Emergence of a localized total electron content enhancement during the severe geomagnetic storm of September 8, 2017

Abstract. In this work, the results of the analysis on total electron content (TEC) data before, during and after the geomagnetic storm of September 8, 2017 are reported. One of the responses to geomagnetic storms due to the southern vertical interplanetary magnetic field (Bz) is the enhancement of the electron density in the ionosphere. Vertical TEC (VTEC) from the Center for Orbit determination in Europe (CODE) along with a statistical method were used to identify positive and/or negative 5 ionospheric storms in response to the geomagnetic storm of September 8, 2017. When analysing the response to the storm of September 8, 2017 it was indeed possible to observe an enhancement of the equatorial ionization anomaly (EIA); however what it was unexpected, was the identification of a local TEC enhancement (LTE) to the south of the EIA (∼40◦ S, right over New Zealand and extending towards the south-eastern coast of Australia and also eastward towards the Pacific). This was a very transitory LTE that lasted approximately four hours, starting at ∼02:00 UT on September 8 where its maximum 10 VTEC increase was of 241,2%. Using the same statistical method, comparable LTEs in a similar category geomagnetic storm, the 2015 St. Patrick’s day storm, were looked for. However, for the aforementioned storm no LTEs were identified. As also indicated in a past recent study for a LTE detected during the August 15, 2015 geomagnetic storm, an association between the LTE and the excursion of Bz seen during the September 8, 2017 storm was observed as well. Furthermore, it is very likely that a direct impact of the super-fountain effect along with travelling ionospheric disturbances may be playing an important role in 15 the production of this LTE. Finally, it is indicated that the September 8, 2017 LTE is the second one to be detected since the year 2016.

The statement By analyzing the latitudinal profiles, it could be determined that the increment of TEC to produce this LTE was of 241.2% (pg 7, lines 5-6) makes no sense. As far as I understand, LTE is a TEC increment. If you mean to show the LTE intensity, just present TEC value. Currently it is unclear what do you take as a 100%.
Dear editor, indeed, LTE is basically a TEC increment. Therefore, the last sentence of the first paragraph of the conclussions section was slightly changed to indicate better this fact. The increment is calculated with respect to the average of a stable day (being the average 100%). For this case an average of the curves for September 7 and 9, 2017 was taken.
Figures 1 and 5 are to be edited since they take a significant space and do not give too much information. Add a grid and more frequent tick marks to Y-axes for Bz and Dst panels.
Dear editor, a grid was added in every subplot of the two figures and also tick marks in the Y-axes for Bz and Dst were made more frequent Pictures of differential VTEC (figs. 2 and 3) looks better, but need higher contrast. Try to use colormap seismic from matplotlib package.
Dear editor, colormap changed to seismic in figures.2, 3, and 6 as per requested.

Introduction
Anomalies in the ionosphere can be product of different natural phenomena (Afraimovich et al., 2013). For instance earthquakes can produce positive or negative ionospheric anomalies (e.g., Zakharenkova et al., 2008;Yao et al., 2012;Guo et al., 20 2015;Li et al., 2015;Sotomayor-Beltran, 2019), although such variations are expected to be localized within the earthquake's preparation region (Dobrovolsky et al., 1979). On the other hand, major changes in the ionosphere are caused by geomagnetic storms (e.g., Buonsanto, 1999;Danilov, 2013). The response of the Earth's ionosphere to the geomagnetic storms are known 1 as ionospheric storms. These ionospheric storms can disrupt technologies relying on transmission of radio frequencies (e.g., Buonsanto, 1999;Borries et al., 2015), and thus they can have an impact in the modern society in general.
In order to understand better ionospheric variability in time and space produced by geomagentic storms, Global Navigation Satellite System (GNSS) receivers, due to its global coverage, are used as one of the tools for ionospheric studies. According to several studies (e.g., Huang et al., 2005;Mannucci et al., 2005;Astafyeva, 2009), one common response to a geomagnetic 5 storm due to the excursion of the southward interplanetary magnetic field is the significant increment in the equatorial and midlatitude total electron content (TEC), which manifests as an enhancement of the equatorial ionization anomaly (EIA; Appleton, 1946;McDonald et al., 2011). Such increase of TEC in the EIA is possible to visualize in global ionospheric maps (GIMs).
Besides changes in the EIA, it was recently observed by Edemskiy et al. (2018) and Sotomayor-Beltran (2018) that localized TEC enhancements (LTEs) can also emerge as a response to geomagnetic storms.

