F 2 enhancements are near the geomagnetic equator or the EIA crests. Therefore, it is crucial to know the behavior of EIA structures over a large area when investigating the low-latitude nighttime N m F 2 enhancements.
EIA is a phenomenon in the equatorial ionospheric F layer with two electron density crests at about ±15° magnetic latitudes and a density trough at the geomagnetic equator (Appleton, 1946). Due to different neutral wind and chemistry on the two sides of the geomagnetic equator, the densities on two crests are mostly different, which is called as EIA asymmetry. There are two types of EIA asymmetry. One is the hemispheric asymmetry, which is the density difference between the two crests in the same season. Another is the EIA annual asymmetry, which is the phenomenon that the December EIA strength is stronger than the June one. The asymmetry of EIA structures has been studied for decades by using data from multiple ionosondes (Thomas, 1968), TEC maps (Khadka et al., 2018;Rama Rao et al., 2006;M. L. Zhang et al., 2009;Zhao et al., 2009) and LEO satellites (Balan et al., 2013;Basu et al., 2009;Henderson et al., 2005;Huang et al., 2018;Kil et al., 2006;Lin et al., 2007;Luan et al., 2015;McDonald et al., 2008;Tallat et al., 2013;Tulasi Ram et al., 2009;Zeng et al., 2008). Zeng et al. (2008) found that the hemispherically asymmetric magnetic field was the most important factor to generate the EIA annual asymmetry. Khadka et al. (2018) investigated the EIA hemispheric asymmetry near 75°W near local sunset. They found that the asymmetry appeared more frequently in solstices and March equinox than September equinox. Tulasi Ram et al. (2009) found that the hemispheric asymmetry appeared near 12-13 LT during solstices. Dang et al. (2016) compared the role of meridional neutral winds, photochemical effects, and magnetic field configuration played in the EIA hemispheric asymmetry in daytime during summer solstices. They suggested that the trans-equatorial neutral wind is a major contributor to the north-south asymmetry, which was similar to the modeling studies carried out by Balan et al. (2013).
It should be noted that all these previous studies focused on EIA asymmetry in a climatological way based on local ground observations and/or LEO satellite observations. LEO satellite observations can only provide cross-track scanned images or profiles at certain LTs near the orbit. These observations cannot delineate the spatial and temporal variability. TEC maps can provide both temporal and spatial variations of the EIA asymmetry with high resolution (as high as 1° × 1° and 5 min). However, most of the GPS receivers are located on continents (excluding the remote area such as Desert), which only counts for 30% of the Earth surface. Furthermore, the uneven distributions of the GPS receivers may also introduce errors The newly launched NASA Global-scale Observation Limb and Disk (GOLD) mission, for the first time, provides two-dimensional (2D), synoptic maps of OI 135.6 nm airglow emission observations. These maps describe unambiguously the dynamic evolution of nighttime N m F 2 , since the 135.6 nm airglow emission rate correlates well with N m F 2 at night (DeMajistre et al., 2004). They provide unambiguous observations of temporal and 2D spatial variations of the postsunset N m F 2 enhancement over a large area that is not feasible in the aforementioned local observations, and thus advance our understanding of the nighttime electron density enhancement phenomenon.
In this paper, we focus on the postsunset N m F 2 /OI 135.6 nm radiance enhancement in low latitudes seen in 2D 135.6 nm airglow images obtained by GOLD. We report a case study of the postsunset N m F 2 enhancement event at the Cachoeira Paulista (CP) station, Brazil (22.7°S, 45°W) (located in an area that is usually poleward of the southern crest of EIA) observed by GOLD on the night of November 19, 2018. In addition, we use TEC maps and digisonde data from the same area, which also observed a similar postsunset enhancement as GOLD did, to provide a comprehensive view of the phenomenon. The main purposes of this study are to validate the GOLD nighttime observation; provide new insight into the postsunset N m F 2 enhancement and the motion of EIA crest, and demonstrate the powerful potential of GOLD observations.

