The Relationship of Exohiss Waves With Plasmaspheric Hiss Distribution and Solar Wind Parameters

Exohiss waves below 0.1 electron cyclotron frequency (fce) are structureless whistler‐mode emissions typically observed in the plasmatrough. Plasmaspheric hiss may possibly propagate from the plasmasphere into the plasmatrough and evolve into exohiss waves. We investigated the relationship of exohiss occurrence and characteristics with plasmaspheric hiss occurrence and solar wind parameters, analyzing Van Allen Probe observations from 1 October 2012 to 28 February 2018. Exohiss waves observed in the plasmatrough occurred more frequently on the dayside than the nightside, which was consistent with the plasmaspheric hiss distribution in the plasmasphere. Exohiss occurrence gradually increased up to ∼4 hr after hiss measurements and showed a magnetic local time dependence on the plasmaspheric hiss amplitude. We also determined the relative contribution of each solar wind parameter to exohiss distribution as based on exohiss measurements made 0–4 hr after plasmaspheric hiss measurements. A stronger southward interplanetary magnetic field (IMF) BZ limited the region of exohiss occurrence to the prenoon sector, again consistent with the distribution of plasmaspheric hiss. Prenoon exohiss was also observed for stronger dynamic pressure (PSW), but the plasmaspheric hiss appeared in the postnoon sector. This discrepancy indicated that prenoon exohiss is locally excited rather than a product of plasmaspheric hiss leakage. In addition, during enhanced solar wind conditions with southward IMF BZ or higher PSW, the intensity of the lower‐band chorus was enhanced even below 0.1fce, corresponding to the frequency range of the exohiss, implying that the nightside exohiss may be related to the evolution of low‐frequency chorus waves.


Introduction
The origin and characteristics of exohiss have been receiving increasing attention (Li et al., 2019;Wang et al., 2020;Zhu et al., 2015Zhu et al., , 2019)).Exohiss waves below 0.1f ce (where f ce is the electron cyclotron frequency) are structureless whistler-mode emissions typically observed on the dayside outside the Earth's plasmasphere (Wang et al., 2020;Zhu et al., 2019).They were first observed by the Ogo satellite outside the plasmasphere and were originally interpreted as a part of plasmaspheric hiss leaking from the plasmasphere (Thorne et al., 1973).Plasmaspheric hiss is another structureless whistler-mode emission around Earth, occurring in the frequency range of a few hundred hertz to several kilohertz (<0.1f ce ).However, in contrast to exohiss, it is usually observed in regions of high plasma density, such as inside the plasmasphere (Dunckel & Helliwell, 1969;Meredith et al., 2004;Thorne et al., 1973) or plasmaspheric plumes (Kim & Shprits, 2019;Summers et al., 2008;Zhang et al., 2019).Previous studies have suggested that exohiss can originate from plasmaspheric hiss inside the plasmasphere (Thorne et al., 1979;Wang et al., 2020;Zhu et al., 2015), functioning as an energy-dissipation mechanism.Additionally, exohiss waves contribute to the loss of relativistic electrons from the outer radiation belt (Li et al., 2019;Zhu et al., 2015).Zhu et al. (2019) were the first to investigate the statistical characteristics of exohiss via Van Allen Probe observations from 2013 to 2016.They found that exohiss waves exhibited a central frequency range of 200-300 Hz (which remained constant regardless of L-value) and a day-night distribution asymmetry (with exohiss occurring more frequently and with higher amplitude on the dayside).These findings are similar to those for plasmaspheric hiss.A recent study by Wang et al. (2020) provided a clearer correlation between exohiss and plasmaspheric hiss using Van Allen Probe A data from 2013 to 2015, revealing a similar occurrence of both in the 8-20 magnetic local time (MLT) region.They also showed that most exohiss waves outside the plasmapause propagated equatorward and that their amplitudes were generally smaller than those of hiss waves.This strongly supports the possibility that the leakage of hiss waves from the plasmasphere at higher latitudes serves as an exohiss origin.
Abstract Exohiss waves below 0.1 electron cyclotron frequency (f ce ) are structureless whistler-mode emissions typically observed in the plasmatrough.Plasmaspheric hiss may possibly propagate from the plasmasphere into the plasmatrough and evolve into exohiss waves.We investigated the relationship of exohiss occurrence and characteristics with plasmaspheric hiss occurrence and solar wind parameters, analyzing Van Allen Probe observations from 1 October 2012 to 28 February 2018.Exohiss waves observed in the plasmatrough occurred more frequently on the dayside than the nightside, which was consistent with the plasmaspheric hiss distribution in the plasmasphere.Exohiss occurrence gradually increased up to ∼4 hr after hiss measurements and showed a magnetic local time dependence on the plasmaspheric hiss amplitude.We also determined the relative contribution of each solar wind parameter to exohiss distribution as based on exohiss measurements made 0-4 hr after plasmaspheric hiss measurements.A stronger southward interplanetary magnetic field (IMF) B Z limited the region of exohiss occurrence to the prenoon sector, again consistent with the distribution of plasmaspheric hiss.Prenoon exohiss was also observed for stronger dynamic pressure (P SW ), but the plasmaspheric hiss appeared in the postnoon sector.This discrepancy indicated that prenoon exohiss is locally excited rather than a product of plasmaspheric hiss leakage.In addition, during enhanced solar wind conditions with southward IMF B Z or higher P SW , the intensity of the lower-band chorus was enhanced even below 0.1f ce, corresponding to the frequency range of the exohiss, implying that the nightside exohiss may be related to the evolution of low-frequency chorus waves.