10
In this paper vertical TEC (VTEC) maps, also known as global ionospheric maps (GIMs), due to its reliability on ionospheric information (Hernández-Pajares et al., 2009), were used to analyse the response to the geomagnetic storm of September 8, 2017. Section 2 introduces the ionospheric data and the technique for the corresponding analysis. In Sect. 3 the results and the discussion are presented. Section 4 presents the final remarks or conclusions.

15
VTEC maps were downloaded via ftp 1 from the Center for Orbit Determination in Europe (CODE) between August 21, 2017 and September 20, 2017. VTEC maps, which have a resolution of 2.5 • x 5 • (latitude and longitude, respectively), come in daily IONnosphere Map EXchange files (Schaer et al., 1998) and they are produced every hour. Due to the format of the IONEX files, which consists of headers and the actual VTEC data, a code entirely written in Python was implemented for this work.
Using the NumPy 2 library, which handles relatively easily N-dimensional arrays, the VTEC data was stored in a 3D cube for 20 further analysis. The x, y and z axes in the 3D cube are longitude, latitude and number of maps, respectively.
In order to indentify ionospheric anomalies a running window of 8 days to every cell in the 3D VTEC cube was applied (e.g., Liu et al., 2004;Zhu et al., 2010;Zou and Zhao, 2010;Li et al., 2015;Sotomayor-Beltran, 2018). Assuming that for each cell or line-of-sight, the VTEC follows a Gaussian distribution, the mean (µ) VTEC and its associated standard deviation (σ) are calculated in order to define the upper and lower bounds: If a VTEC value for a certain day at a particular time falls above the U B, then a positive ionospheric anomaly is detected with a confidence level of 95%. The difference between the VTEC and U B or LB is defined as differential VTEC (∆VTEC).
On the other hand, if the VTEC falls bellow the LB, then a negative anomaly is detected. In this way, a cube of ∆VTEC was created, with a total of 744 maps. If U B > VTEC > LB, then ∆VTEC = 0 Some important geomagnetic parameters were also needed to be taken into account for the analysis. The Dst index (Sugiura,5 1964) provides information about the strength of the ring current around the Earth. According to Loewe and Prölss (1997)  is a strong southward B z for more than 3 hours a geomagnetic storm is in development (Gonzalez et al., 1994;Liu and Li, 2002). Hourly averages for B z also for the month of September 2017 were dowloaded from the OMNI datasabe 5 . In Fig The origin of this geomagnetic storm lies in the coronal mass ejection (CME) that occurred on September 6, 2017 at ∼12:40 25 UT. This CME was observed with the Camera 2 of the Large Angle and Spectrometric Coronograph on board of the Solar and Heliospheric Observatory (SOHO 7 ). Figure 1 also shows that on September 8 at ∼00:00 UT the vertical interplanetary magnetic field decreased significantly to a minimum of -24 nT. One hour before (September 7 at 23:00 UT), B z already decreased considerably to -20.6 nT, time of the storm sudden commencement (Fig. 1). In addition, it can be noticed that almost simultaneously with the drastic change of B z , the Dst index reached its peak at 01:00 UT on September 8, 2017. As it is already well-known, this relationship between B z and the Dst index hints to a physical response of the ring current in the magnetosphere to the interplanetary field B z (Patel and Desai, 1973;Gonzalez and Echer, 2005).