Data Source
The GOLD instrument is onboard the SES-14 communication satellite, which was launched on January 25, 2018. The satellite is located in a geostationary orbit over 47.5°W. The only scientific instrument onboard is a Far Ultraviolet (FUV) imager, which has two similar and independent channels (Channels A and B) to perform airglow measurements (Eastes et al., 2017(Eastes et al., , 2019. Since GOLD makes measurements from the geostationary orbit, it is able to provide airglow images of the same region during the same time range every day (Cai et al., 2020a(Cai et al., , 2020bKaran et al., 2020;Laskar et al., 2020). At night, it observes the OI 135.6 nm with a spectral resolution of 0.04 nm. The data we utilize for this study are the GOLD Level 1C OI 135.6 nm radiance over the equatorial region in the South American continent between 21:55 universal time (UT) and 23:55 UT at local night. The nighttime OI 135.6 nm emission is produced mainly due to the recombination reaction between the oxygen ions (O + ) and electrons (McDonald et al., 2008;Rajesh et al., 2011). For the GOLD observation along a given line of sight (Kelley et al., 2003), it can be modeled as as having the same variability in the following sections.
The critical frequency data of the F 2 layer (f 0 F2) measured by the digisonde at CP were downloaded from the Global Ionospheric Radio Observatory (GIRO) web portal (Galkin et al, 2012;Reinisch & Galkin, 2011). The data are auto-scaled, and we checked every corresponding ionogram to ensure accuracy. N m F 2 was derived from a relationship between the electron density and critical frequency  The ionospheric TEC data with a 5-min temporal resolution and 1° spatial resolution (in both latitude and longitude) were obtained from the Madrigal database (Rideout & Coster, 2006). In order to match the resolution of GOLD nighttime observations, especially after 22:55 UT, we binned the TEC data with a 15-min resolution. Although TEC is the integration of electron density along the GPS path, it can be assumed that the TEC morphology is similar to that of N m F 2 , especially during solar minimum (Solomon et al., 2018). In addition, Rajesh et al. (2011) demonstrated that the OI 135.6 nm radiance had a linear relationship with the square of TEC (TEC 2 ). In the following, we use TEC 2 , instead of TEC to compare with GOLD observation.
In order to extract the radiance at CP from the GOLD data, we calculated the average of the radiance in 1°, 2° and 5° diameter circular bins centered at CP. For the 1° bin, we collected all data points within a distance to CP (22.7°S, 45°W) less or equal to 0.5° of latitude and longitude. We then calculated the average of the radiance values in the bin. The distance to CP was increased to 1° and 2.5° when we utilized 2° and 5° bins, respectively. The 1° bin is adapted to match the spatial resolution of the TEC data so that a direct comparison can be carried out. We checked the skymap of the digisonde at CP from the drift database of GIRO, and found that the horizontal scale of the digisonde measurements is in a circle with a maximum radius of 110 km (∼1°). Therefore, the 2° bin was employed for comparison with the digisonde measurements. Finally, we added the 5° bin because sometimes the GOLD measurements in the nighttime may have high noise levels; a larger bin size improves signal-to-noise level and ensures data quality.