10.1029/2023JA031777
2 of 13 Both plasmaspheric hiss and exohiss significantly depend on geomagnetic activity.Generally, with increasing geomagnetic activity, the occurrence and amplitude of plasmaspheric hiss are enhanced on the dayside (Meredith et al., 2004(Meredith et al., , 2007;;Orlova et al., 2014).Exohiss that originate from hiss can occur more frequently with increases in the Kp index; however, Landau damping dominates over the source of plasmaspheric hiss, thereby limiting its occurrence at Kp > 2 (Zhu et al., 2019).Kim et al. (2015) suggested that different aspects of plasmaspheric hiss distribution are based on past solar wind conditions (see also Golden et al., 2012;Kim et al., 2015Kim et al., , 2020)), which are not captured by geomagnetic indices such as AE and Kp.They suggested that two solar wind parameters, namely, the B Z component of the interplanetary magnetic field (IMF) and the solar wind speed V SW , contribute differently to the variation in hiss wave amplitude.Such a relationship would imply that exohiss, originating from the leakage of plasmaspheric hiss, can also respond to solar wind conditions.This study is the first to statistically investigate the dependence of exohiss occurrence on solar wind parameters and the relationship of exohiss with the occurrence of plasmaspheric hiss using Van Allen Probes measurements.
The remainder of this paper is organized as follows: Section 2 describes the Van Allen Probe data analysis; Section 3 presents the occurrence and characteristics of exohiss and plasmaspheric hiss; Sections 4 and 5 examine the relationship of exohiss occurrence with plasmaspheric hiss and solar wind parameters, respectively; our conclusions are provided in Section 6.