GIM maps
In the left column of Fig. 2, GIMs for September 7, 8 and 9, 2017 at 02:00 UT are presented. It is clearly seen in the GIM 5 of September 8 at 02:00 UT (just three hours after the storm sudden commencement) that the VTEC was enhanced in the EIA region with respect to the day before (September 7) and the day after (September 9) at the same hour. A recent study by Lei et al. (2018), using diverse instruments (e.g., satellites and ionosondes), has also observed this TEC enhancement in

Differential VTEC maps
What it was quite compelling was the detection of a ionospheric localized anomaly (∼40 • S), or as named by Edemskiy et al.
(2018) a localized TEC enhancement (LTE), to the south of the southern conjugate geomagnetic region of the EIA. This LTE can be identified in the GIM map of September 8, 2017 at ∼02:00 UT (Fig. 2). In the right column of Fig. 2, ∆VTEC maps for 15 September 7, 8 and 9, 2017 at 02:00 UT are also presented. It can be seen from these ∆VTEC maps that a day before and after that the LTE appeared, no anomalies were visible. However as already indicated, the day that the ionospheric storm occurred (September 8), the dramatic enhancement of the VTEC to the south of the EIA, manifested as a LTE, was observed.
In Fig. 3 the dynamics of the LTE can be clearly seen. It can be noticed that this LTE was very transitory, in the ∆VTEC maps it appeared at ∼02:00 UT on September 8 and at ∼06:00 UT it was already gone. This unforseen positive ionospheric 20 storm covers most of New Zealand and extends westward towards the south-eastern part of Australia and eastward towards the Pacific. The maximum peak of this LTE happened as well on September 8 at 02:00 UT with ∆VTEC = 6.47 TECU (where 1 TECU = 10 16 electrons/m 2 ).

Shape of the EIA
To better visualize this LTE to the south of the EIA, the shape of the VTEC along the meridional line of 170 • E is shown in 25 Fig. 4 between September 7 and 9, 2017 at 02:00 UT. From the ∆VTEC maps (Fig. 2), it can be confirmed that the EIA follows its normal variability one day after (September 9 at 02:00 UT) and before (September 7 at 02:00 UT) that the storm occurred (no anomalous VTEC enhancements are visible). However, on September 8 at 02:00 UT the EIA was significantly enhanced and hence this translated in a much sharper definition of the double-crest with a trough shape observed in Fig. 4. This shape is expected because when the LTE is above New Zealand, it is still day time, the local time is 14:00 (02:00 UT). In addition to the 30 two crests from the EIA, a third one in the southern hemisphere is visible (Fig. 4). This third crest is simply the LTE observed 4 in the ∆VTEC and GIM maps for September 8 at 02:00 UT ( Fig. 2 and 3). The peak increment for this day and this time in the southern crest of the EIA is of 172% and in the LTE of 241,2%. Edemskiy et al. (2018) have also reported for the August 15, 2015 G3 geomagnetic storm that the two LTEs they observed were located to the south of the EIA (between Africa and Antarctica), whereas Sotomayor-Beltran (2018) has also identified to the south of the EIA a LTE over the Indian ocean during the G2 moderate storm of April 20, 2018.  These TADs, which originate in the polar regions, can transport equatorward winds that drive plasma upwards in the middle and lower latitudes and as a consequence the ionosphere moves to higher altitudes (Chen et al., 2016). It is very likely then, as suggested by Lei et al. (2018), that the combined effect of TADs and the PPEFs are responsible for the creation of the LTE observed in Fig. 2 and 3. As per to the overall enhancement of the EIA (Fig. 2 and 3) and shifting of the crests in the direction of the poles observed in Fig. 4, as previously mentioned and suggested by many studies (e.g., Tsurutani et al., 2004;Mannucci et al., 2005;Astafyeva, 2009;Astafyeva et al., 2014;Chakraborty et al., 2015) the mechanism at work is the ionospheric super-5 fountain effect. Finally, it is also worth mentioning that this would be the second time a LTE is detected since 2016, as the first one was the one observed during the April 20, 2018 geomagnetic storm (Sotomayor-Beltran, 2018).