On the night of November 19, 2018, GOLD began to image the regions of West Africa and the Atlantic Ocean using Channel B from 20:10 UT. It imaged both the northern and southern hemispheres every 15 min in turn (for example, at 20:10 UT, GOLD imaged the northern hemisphere near west Africa, then the southern hemisphere at 20:25 UT and again the northern hemisphere at 20:40 UT. The region that was imaged was moving westward due to the movement of the day night line). Channel B of the FUV imager began to image South America from 21:55 UT, and then at 22:25 UT, and again from 22:55 UT to 23:55 UT with a 15-min cadence. Between 23:10 UT and 23:55 UT, Channel A started to image the northern hemisphere simultaneously with Channel B. In a word, the temporal resolution is 30-min from 20:10 to 22:55 UT, and 15-min between 23:10 and 23:55 UT. The spatial resolution of the measured OI 135.6 radiance varies from geomagnetic equator to EIA crests, but mostly between 0.5° × 0.5° and 1° × 1°. The nighttime OI 135.6 nm radiance measurement error is around 10% . We have four images covering both the northern and southern hemispheres in that time period, which provided the 2D evolution of the nighttime equatorial ionosphere. UT, GOLD channel B scanned the southern hemisphere every 15 min, so we have GOLD data at CP at 23:10, 23:25, 23:40, and 23:55 UT. The comparison between GOLD and local observations over a single location has great advantages over previous comparisons of monthly climatology provided by LEO satellites because much variability can be removed by doing monthly climatology, such as the comparison between Constellation Observing System for Meteorology, Ionosphere, and Climate and ionosonde measurements in Lei et al. (2007). In addition, the GOLD measurements are 2D synoptic maps, providing information over a much larger area, which is not available from LEO cross-track scanning. When binned at 1° (black dots), the radiance increased from 12. available after 23:30 UT because spread F appeared at CP (see Figure A1 in Appendix A), which prevents the observation of f 0 F 2 . With the detailed variations of radiance from three bin sizes presented, we found that the 1° and 2° binning results were closer to the variation of 2 2 F m N compared with the 5° binning. This is because the results with a larger bin size can be influenced more easily by changes in the areas further away from the center (CP).

Observations over Cachoeira Paulista
The corresponding h m F 2 is shown in Figure 1b. It increased from 269 km at 22:00 UT to 325 km at 22:40 UT, and then decreased to 293 km after 22:50 UT. There was a small increase of h m F 2 at 23:00 UT, but it decreased with time after that. In addition, we also present the corresponding TEC 2 at CP (from the 1° binned Madrigal data containing the CP site) in Figure 1c  are 0.9168, 0.8545, 0.8283, 0.8227, 0.8464, and 0.8668, respectively. All are above 0.8. This demonstrates that the OI 135.6 nm and the N m F 2 and TEC 2 are correlated well in 2018 November. More comparisons between GOLD and local observation in other months in the same time range will be carried out as future work (Cai et al., 2020b).

GOLD and TEC Observation over South American Continent
After showing the comparisons of GOLD, digisonde, and TEC data at a single location, we turn to the observations over a large area. Figure 2 shows an example of the comparison between the TEC 2 map and GOLD radiance map in the South American sector (coast-line shape marked in red) at 23:10 UT. The location of CP is marked on both maps (white cross). The spatial resolution of the airglow radiance and TEC data is 1°, and both of them have been processed by a moving window of 5° along each latitude to remove the small-scale perturbations on the EIA crests. We also marked the location of the peak density of the southern EIA crests in both maps (black dots). The distance in latitude between CP and the corresponding peak of the southern crest in the GOLD map was 4.1°, and 3.7° in the TEC 2 map, with a difference of less than 1°. The locations of the peaks of the southern EIA crest in both two maps were close to each other, apart from several longitudes (such as 47°W and 39°W). It should be noted that the peak location on TEC 2 map is not available outside of the continent due to the TEC data unavailability over the Ocean. In all, the nighttime GOLD observations match well with the observation made by other instruments in both single locations and over a large area. Figure 3 shows the airglow radiance in the South American sector (35°S to 5°N and 30°W to 70°W) imaged by GOLD Channel B from 22:25 to 23:55 UT, with both CP (white cross) and the corresponding location of the southern crest EIA peak at 45°W (black dot) marked on. At 22:25 UT, CP was poleward the EIA southern crest and had a distance of 7.4° from the peak of the southern EIA crest at 45°W. At 22:55 and 23:10 UT, the south crest was moving southward toward CP, and the distance between CP and the peak at 45°W decreased to 4.9° and 4.1°, respectively. As the EIA southern crest moved closer to CP after 22:25 UT, N m F 2 at CP increased from 22:30 UT, as shown in Figure 1b. At 23:25 UT, the southern crest peak was 3.3° north from CP. At 23:40 UT, the southern crest was almost over CP (with a distance of only 0.2°). The peak was finally passed over CP and located 0.6° south of CP at 23:55 UT. Therefore, based on the dynamic evolution of the GOLD 135.6 nm radiance map shown in Figure 3, we can see that the postsunset enhancement of CAI ET AL.