Data Analysis
First, we introduce a methodology for capturing exohiss waves via Waveform Receiver (WFR), a part of the Electric and Magnetic Field Instrument Suite and Integrated Science (EMFISIS) (Kletzing et al., 2013) suite onboard the Van Allen Probes.This instrument collects wave power spectral density (PSD) data covering the frequency range of 10 Hz-12 kHz.In this study, exohiss signals were defined as those with magnetic PSDs below 0.1f ce and located outside the plasmapause, as per Zhu et al. (2019).
The more detailed process of exohiss signal identification was as follows: First, the plasmapause location was identified based on plasma density values, as inferred from the upper hybrid resonance frequencies measured with the high-frequency receiver on EMFISIS (Kurth et al., 2015).The plasmapause was determined by requiring a density gradient by a factor of 15 within a radial distance of 0.5 L*, with the density level prior to the drop being ≥100/cm 3 at all L* (Cho et al., 2015).These criteria can be satisfied for multiple-density drops in a bound orbit.In such a case, the innermost location was selected as the plasmapause.Identification was performed at every half-orbit.Additionally, to supplement the plasmapause location, we estimated the ratio of the plasma frequency (f pe ) to the electron cyclotron frequency (f ce ) and assumed that the region where f pe /f ce < 10 was located outside the plasmapause (Meredith et al., 2006).The Roederer's L* value, as estimated based on the Tsyganenko magnetospheric TS04D model (Tsyganenko & Sitnov, 2005), was used as a radial distance measure.
Second, we identified other wave types using the singular value decomposition method (Santolík et al., 2003), distinguishing these waves from exohiss when the following criteria were satisfied (Li et al., 2015): (a) a plasmaspheric hiss in the frequency range of 20 Hz to 4 kHz inside the plasmapause, with a wave ellipticity of >0.7 and planarity of >0.2;(b) a whistler-mode chorus between 0.1f ce and 0.8f ce outside the plasmapause, with a wave ellipticity of >0.7 and planarity of >0.2; and (c) equatorial magnetosonic waves with a wave normal angle (WNA) of >80° and absolute ellipticity of <0.2, regardless of whether the probe was inside or outside the plasmapause.
Our analysis was based on WFR signals with the background noise removed using signal-to-noise ratio (SNR) thresholds.The SNR is defined by fitting a Gaussian distribution to the peak of the PSD distribution in the lowest frequency band and determining two standard deviations (Malaspina et al., 2017).We calculated SNR thresholds for each spacecraft and year.to 0.1f ce (dotted) and 0.8f ce (dashed), respectively.Figures 1h and 1i depict the root-mean-square amplitude (B w ) and power-weighted WNA, respectively.The values at time i were estimated using Equations 1 and 2: (1) 10.1029/2023JA031777 4 of 13 where f uc and f lc indicate the upper and lower limit frequencies, respectively.They were set to 20 Hz and 4 kHz for the plasmaspheric hiss and 70 Hz and 0.1f ce for the exohiss.The f lc of the exohiss was determined based on Figure 3a in Zhu et al. (2019), which presents a statistical magnetic PSD distribution in terms of L-value and frequency.

