Red Line Diffuse‐Like Aurora Driven by Time Domain Structures Associated With Braking Magnetotail Flow Bursts

Magnetotail earthward‐propagating fast plasma flows provide important pathways for magnetosphere‐ionosphere coupling. This study reexamines a flow‐related red‐line diffuse‐like aurora event previously reported by Liang et al. (2011, https://doi.org/10.1029/2010ja015867), utilizing THEMIS and ground‐based auroral observations from Poker Flat. We find that time domain structures (TDSs) within the flow bursts efficiently drive electron precipitation below a few keV, aligning with predominantly red‐line auroral intensifications in this non‐substorm event. The diffuse‐like auroras sometimes coexisted with or potentially evolved from discrete forms. We forward model red‐line diffuse auroras due to TDS‐driven precipitation, employing the time‐dependent TREx‐ATM auroral transport code. The good correlation (∼0.77) between our modeled and observed red line emissions underscores that TDSs are a primary driver of the red‐line diffuse‐like auroras, though whistler‐mode wave contributions are needed to fully explain the most intense red‐line emissions.


Introduction
Since their discovery in the 1970's (Gurnett et al., 1976;Scarf et al., 1974), broadband electrostatic noise in the frequency range of tens of Hz up to several kHz have been observed to be ubiquitous in various space plasmas (Cattell et al., 2005;Ergun et al., 1998;Williams et al., 2006).These fluctuations, when examined in the time domain (Matsumoto et al., 1994), manifest unique localized electrostatic solitary structures, comprising mostly electron phase-space holes and double layers, which are collectively termed as time-domain structures (TDSs) (Mozer et al., 2015).
Although individual conjunction events have revealed moderate correlations between TDSs and diffuse auroras (see Supporting Information in Shen et al., 2020), and their similar statistical distributions in the nightside plasma sheet lend support (Malaspina et al., 2015;Newell et al., 2009), direct evidence remains elusive.
Prior statistical studies revealed that TDSs are abundant in the plasma sheet within most plasma injections and braking fast ion flows or bursty bulk flows (BBFs) (Angelopoulos et al., 1992;Ergun et al., 2015;Gurnett & Frank, 1977;Malaspina et al., 2015).These fast ion flows often manifest in the ionosphere as north-south auroral structures or "streamers" (e.g., Henderson, 2012;Lyons et al., 2012;Nishimura et al., 2011).Auroral streamers are typically discrete in nature and are linked to upward field-aligned currents forming on the western edge of flow bursts (Nakamura et al., 2001;Nishimura et al., 2011).However, some streamers may coexist with or evolve into diffuse aurora enhancements as they move equatorward and reach the inner edge of the plasma sheet (Henderson et al., 1998).Within the braking flows, TDSs may contribute to electron precipitation into structured diffuse and non-accelerated auroras (Sergeev et al., 2004;Shen et al., 2020).
Low-energy (<∼1 keV) electron precipitation more effectively excites red-line (630 nm) auroras, whereas higherenergy (>1 keV) electrons predominantly induce green-line (557.7 nm) and blue-line (427.8 nm) auroras (Liang et al., 2016;Paschmann et al., 2002).Thus, TDSs are likely linked to red-line-dominated auroras, especially when the plasma sheet state is prime for such low-energy precipitation, like quiet times with low temperatures.In fact, to reliably identify TDS-driven diffuse auroras, non-substorm conditions are more suitable, as it is often challenging to disentangle aggregated effects from different, concurrent drivers during substorms that preclude direct linkage to flow bursts and TDSs (e.g., Sergeev et al., 2012;Shen et al., 2023).
One potential example has been reported by Liang et al. (2011), where conjugate THEMIS and ground-based optical and radar observations revealed flow-related auroral signatures, including red-line diffuse-like auroras that are associated with discrete forms.In that study, THEMIS waveform data and potential TDS signatures have not been investigated, leaving the impact of TDS-induced electron scattering on red-line auroras unresolved.In this paper, we revisit this event, aiming to provide the first forward modeling and data-model comparison of redline diffuse auroras driven by TDSs.