10.1029/2020JA028108 6 of 18  airglow radiance/N m F 2 observed at CP is a part of the poleward movement of the southern EIA crest. They revealed clearly that the southward movement of the southern EIA crest resulted in the postsunset enhancement in GOLD airglow radiance, digisonde N m F 2 and GPS receiver TEC at CP. Therefore, the postsunset enhancement of N m F 2 /OI 135.6 radiance is due to spatial variation of EIA. This is different from the previous explanations that the PRE near the geomagnetic equator caused the postsunset of N m F 2 in low latitude ionosphere (Anderson & Klobuchar, 1983;Zhao et al., 2008). In other words, it is the spatial variations of EIA that directly cause this postsunset enhancement of N m F 2 /OI 135.6 radiance, not the temporal variation of E × B drift. The linkage between the postsunset enhancement of N m F 2 and the EIA crest movement in a short time period is discovered. This new finding will enhance the understanding of nighttime equatorial ionosphere and help avoid possible misinterpretation obtained from single local ground observation. More similar cases are searched for and will be studied in the future. Two-dimensional images from GOLD not only provide the evolution of the EIA southern crest, but also provides the behavior of the EIA northern crest simultaneously from 23:10 UT. Figure 4 presents the OI 135.6 nm radiance images by both Channel A and Channel B from 23:10 UT to 23:55 UT with the coastline of continent marked on. Unlike the 5° poleward motion of the southern EIA crest, the northern crest moved much less poleward between 23:10 UT and 23:55 UT. To the east of 40°W, the northern EIA crest did not even move, when compared with the 18°N line during this time period. On the other hand, the distance between the two crests between 40°W and 50°W was increasing during the same time period. For example, at 23:10 UT, the distance between the two EIA crests near 45°W was around 35.6° in latitude, and it increased to 41.1° at 23:55 UT. Therefore, the meridional movement of the two EIA crests showed hemispheric asymmetry. For the corresponding TEC map, it cannot show the variation like GOLD OI 135.6 nm map did due to the difficulty to situate GPS receivers in the Ocean and remote area. The hemispherically asymmetric movement of the EIA can only be captured by GOLD in this region because the scanning from the geostationary orbit is not restricted by surface conditions. Unlike those previous studies about EIA asymmetry that only reported and simulated the climatological electron density difference of EIA crests (Dang et al., 2016;Luan et al., 2015;Tulasi Ram et al., 2009), GOLD observations here capture the process of asymmetric motion of two EIA crests in a short time period (<2 h). This again demonstrates the powerful capacity and advantages of the GOLD 2D OI 135.6 nm radiance map.