Occurrence and Characteristics of Exohiss and Plasmaspheric Hiss
Based on the aforementioned criteria, we surveyed both exohiss and plasmaspheric hiss utilizing Van Allen Probes A and B data from 1 October 2012 to 28 February 2018 (approximately 5.5 years), covering the entire MLT region three consecutive times.This observational period is longer than that used in previous studies by Zhu et al. (2019) and Wang et al. (2020).Figure 2 shows the L*-MLT distributions of (columns one to four) the number of samples, occurrence rate, median wave amplitudes B w , and WNA in each bin of the 0.2 L* × 1 h MLT, without considering its latitudinal distribution for the exohiss (upper row) and plasmaspheric hiss (lower row).The occurrence rate was defined as the ratio of the wave detection time to the spacecraft dwell time in each bin.The dwell time was estimated only for events in which plasmapause was defined.Exohiss appeared mainly on the dayside, with an average amplitude of ∼20 pT and WNA < ∼20° when limited to L* <3, where intense exohiss waves were mainly observed.This is qualitatively similar to the results of Zhu et al. (2019) and Wang et al. (2020).The exohiss amplitude decreased as L* increased, reaching a maximum near the plasmapause.This may reflect the fact that an exohiss wave at a larger L* is further away from the plasmaspheric hiss and therefore experiences stronger Landau damping (Zhu et al., 2019).Meanwhile, plasmaspheric hiss waves were evenly distributed throughout the entire MLT region, but intense field-aligned (<20°) waves appeared mostly on the dayside, which is qualitatively similar to the results of previous studies (e.g., Kim et al., 2020;Yu et al., 2017).Such MLT asymmetry of the wave amplitudes is related to the frequently occurring region of the exohiss, implying that dayside exohiss is closely related to the intense dayside plasmaspheric hiss.
In contrast, compared to the dayside exohiss, highly oblique (>50°) exohiss waves with relatively weak amplitudes (<10 pT) were observed in the outer L* (>3) of the postmidnight sector (0-6 MLT), with an occurrence rate of <10%.One of the main reasons for this MLT dependence is strong Landau damping, which results in a weak exohiss on the nightside (Bortnik et al., 2006(Bortnik et al., , 2011)).The possibility of capturing low-frequency (<0.1f ce ) chorus 10.1029/2023JA031777 5 of 13 waves in our nightside exohiss data set cannot be ignored; however, the low-frequency chorus may be absent at all magnetic latitudes for all geomagnetic activities, as measured by AE (Meredith et al., 2014).Nevertheless, previous studies have suggested that features of plasma waves in the magnetosphere may differ depending on solar wind conditions, which the geomagnetic index may not capture (e.g., Golden et al., 2012;Kim & Shprits, 2017;Kim et al., 2013Kim et al., , 2015Kim et al., , 2020)).

Dependence of Exohiss Occurrence on Plasmaspheric Hiss
To investigate the relationship between plasmaspheric hiss and exohiss quantitatively, we sorted the occurrence of exohiss in the L*-MLT plane according to time elapsed after plasmaspheric hiss measurement, from 0 to 9 hr at 1-hr intervals (Figure 3).The occurrence rate was estimated by normalizing the samples in each bin to the total number of samples for all bins from 0 to 9 hr.The corresponding number of samples in each bin is shown in the small insets of Figure 3 panels.In this analysis, we hypothesized that, if an exohiss originated from a hiss, it would require time to evolve into an exohiss.Thus, it would be observable outside the plasmapause some time after hiss measurement inside the plasmapause.This approach differs from the method adopted in the previous statistical studies of Zhu et al. (2019) and Wang et al. (2020).Zhu et al. (2019) sorted waves according to their location at the time of observation.This does not provide temporal variation, but only spatial differences in the distributions of the two wave types.Wang et al. (2020) similarly also analyzed differences in the spatial distribution of two waves by considering only events occurring