Instrumentation and Models
We use the following data from THEMIS (Angelopoulos, 2008): electron and ion fluxes and moments measured by the Electrostatic Analyzers (ESA) and the Solid State Telecope (SST) instruments in the energy range of several eV up to 900 keV (Angelopoulos et al., 2008;McFadden et al., 2008), DC vector magnetic fields at spin resolution (∼3 s) measured by the Fluxgate Magnetometers (FGM) (Auster et al., 2008), electric and magnetic field wave power spectra within 1 Hz-4 kHz (FBK with 6 frequency bands and FFP with 64 bands), and waveform data at 8,192 samples per second (sps, DC-coupled), measured by the Electric Field Instrument (EFI), the search coil magnetometer (SCM), and the Digital Fields Board (DFB) (Bonnell et al., 2008;Cully et al., 2008;Le Contel et al., 2008).
The flow-related red-line auroral event occurred during 10:30-11:30 UT on 3 March 2009.Following Liang et al. (2011), we primarily use the ground-based Poker Flat meridian spectrograph (PFMSP) for auroral observations, recording every 15 s and capturing emissions in the red (630 nm), green (557.7 nm), blue (427.8nm), and proton H β (486.1 nm) lines along the meridian.Complementarily, the Poker Flat digital all-sky camera (DASC) provided 2D auroral data to elucidate the longitudinal or magnetic local time (MLT ) extent of the red-line diffuse auroras.Note that the DASC does not provide absolute intensities of 630 nm emissions and it recorded images at 41-s cadence for this event.We also use auroral observations from the THEMIS whitelight all-sky imager (ASI) at Fort Yukon with 3-s cadence (Mende et al., 2008).
Although TDSs include different types of nonlinear electrostatic structures as noted by Mozer et al. (2015), electron phase space holes are the predominant type in the near-Earth plasma sheet (e.g., Ergun et al., 2015;Malaspina et al., 2018).Electron scattering by electron holes can be quantified using the quasi-linear approach initially developed for an ensemble of plane waves (Kennel & Engelmann, 1966;Lyons, 1974), assuming statistically uniform, stationary, and weak TDS turbulence and electron random gyrophases during interactions (Vasko et al., 2018).We use the formulation of local pitch-angle diffusion coefficients D αα derived in Vasko et al. (2018) and refined by Shen et al. (2021).We bounce average the local diffusion rates using the standard procedure (Glauert & Horne, 2005;Lyons et al., 1972): Geophysical Research Letters where the integration is over the period of bounce motion τ B , between the south and north mirror points λ m,south and λ m,north , and s = s(λ) is the length of a field line.We use the relatively realistic T89 magnetic field model for bounce-averaging (Ma et al., 2012;Tsyganenko, 1989).We modify the latitudinal profile of B near the equator using a spline function fit to match the locally measured B by THEMIS.As a result, the magnetic field stretching is reduced near the equator (e.g., Ni et al., 2012).In addition, bounce-averaged pitch angle diffusion coefficients due to transient whistler-mode waves are also calculated using the Full Diffusion Code (Ma et al., 2020;Ni et al., 2008), including the Landau (n = 0) and higher order (n up to ±10) cyclotron harmonic resonances.
We use the full precipitating electron distributions calculated from quasilinear diffusion theory to forward model TDS-driven red-line auroras.Because the varying timescale of precipitation is typically shorter than the radiative timescale (∼110 s) of 630 nm emission, it is imperative to use a time-dependent auroral transport code, such as the TREx-ATM model (Liang et al., 2016).TREx-ATM is a time-dependent ionosphere aurora model, adopting the two-stream electron transport code embedded in the GLOW model (Solomon et al., 1988) and ambipolar diffusion to compute the electron transport in the atmosphere, with additional capabilities to compute the impact ionization, secondary electron production, and impact excitation of neutrals.

Results
The event featuring bursty bulk flows and red-line auroras was observed from 10:40 to 11:20 UT on 3 March 2009.During this time, THEMIS-C spacecraft detected a series of high-speed flow bursts (>400 km/s) in the Earth's plasma sheet, as shown in Figure 1.Figures 1a-1c illustrate the magnetic fields threading all five THEMIS spacecraft, modeled by T89, alongside the 2D ionospheric red-line auroras observed by Poker Flat DASC.The red-line auroras spanned nearly 45°in magnetic longitude or 3 hr in MLT, primarily drifted eastward, occasionally moving equatorward, and exhibited both diffuse-like and discrete enhancements.
For this event complementary ground-based measurements, including THEMIS ASIs and Poker Flat incoherent scatter radar, provided synergistic auroral and ionospheric density observations as detailed by Liang et al. (2011).