Other Related Local Observations
We checked the ionograms during the same time period at two stations near the geomagnetic equator: Sao Luis (2.6°S, 45°W) and Fortaleza (3.9°S, 38.4°W). Spread F occurred over both stations between 22:00 UT and 23:50 UT (see Figures A2 and A3   checked the vertical ion drift velocity (VID) from the digisonde at CP on November 19, 2018, as shown in Figure 5. It was upward with a maximum magnitude of 20 m/s before 22:50 UT and then changed to downward after 23:00 UT, which matched well with the variations of h m F 2 and 2 2 F m N in Figure 1. When there was upward VID, h m F 2 moved upward whereas it decreased with downward VID. By looking at the VID at CP during this time period on other days in November 2018 without such a postsunset N m F 2 enhancement occurring at CP, we found downward VID or an upward one with a much smaller value (<10 m/s) or a combination of both. The magnitude of VID at CP on November 19 was the largest during the whole month from 22:30 to 23:00 UT. All this information suggests that there existed large southward and upward ion drifts in the southern EIA crest on November 19th from about 42°W to 53°W. For the corresponding northern crest, there were two digisondes nearby. One is in Boa Vista (2.8°N, 60°W), another is in Belem (1.43°N, 48.44°W). Unfortunately, the digisonde at Belem did not have observation on this day. The digisonde at Boa Vista only observed spread F after 23:10 UT, suggesting the unavailability of N m F 2 and vertical ion drift results.

Discussion
Based on the observation results in previous sections, it is found that this observed postsunset enhancement is due to the southward motion of EIA southern crest. However, the corresponding EIA northern crest with weaker density did not move northward. In the following, the formation mechanism of this special EIA asymmetric movement is explored.
We ran the National Center for Atmospheric Research (NCAR) Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM) (Richmond et al., 1992;Roble et al., 1988) with a resolution of 2.5° × 2.5° × 1/4 scale height for this observed postsunset enhancement event and the asymmetry movement of the crests on November 19, 2018. Unfortunately, the TIE-GCM with the high-latitude input of either Heelis (Heelis et al., 1982) or Weimer (Weimer et al., 2005) could not reproduce this observed southward movement feature of the southern crest. Therefore, our explanation has to be based on the current available observations only.
From the geomagnetic activity shown in Figure 6, there might also be effects of penetration electric fields on the ionosphere in low-latitudes . However, we found that in the most of the days in November 2018, the EIA crests did not exhibit hemispherically asymmetric movement as they did on November 19 between 22:25 and 23:55 UT, even with stronger geomagnetic disturbances. Figure A4 shows geomagnetic activity from November 4 to November 6, 2018. From the AE index shown in Figure A4, AE also increased around 15 UT on November 4, which is similar to its behavior on November 19. However, with that strong AE in this geomagnetic storm, the corresponding EIA structure ( Figure A5) did not exhibit the asymmetry movement observed on November 19. Therefore, the geomagnetic activity influence on this EIA asymmetric movement is apparently weak.
As shown in Figure 3, the southern EIA crest was approaching CP (but had not passed it) from 22:25 UT to 23:25 UT. The magnitude of the crest at 45°W increased from 28.4 Rayleigh at 22:25 UT to 40.3 at 23:25 UT. Furthermore, the magnitude of OI 135.6 nm airglow emission over CP also increased during this time period. Based on Su et al. (1994), the postsunset enhancement of N m F 2 was attributed to the strong equatorial PRE. Therefore, the E × B drifts over the geomagnetic equator should play a role in this observed postsunset enhancement of N m F 2 or OI 135.6 nm radiance. It should be noted the E × B drifts over the geomagnetic equator should have the same impact on both sides of the equator (Chen & Lei, 2019;Eccles et al., 2015), namely the two crests should exhibit the same behavior. However, the two crests of the EIA actually exhibited asymmetric movement. Unlike the poleward movement and the enhancement of the southern crest near CP, the northern crest did not move poleward and there was also no apparent enhancement of the magnitude. Therefore, the E × B drifts should not be the only reason for this asymmetric movement of EIA crests. The behavior of electron density in the ionosphere F 2 region is determined by chemistry, E × B transport, neutral wind transport and ambipolar diffusion (e.g., J. Liu et al., 2016). Apart from E × B drifts, the other three terms could also contribute to this asymmetric movement. It should be noted that GOLD nighttime observations can only provide N m F 2 , not hmF 2 . It is possible that the hmF 2 of the northern and southern crests are different, causing