near the plasmapause (<0.3 R E ), while ignoring events 10.1029/2023JA031777 6 of 13 occurring far from the plasmapause (>0.3 R E ).In contrast, our analysis incorporates both the spatial extent and temporal evolution of exohiss waves by sorting them in terms of elapsed time from hiss measurement.For our analysis, we used 1-min averaged data for both plasmaspheric hiss and exohiss waves.Exohiss occurrences on the dayside-where intense hiss waves were mainly observed (Figure 2)-gradually increased up to ∼4 hr after the measurement of hiss waves and decreased up to ∼9 hr.This may indicate the time required for the plasmaspheric hiss to escape into the plasmatrough.Nightside occurrences of exohiss also increased with time but were relatively fewer than those on the dayside.
However, this result should be viewed cautiously in view of some confounding factors.First, we only assessed exohiss occurrence up to 8-9 hr after plasmaspheric hiss measurements.When this analysis was extended to longer times, the trend shown in Figure 3 repeated every 8-9 hr (figure not shown), which may reflect the orbit period of the Van Allen Probes.Second, we included cases where data from only one of the two spacecraft was used as well as cases where data from both were used.In the case of only one spacecraft, time elapses while the probe crosses the plasmasphere after collecting measurement inside the plasmapause before it can observe the exohiss.However, when utilizing both spacecraft, located at different positions (i.e., inside and outside the plasmapause), it is possible to retrieve observations of both waves simultaneously; this can result in recording more exohiss samples within the time period immediately following hiss measurements.If we analyze only data that included both probes, the time of maximum exohiss occurrence following plasmaspheric hiss reduces to 2-3 hr, which is shorter than the 3-4 hr shown in Figure 3. Third, we cannot guarantee that our delayed time of exohiss observation does not imply its evolution time from plasmaspheric hiss because the exohiss already exists outside the plasmasphere and has been accidently observed by a probe.Because it is difficult to specify how long it may take for plasmaspheric hiss to evolve into exohiss, we included exohiss occurrences up to 3-4 hr post plasmaspheric hiss to consider all possibilities in the following statistical analyses (Figures 4-9).The Supporting Information S1 provides all statistics corresponding to Figures 3-9 separately for each of these two analyses.One analysis used data from only a single probe, whereas a second analysis considered the data from both probes.We confirmed that results from neither of the sole cases differed significantly from the combined statistical results for the two probes (Figures 3-9), except for the maximum time delay of exohiss occurrence after hiss measurements; this delay was 3-4 hr when considering only one probe and 2-3 hr when analyzing the collective data from both probes, yet the same interpretation can be applied to both scenarios.
To further examine the contribution of hiss amplitude in the evolution of hiss to exohiss (i.e., the correlation between hiss amplitudes and exohiss distribution), we sorted the exohiss waves according to their time of occurrence after hiss measurements and hiss amplitudes, as depicted in Figure 4.The first column indicates the global distribution of hiss occurrence for different amplitudes: B w < 10 pT (Figure 4a), 10 ≤ B w ≤ 30 pT (Figure 4b), and B w > 30 pT (Figure 4c).The second to fifth columns show the exohiss occurrence rate from 0-1 hr to 3-4 hr at 1-hr intervals.The corresponding sampling distributions are shown in the insets.The occurrences for each column were normalized by the total number of samples in that column.As evident from the first column, stronger hiss is more likely to occur on the dayside and at late MLT, which is qualitatively similar to the results of Kim et al. (2020).The peak exohiss occurrence region shifted accordingly across all time intervals, implying a relationship between these variables.