While not described again here, the key findings include: (a) flow-related auroras featured intermittent poleward boundary intensifications (PBIs) and streamer-like green line emissions, occasionally coinciding with or transitioning to predominantly red-line diffuse-like auroras at lower latitudes; (b) ionospheric density altitude profiles were consistent with the red-line emissions and soft (<1 keV) electron precipitation; (c) southward-moving density patches corresponded to earthward flow bursts and red-line auroral enhancements.It was noted that the absence of substantial density enhancements above 200 km and the latitudinally extended red-line diffuse-like emissions were inconsistent with typical small-scale Alfvénic auroras and broadband acceleration (Chaston et al., 2003;Liang et al., 2019).However, for some observed discrete auroras, such as the bright stripes around 100°magnetic longitude and discrete structures near the poleward boundary (likely linked to PBIs) in Figure 1c, Alfvénic acceleration cannot be ruled out (Damiano et al., 2015;Hull et al., 2022;Tian et al., 2021), though these aspects are not the main focus of this study.Auroral features will be further examined using multi-spectra PFMSP measurements.
Figures 1b and 1c indicate that of the five THEMIS spacecraft, THB and THC were closest to midnight, with their footprints less than 1 hr MLT east of Poker Flat.THC detected all flow bursts exceeding 400 km/s within BBFs, whereas THB, positioned further earthward, registered only a weaker flow burst (∼200 km/s) that had waned during inward propagation.Figure 1d shows that the AE index remained below 80 nT throughout the event.Three strong flow bursts were recorded between 10:40 and 11:20 UT, each lasting ∼5 min.The latter two, occurring closer to the equator, were accompanied by |B x | decreases, notable B z dipolarizations (Figure 1e), and slightly energized electron spectra (Figure 1h).Despite heightened flow activity, the plasma sheet temperature remained below 1 keV (Liang et al., 2011).Figures 1i and 1j show that the flow bursts corresponded to enhanced broadband electrostatic fluctuations (or TDSs) with frequencies ranging from tens of Hz to several kHz.Additionally, sporadic electromagnetic wave power enhancements, mostly below ∼20 Hz, indicate the presence of kinetic Alfvén waves (KAWs) (e.g., Chaston et al., 2012;Malaspina et al., 2018;Shen et al., 2023).
Figure 2 provides a detailed view of wave spectra and waveform measurements associated with two flow bursts near 10:58 UT and 11:17 UT.Throughout the flow intervals, there were continuous enhancements in broadband electrostatic fluctuations.Figures 2d, 2e, 2i, and 2j clearly show that these broadband electrostatic fluctuations predominantly consisted of electron phase-space holes, identified by bipolar parallel (in the z direction) and unipolar perpendicular (in the x and y direction) electric fields on the timescales of milisecond or on the spatial scales of the local Debye length (Hutchinson, 2017;Lotekar et al., 2020;Muschietti et al., 1999).In contrast, whistler-mode waves were more ephemeral, characterized by sporadic, narrow-band magnetic power enhancements lasting less than 1 minute within the flow intervals (Figures 2b and 2g).
Based on the observed characteristics of TDSs and whistler-mode waves, we proceed to calculate bounceaveraged electron pitch-angle diffusion rates 〈D α eq α eq 〉 and resulting electron distributions precipitating into the ionosphere (Ma et al., 2012(Ma et al., , 2020;;Shen et al., 2021).For TDSs, local diffusion coefficients do not depend on individual electron hole electric fields, but on the root-mean-square intensity E w derived from spectra data, which take into account the occurrence of electron holes (Shen et al., 2021;Vasko et al., 2018).Following Shen et al. (2021), we also adapt statistical electron hole velocity and parallel scale distributions based on the local plasma parameters for diffusion rate calculations.TDSs can be generated locally at various latitudes and continuously along the propagation of flow bursts and injections toward Earth (Ergun et al., 1998;Franz et al., 2005;Malaspina et al., 2015;Mozer et al., 2015).We assume a latitudinal distribution of electron holes to be within ±25°near the equator (Shen et al., 2021;Vasko et al., 2017); increasing the latitude extent to ±45°enhances the diffusion rate near the loss cone by <20%.