different recombination effects. Ambipolar diffusion depends on gravity and the ion/electron pressure gradient force. The latter includes the ion/electron density and plasma temperature . It tends to react to the changes caused by other physical processes. It is possible that the plasma temperatures in the two crests are different, causing different ambipolar diffusion. There is also a feedback effect insofar as the occurrence of a different density in each hemisphere then changes ambipolar diffusion. Balan et al. (2013)   where U and V stand for the zonal and meridional neutral winds, and D represents the declination angle of the Earth's magnetic field. Equation 2 also suggests that the direction of the second term on the righthand side is determined by both the declination angle and zonal wind, while the direction of the first term on the right-hand side is determined by the meridional wind only. The magnetic declination angle, from the International Geomagnetic Reference Field (IGRF) model (Finlay et al., 2010), has both positive and negative values between 0° and 60°W, where the variation of the geomagnetic equator is also the strongest. In addition, in November, meridional winds blow northward from the southern hemisphere to the northern hemisphere due to the increased pressure gradient between the two hemispheres (generated by different solar zenith angles). The northward meridional wind pushes the ions along the field line toward the geomagnetic equator to higher altitudes, generating less recombination in the southern hemisphere. While it pushed the ions away from the geomagnetic equator to lower altitudes, resulting in more recombination in the southern hemisphere. Neutral wind transport may also create a suitable foundation and precondition for the asymmetric movement. Regarding the chemistry, molecular nitrogen (N 2 ) near the F 2 region dominates the chemical recombination process during nighttime. As mentioned earlier, the behavior of E × B drifts and neutral wind transport can also modulate the chemical loss by transporting plasma to higher or low altitudes.
The EIA asymmetric movement reported in this paper is a specific space weather event, which may require a more accurate description of the thermosphere and ionosphere state and its variations during the event before it can be properly understood. Further investigation of this phenomenon with more observations and numerical simulations are needed.

Conclusions
In summary, we report a postsunset OI 135.6 nm radiance/N m F 2 enhancement over CP (a location normally south of the southern EIA crest) observed by GOLD, digisonde, and TEC. Further analysis of GOLD images leads to the discovery of a special EIA asymmetry event. Our main findings are summarized as follows: 1. It is the first time that the 2D synoptic measurements from a geostationary satellite are compared with the corresponding local observations (digisonde and GPS receiver) during the same time period (22:25-23:55 UT). The comparison is reasonably good in both single location (CP) and over a large area (South American continent) 2. The dynamic evolution of the 2D OI 135.6 nm radiance observed by GOLD revealed that the postsunset enhancement at CP was caused by the southward movement of the southern EIA crest that passed over CP. The EIA southern crest moved 5° southward between 22:25 and 23:55 UT. Therefore, it was a result of a spatial variation instead of a local temporal variation. This is different from the explanation provided by previous observational studies, which is the postsunset enhancement of E × B drift. This finding provides new insight into the postsunset enhancement of N m F 2 in the low latitude ionosphere 3. The corresponding EIA northern crest did not show such apparent motion during the same time range, indicating asymmetric movement of EIA crests. Compared with previous studies of climatological EIA asymmetry, this EIA asymmetric movement occurring in a short time (<2-h) includes not only different plasma densities but also different motions. And it can only be captured by GOLD due to the difficulty in situating the local observation equipment in the ocean and remote area of the continent 4. The detailed mechanism of this EIA asymmetric movement is still unknown, but shall be the combination of chemistry, E × B transport, neutral wind transport and ambipolar diffusion. The geomagnetic activity may play minor roles Further understanding relies on future observational and theoretical studies. More equipment such as Fabry-Perot interferometer (FPI) (measuring neutral wind around 250 km altitude) and meteor radars (measuring neutral wind between 80 and 100 km altitudes) are needed in this area to know the behavior of neutral wind. On the other hand, numerical models with improved descriptions of geomagnetic activity and low atmosphere forcing to better reflect the ionosphere daily conditions are also necessary. Figure A1 shows      10.1029/2020JA028108