Dependence of Exohiss Occurrence on Solar Wind Parameters
The generation and evolution of plasmaspheric hiss are influenced by past solar wind conditions (e.g., Golden et al., 2012;Kim et al., 2015).Therefore, we analyzed the occurrence of exohiss according to solar wind conditions at the time of plasmaspheric hiss measurements to assess the solar wind conditions under which hiss waves could evolve into exohiss.Here, we present the global distribution of cumulative exohiss occurrences during the 0-4 hr after hiss measurement, along with the distribution of plasmaspheric hiss.
Figure 5 shows the global distributions of exohiss and hiss for different IMF B Z ranges at the time of plasmaspheric hiss measurement.The first three columns show the distribution of exohiss occurrence (normalized using the total number of samples in the first column), B w , and WNA, respectively, as based on the cumulative data of both probes obtained 0-4 hr after hiss measurements.The fourth to sixth columns display the distribution of hiss occurrence, B w , and WNA, respectively.The corresponding sampling distributions are shown in the insets of the first and fourth columns.The rows, from top to bottom, are in the order of southward-increasing IMF B Z , along with the northward IMF B Z condition in the top row.When the IMF B Z was northward, exohiss waves appeared with a relatively higher occurrence on the dayside than on the nightside (Figure 5a).The dayside exohiss waves exhibited a relatively high amplitude (>10 pT) and were more field aligned compared to those on the nightside.However, with increasing IMF B Z southward (Figures 5b-5d), the region of maximum exohiss occurrence shifted from the noon toward the prenoon sector.In other words, the low-occurrence regions shifted from the midnight toward the dusk sector.This resembles the shift in the distribution of intense hiss waves on the dayside (compare Figures 5r-5t).The plasmaspheric hiss in the dusk sector (∼18 MLT) weakened, whereas the noon sector (∼12 MLT) hiss strengthened, which is related to enhanced particle injection during substorms.In addition, with increasing IMF B Z southward, the occurrence of nightside exohiss slightly increased (compare Figures 5b-5d).
Figure 6 shows the global distributions of the exohiss and hiss for different ranges of solar wind speed V SW at the time of the plasmaspheric hiss measurement.The exohiss measurements are those recorded 0-4 hr after the hiss measurements.The column layout is the same as that of Figure 5.The rows from top to bottom are in the order of increasing V SW.Across all conditions, exohiss appeared with a relatively higher frequency on the dayside than on the nightside (first column).As V SW increased, the amplitudes of both exohiss and hiss slightly increased (second and fifth columns), while there was no significant change in the MLT of exohiss occurrence on the dayside (first column), where the most intense hiss was encountered.
We also examined the contribution of the solar wind dynamic pressure (P SW ) to the distributions of exohiss and hiss.The results are shown in Figure 7, with the column layout again reflecting that of Figure 5.The rows from top to bottom are in the order of increasing P SW .As P SW increased, the region of the frequent occurrence of exohiss was limited to the prenoon sector (Figure 7a-7c), whereas there was no significant change in the MLT with a strong hiss on the dayside (Figures 7m-7o).One notable factor is that the exohiss occurrence increased on the nightside (Figures 7a-7c).In addition, the amplitude of exohiss and hiss waves seemed to gradually weaken on the dayside (second and fifth columns).
The solar wind parameters are interconnected.Thus, to identify the pure effect of each parameter on exohiss occurrence, we divided the statistics presented in Figure 5 into two groups: one dominated by high-speed V SW (>500 km/s) under the northward IMF B Z (>2 nT), and the other dominated by the southward IMF B Z (<−4 nT) under low-speed V SW (<400 km/s).The cutoff values were adopted from Kim et al. (2015), since these authors clearly demonstrated the dependence of plasmaspheric hiss on solar wind parameters by using these values.The corresponding results are shown in the first and second rows of Figure 8, which reflect the pure effect of the IMF B Z or V SW .The column layout is the same as that of Figure 5.A higher V SW led to more occurrences of exohiss over the whole dayside (Figure 8a).In contrast, a strong southward IMF B Z resulted in a minimum occurrence of exohiss in the dusk sector (Figure 8b).Interestingly, this tendency followed a change in the distribution of intense hiss amplitude on the dayside (Figures 8i and 8j).Meanwhile, the amplitude of nightside exohiss waves was slightly enhanced and they displayed a higher occurrence.
We further identified the relative contribution of the P SW to exohiss distribution by dividing the exohiss measurements into two groups: strong and weak (with both groups under the same conditions of northward IMF B Z (>2 nT) and weak V SW (<400 km/s)).In the case of weak P SW , an exohiss occurred on the entire dayside (Figure 9a), which is quite similar to the MLT distribution of the wave amplitude of plasmaspheric hiss in Figure 9i.In contrast, intense P SW limited the occurrence of exohiss to the prenoon sector (Figure 9b), which is not consistent with the distribution of hiss amplitude (Figure 9j).Unlike other solar wind parameters, when the P SW was strong, it appeared to prevent the leakage of postnoon hiss into the plasmatrough region.The prenoon exohiss was likely to constitute a wave caused by locally excited electrons (Gao et al., 2018;Zhu et al., 2018) rather than originating from the leakage of plasmaspheric hiss.
Finally, to examine the possibility of chorus waves as a source of nightside exohiss, in Figure 10, we evaluate the dependence of nightside chorus wave frequency spectra on different solar wind parameters, that is, V SW , northward IMF B Z , southward IMF B Z , and P SW .We first interpolated the chorus wave PSD onto a frequency normalized by the electron cyclotron frequency f ce .The f lhr (lower hybrid resonance frequency) up to 0.8f ce was divided into 80 bins for an estimation of the average PSD values for each frequency.Compared to the results for V SW (Figure 10a) and northward IMF B Z (Figure 10b), the nightside chorus wave PSD was significantly enhanced at ∼0.1f ce with the southward IMF B Z (Figure 10c) and P SW (Figure 10d).This may imply that, during the enhanced solar wind conditions of the southward IMF B Z and higher P SW , the peak frequency of lower-band chorus waves extends to below 0.1f ce, corresponding to the frequency range of the exohiss.Therefore, the nightside exohiss may be related to the evolution of low-frequency chorus waves.However, the wave amplitude of the nightside exohiss was generally lower than that of the dayside exohiss, suggesting an insignificant contribution of electron loss in the radiation belt.