For transient whistler-mode waves, we determine the average wave spectra near 11:15 UT (Figure 2g) as having a Gaussian with a lower limit ω 1 = 0.15Ω ce , a mean frequency ω 2 = 0.3Ω ce , an upper limit ω m = 0.45Ω ce , a bandwidth δω = 0.075Ω ce , and a mean wave amplitude B w = 50 pT.No whistler-mode waves were measured by waveform data in our event, thus we do not have direct information on their wave normals.For the nightside Figure 2. Zoomed-in view of the plasma waves within the two flow bursts near 10:58 UT (a-e) and 11:17 UT (f-j).Panels from top to bottom are: ion flow velocities, FFP magnetic field spectra, FBK electric field spectra, waveform measurements of high-frequency (>20 Hz and >50 Hz) electric fields in the field-aligned coordinates (B is in the z direction), and expanded views of milisecond-scale electron phase-space holes, also known as time-domain structures (TDSs).equatorial plasma sheet (Agapitov et al., 2013;Li et al., 2011;Meredith et al., 2021), we can assume parallel whistlers confined within ±15°latitude near the equator, having Gaussian wave normals with a width θ w = 10°, a minimum θ min = 0°, and a maximum θ max = 30°.We set the background n e ∼ 0.6 cm 3 , f pe /f ce ∼ 14.4, and L ∼ 14.0 based on THEMIS observations.For bounce averaging, we use the T89 model with slight adjustments near the equator to align with THEMIS-C local B observations.Applying different magnetic field models, such as T96 and T01, we have obtained variations of less than a factor of 2 in 〈D α eq α eq 〉 near the loss cone for TDS and whistler-mode waves (Ma et al., 2012), which further translates to a variation in precipitating flux of less than 40%.
Figures 3a and 3c present the calculated 〈D α eq α eq 〉 as a function of energy and pitch angle, resulting from electron scattering by TDSs and whistler-mode waves, respectively.Near the loss cone (∼1°), TDSs efficiently induce pitch-angle scattering at energies below a few keV, with 〈D α eq α eq 〉 on the order of 10 4 -10 3 s 1 , assuming E w ∼ 1 mV/m, compared with the strong diffusion limit of D sd ∼ 10 5 s 1 at L ∼ 14.0 (Kennel & Petschek, 1966).At any instance during the event, we may obtain 〈D α eq α eq 〉 by multiplying a factor of E 2 w based on quasilinear theory (Kennel & Engelmann, 1966;Vasko et al., 2018).Due to large f pe /f ce for this event, whistler-mode waves can also drive efficient electron scattering at energies as low as 200 eV, but the scattering rates exhibit a broad  peak near 1 keV, extending up to 100 keV.This suggests that whistler-mode waves tend to produce more energetic precipitation spectra compared with TDSs.
Example precipitating electron energy distributions are shown in Figures 3b and 3d.The differential energy flux (red pluses) within the loss cone can be estimated as x(E)J(E,α LC ) , where being the index of loss cone filling, J(E, α LC ) is the electron differential energy flux at the edge of the loss cone (∼1°, blue stars), I 0 is the modified Bessel function with an argument Petschek, 1966), and τ loss is assumed to be half of the bounce period.
Simultaneously measured time-varying precipitating electron full distributions are input into the time-dependent TREx-ATM code to forward model red-line auroras and compare their evolution with that of the measured redline diffuse-like auroras by PFMSP.We begin by examining PFMSP auroral observations.Figures 4a-4d display approximately correlated ion flow bursts, TDS E w , and enhanced auroral activities in the green line (577.7 nm), blue line (427.8nm), and red line (630 nm).To obtain wave amplitudes of TDSs from FBK spectra (Figure 2), we integrate wave power in the frequency range of 20-900 Hz, as informed by TDS wavelet analyses (Shen et al., 2021).The flow bursts (>400 km/s) were associated with intermittent poleward boundary intensifications (PBIs) near 160°elevation angle (or ∼68°magnetic latitudes) in the green and blue line emissions, exhibiting distinct equatorward movement.These features align with auroral streamers associated with earthward flow channels in the plasma sheet (e.g., Henderson, 2012;Lyons et al., 2012;Nishimura et al., 2011).
However, the majority of the auroral enhancements were noted in the red line, predominantly equatorward of the green-line enhancements.These enhanced red-line emissions, characterized by "patchiness" or short-lived activations, coincided with TDS amplitude increases and exhibited discernible equatorward movement.As noted by Liang et al. (2011), this movement was also in accord with the equatorward shift of soft electron precipitation and enhanced density patches at altitudes above 150 km. Figure 4f suggests that TDS-driven precipitation has characteristic energies mostly below 1 keV, consistent with the overwhelmingly red-line emissions and soft electron precipitation.In contrast, assuming the presence of whistler-mode waves throughout the flow bursts near 10:58 UT and 11:17 UT, these would produce precipitation with characteristic energies exceeding 1 keV.