Discussion and Conclusions
We investigated the relationship of exohiss occurrence and properties with plasmaspheric hiss occurrence and solar wind parameters, based on Van Allen Probes observations covering the entire MLT region three consecutive times over almost five and half years.Our findings can be summarized as follows: 1. Exohiss waves occurred mainly on the dayside, which showed a relatively higher occurrence of exohiss than the nightside.The region of dayside exohiss occurrence coincided with regions where intense plasmaspheric hiss waves frequently appeared.2. The occurrence of exohiss waves gradually increased up to ∼4 hr after plasmaspheric hiss measurements and showed a clear MLT dependence.Postnoon exohiss waves usually appeared in regions where a large-amplitude hiss is commonly observed, while prenoon waves appeared in regions where a relatively low-amplitude hiss is commonly observed.3. Dayside exohiss occurrence depended on solar wind conditions.A stronger southward IMF B Z limited the region of exohiss occurrence to the prenoon sector, whereas exohiss was more widely distributed over the dayside at high solar wind speeds.These findings were consistent with the distribution of the plasmaspheric hiss.10.1029/2023JA031777 11 of 13 4.The distributions of plasmaspheric hiss and exohiss waves did not match in the case of a strong P SW , in contrast to the scenario for other solar wind conditions.This may indicate locally excited exohiss waves rather than leakage from plasmaspheric hiss. 5.A low-frequency (<0.1f ce ) nightside chorus was likely to be selected as nightside exohiss.During enhanced solar wind conditions with southward IMF B Z or higher P SW , the lower-band chorus intensity was enhanced to below 0.1f ce, which corresponds with the frequency range of the exohiss.
The finding summarized in point 3 above indicates that an enhanced southward IMF B Z and V SW are favorable conditions for the evolution of plasmaspheric hiss into exohiss.Unlike this, point 4 points to different origins of exohiss during enhanced P SW that are not related to leakage from plasmaspheric hiss.This may be due to the excitation of exohiss via locally injected electrons (Gao et al., 2018;Zhu et al., 2018).The stronger the P SW , the more significant the day-night asymmetry in the field structure, resulting in favorable conditions for drift shell splitting (Roederer, 1970) or bifurcation (Kim et al., 2008;Öztürk & Wolf, 2007;Shabansky, 1971), which naturally favors the anisotropic enhancement of particle distribution and, consequently, a growth of plasma waves.However, we cannot ignore a shift in the frequency band to above 0.1f ce during enhanced P SW , as shown in the case study of Yu et al. (2018), which may result in relatively less exohiss occurrence at prenoon than at postnoon.Such variation may not be captured in our data set, which included only events occurring below 0.1f ce .A more comprehensive approach in future research should include an analysis of ambient particle distribution and the direction of exohiss wave propagation-data that can be collected from spacecraft that cover high altitude regions-to clearly distinguish exohiss originating from plasmaspheric hiss.In addition, our results indicate that exohiss modeling that employ solar wind parameters as model input (instead of using only the geomagnetic Kp index) can more accurately describe the MLT distribution.Although Zhu et al. (2019) reported the Kp dependence of exohiss occurrence, the authors did not emphasize the MLT-dependent nature of this response to Kp.
Since exohiss has the potential to interact with the relativistic electrons of the outer radiation belt (Li et al., 2019;Zhu et al., 2015), solar wind dependent-modeling could provide more accurate estimations of the MLT dependence of electron distribution in radiation belts.