To compare with TREx-ATM model output, we need to identify PFMSP red-line diffuse-like aurora intensifications that were approximately conjugate with THEMIS observations.First, we focus on the time interval between 10:45 and 11:30 UT, during which the longitudinal separation between the PFMSP meridian and projected THEMIS-C footprints was less than 15°, or equivalently |ΔMLT| < 1 hr (Figure 4g).Second, we adjust the PFMSP red-line data counts by subtracting a uniform background attributed to ambient diffuse auroras, likely arising from non-TDS mechanisms like diffuse proton precipitation (e.g., Lyons et al., 2015) as indicated by PFMSP H β -line emissions.Third, after applying Van Rhijin correction to data counts for aspect angle effects, we identify an optimum range of elevation angles of 139°-146°, which probably corresponded to THEMIS measurements.These optimal angles were determined by searching the maximum correlation coefficients between the modeled and observed red-line emissions across all elevation angles (Figure 4k).
The event exhibited coexisting auroral types due to various mechanisms.TDS-induced soft electron precipitation leads to a broader altitude distribution of both 630 and 557.7 nm emissions, unlike the narrow altitude range associated with monoenergetic electron acceleration in discrete arcs.In oblique optical observations, like those from PFMSP and THEMIS ASI, this broader distribution may appear as more "diffusive" auroral forms (see Figure 4j around the PFMSP meridian).Additional ASI auroral images are provided in Supporting Information S1.The 630-nm emission in our event, particularly sensitive to soft electron precipitation, aligns closely with TDS effects.We have carefully examined the observations from the Fort Yukon ASI (see Supporting Movie in Liang et al. (2011)) and excluded three periods when discrete arcs predominately contribute to PFMSP red-line emissions.
Figure 4h shows a correlation coefficient of 0.77 between PFMSP observed and TREx-ATM forward-modeled red-line emissions, using inputs solely from TDS-driven precipitation.The TREx-ATM model results, indicated by the black and gray lines, are based on maximum and minimum flux levels measured by THEMIS near the equator during 10:40-11:20 UT.These in-situ measured fluxes were largely controlled by the spacecraft's proximity to the true equator, where |B x | is minimum (see Supporting Information S1). Figure 4h reveals that the majority of the observed red-line diffuse-like auroral enhancements during flow bursts can be explained by TDSdriven model results, albeit slightly underestimated for the most intense emissions.Green bars highlight the time intervals of discrete emission peaks, unrelated to TDSs, near 11:00, 11:05, and 11:10 UT.Including these intervals in our analysis lowers the correlation to 0.67. Figure 4i suggests that including potential whistler-mode wave effects along with TDSs in the model offers a better explanation for the two bright red spots near 10:58 UT and 11:17 UT.This implies that whistler-mode waves might also contribute to the most intense red-line diffuse emissions during flow bursts.The compound effects of whistler-mode waves and field-aligned acceleration on energetic electron precipitation obscure TDS's role in 557.7-nm emissions.Nonetheless, we have analyzed TDS-driven TREx-ATM 557.7-nm outputs and observed a weaker correlation with PFMSP's green line (∼0.6).

Discussion
Plasma sheet flow bursts are typically linked to ionospheric auroral PBIs and equatorward-moving auroral streamers, as evidenced by white-light all-sky imager (ASI) observations, often around substorm times (Donovan et al., 2008;Forsyth et al., 2020;Nishimura et al., 2011;Yadav et al., 2022).In our non-substorm event, while green-line PBIs and forms resembling equatorward-moving auroral streamers were noted, the most prominent feature was the red-line auroras.These auroras extended toward lower latitudes and often appeared more diffuse than discrete near PFMSP, likely due to wave scattering.Liang et al. (2011) suggests that the predominance of red-line emissions was probably due to a cooler inner plasma sheet (T e < ∼1 keV).Despite strong flows and potential association with hot electrons in the tail leading to streamer-inducing energetic precipitation (Angelopoulos et al., 1992;Nishimura et al., 2020), there was an absence of significant energetic injections intruding the inner plasma sheet around midnight (Figure 1g).This could be attributed to differences in adiabatic motion or Alfvén layers between soft and energetic electrons as they undergo convective, curvature and gradient drifts during inward propagation (e.g., Gabrielse et al., 2017).While red-line emissions naturally accompany green-line emissions due to the O( 1 S) to O( 1 D) transition linked with 557.7 nm photo emission (Solomon et al., 1988), the dominance of red-line auroral emissions indicates a major low-energy precipitation population, distinct from the energetic precipitation responsible for green line excitation.This low-energy precipitation was confirmed by ionospheric density altitude profiles measured by PFISR (Liang et al., 2011).During the event period of our interest, ECH waves were not observed, ruling them out as a cause for the auroral precipitation (Ni et al., 2012;Zhang et al., 2015).Our investigations point to electron scattering by TDSs as the primary driver of the observed low-energy precipitation and red-line diffuse-like auroras.

Conclusion
This letter revisits a flow-related red-line aurora event initially reported by Liang et al. (2011).Analyzing THEMIS spectra and waveform measurements, we have revealed an abundance of time-domain structures (TDSs) associated with magnetotail flow bursts.Applying quasilinear calculations, we have estimated TDSdriven electron distributions precipitating into the ionosphere.Employing the time-dependent TREx-ATM auroral transport code, we have forward-modeled red-line auroras due to TDS-driven precipitation.The good correlation of ∼0.77 between modeled and observed red-line emissions suggests that TDSs likely cause the observed red-line diffuse-like auroras, which sometimes coexisted with or likely evolved from discrete forms.However, to fully explain the most intense red-line emissions, contributions from whistler-mode waves are also necessary.Our study suggests that wave-driven red-line diffuse auroras, associated with braking flow bursts, offer a distinct yet complementary pathway of MI-coupling compared with well-established flow-driven discrete auroral streamers, typically seen in the green line and caused by field-aligned potential acceleration.

Figure 1 .
Figure 1.Overview of the flow-related red-line diffuse aurora event observed by THEMIS-C and Poker Flat all-sky camera (DASC).(a, b) Field line configuration threading the five THEMIS spacecraft in the GSM coordinates.Colored vectors represent ion flow velocities for each spacecraft.(c) Poker Flat DASC red-line auroras measured near 11:17 UT.The grids represent AACGM magnetic latitude and magnetic longitude, while the blue line marks the actual local midnight.The assumed emission height and approximate footprints of THEMIS-B (red box), THEMIS-C (green box), and Poker Flat have been mapped to ∼200 km altitude using different Tsyganenko models of T89, T96, and T01.(d) THEMIS AE index.(e) Magnetic fields.(f) Ion flow velocities.(g, h) Electron energy spectrogram from 7 eV up to 900 keV.(i) Magnetic field FFP spectra in 10 Hz-2 kHz.(j) Electric field FBK spectra up to 2 kHz.

Figure 3 .
Figure 3. (a, c) TDS-and whistler-driven bounce-averaged electron pitch-angle diffusion rates (〈D α eq α eq 〉) as a function of energy and pitch angle.We use adjusted T89 model for bounce averaging.(b, d) Example energy flux distributions extrapolated to the loss cone (blue) and calculated within the loss cone (red) based on 〈D α eq α eq 〉.The gray-dashed curves indicate Maxwellian fittings of the precipitating distributions due to TDSs and whistlers, indicating characteristic energies of 1.5 and 1.8 keV near 11:15:18 UT.

Figure 4 .
Figure 4. Auroral features and TREx-ATM model results.(a) Ion flow velocity V x .(b) Root-mean-square TDS electric field amplitudes |E w | (20-900 Hz).(c-e) PFMSP 557.7-nm green-line, 427.8-nm blue-line, and 630-nm red-line intensity; dashed lines mark optimal elevation angle range for THEMIS conjunction.(f) Characteristic energies of electrons precipitated by TDS (black) and whistler-mode waves (blue).(g) Geographic longitudes of PFMSP and THEMIS-C footprints.(h) 630-nm intensity: PFMSP data (red) versus TDS-driven TREx-ATM model results (black, gray) based on THEMIS maximum and minimum fluxes around the equator (see Supporting Information S1).(i) TREx-ATM results with additional inputs from assumed, persistent whistler-mode waves; only maximum electron fluxes used.(j) Example ASI whitelight image showing diffuse-like auroras near PFMSP that likely evolved from discrete auroras.(k) Correlation coefficients for PFMSP and TREx-ATM modeled TDSdriven red-line auroras at various elevation angles.