Figure 1
Figure 1 shows an example of the exohiss wave signals observed on 17-18 March 2016 by Van Allen Probe A. Figures 1a-1g respectively show the plasma number density, f pe /f ce ratio, magnetic field PSD (B PSD ), signed ellipticity, planarity, WNA, and wave flag (blue for the plasmaspheric hiss, red for the exohiss, and yellow for the chorus).The vertical dashed line in each panel indicates the plasmapause location.The two (one dashed and one correspond

Figure 1 .
Figure 1.An example of an exohiss event, as observed by Van Allen Probe A on 17-18 March 2016.Panels depict (a) electron number density, N e (cm −3 ), (b) electron plasma to cyclotron frequency ratio, f pe /f ce , (c) magnetic power spectral density, (d) signed ellipticity, (e) planarity, (f) wave normal angle (WNA), (g) a flag indicating wave type (blue: plasmaspheric hiss, red: exohiss, and yellow: chorus waves), (h) root-mean-square wave amplitudes of exohiss and plasmaspheric hiss, and (i) wave power-averaged WNA of exohiss and plasmaspheric hiss.In each panel, the vertical dashed line indicates the plasmapause location, and the curves in panels 1c-1g represent 0.8f ce (dashed) and 0.1f ce (dotted).

Figure 2 .
Figure 2. L*-MLT distribution of (upper panels) exohiss and (lower panels) plasmaspheric hiss.Columns one to four correspond with the number of samples, occurrence rates, root-mean-square amplitude (B w ), and wave normal angle for each wave from 1 October 2012 to 28 February 2018.

Figure 3 .
Figure 3. L*-MLT distribution of exohiss occurrences (with their sampling distributions being presented in insets), as sorted by time elapsed from plasmaspheric hiss for every 1-hr interval.

Figure 4 .
Figure 4. L*-MLT distribution of exohiss occurrence sorted by time elapsed from plasmaspheric hiss measurement for every 1-hr time interval (second to fifth columns) for different amplitudes of plasmaspheric hiss (top to bottom rows).In each panel, the corresponding sampling distribution is displayed in an inset.

Figure 5 .
Figure5.The exohiss (first to third columns) and plasmaspheric hiss (fourth to sixth columns) distributions for different ranges of interplanetary magnetic field B Z in an increasing southward order (top to bottom rows).The first to third (and fourth to sixth) columns correspond to wave occurrence rates (with sampling distributions being presented in the insets), wave amplitude, and wave normal angle.

Figure 6 .
Figure 6.The exohiss and plasmaspheric distributions for different ranges of solar wind speed (V SW ), presented in the same layout as Figure 5.

Figure 7 .
Figure 7.The exohiss and plasmaspheric distributions for different ranges of solar wind dynamic pressure (P SW ), presented in the same layout as Figure 5.

Figure 9 .
Figure9.The exohiss and plasmaspheric hiss distributions associated with (upper row) weak and (lower row) strong solar wind dynamic pressure (P SW ) under conditions of northward interplanetary magnetic field B Z (>2 nT) and low solar wind speed (<400 km/s), presented using the same layout as Figure8.

Figure 8 .
Figure 8.The exohiss and plasmaspheric hiss distributions (upper row) associated with northward interplanetary magnetic field (IMF) B Z and high solar wind speed and those (lower row) characterized by southward IMF B Z and low solar wind speed.A similar layout to Figure 5 is used.

Figure 10 .
Figure 10.The relationship between wave frequency spectra of nightside (0-6 magnetic local time) chorus waves and (a) solar wind speed V SW , (b) northward interplanetary magnetic field (IMF) B Z , (c) southward IMF B Z , (d) dynamic pressure P SW .