Inverted radial Alfvén continua: first results from “heliophysics audified: resonances in plasmas”

Ultra Low Frequency (ULF) waves with periods of $\sim$10–1,000 s can lead to space weather impacts such as induced electrical currents in power grids, thus it is important to understand the factors controlling wave dynamics. This is challenging, however, as waves (1) are affected by multiple factors simultaneously, (2) are non-stationary which in some cases precludes use of identification methods that assume stationarity, (3) can occur in superposition with each other making them difficult to separate and identify. Past studies have addressed these challenges through combined audiovisual analysis tools to identify complex but recurring patterns in ULF wave activity that eluded standard visual inspection and automated detection algorithms, as well as through crowd-sourced wave identification. The “Heliophysics Audified: Resonances in Plasmas” NASA citizen science project follows these studies by deploying a Graphical User Interface (GUI) for crowd-sourced ULF wave identification to a large online audience before and during the Heliophysics Big Year (HBY). In this study, we discuss the initial development, beta testing, and deployment of the GUI in April 2023. We further discuss the key initial scientific findings of the HARP project, in particular the discovery by volunteers of anomalous standing Alfvén wave activity with frequency increasing with distance from the Earth. Finally, we discuss participant impacts and lessons learned, as well broader impacts beyond the scope of the original project such as collaborations with museums and musicians. We place these results in context with previous work and discuss implications for future studies.

1 Introduction

Many different types of plasma waves can occur in the near-Earth space environment region dominated by the Earth’s magnetic field, or magnetosphere. The lowest frequency waves, referred to as Ultra Low Frequency (ULF) waves (Jacobs et al., 1964), often behave like standing waves bounded by different regions of the magnetosphere and ionosphere (ionized portion of the Earth’s upper atmosphere). For example, standing Alfvén waves with frequencies $\sim <$100 mHz form along magnetic field lines inside the magnetosphere and, like waves on string, are bounded at either end by the ionosphere (Sugiura and Wilson, 1964) with frequencies depending on the distance between the two end points of the string (north and south ionosphere) and the properties of the string (Alfvén speed which depends on magnetic field strength and plasma density along the field line). Many analogies can be made between musical instruments and ULF waves: (1) like a harp, the frequency of standing Alfvén waves typically decreases with distance from the Earth as magnetic field lines become longer (e.g., Sugiura and Wilson, 1964), (2) the outer boundary of the magnetosphere behaves like a stretched membrane with properties like a drum (Archer et al., 2019), (3) in the interior of the magnetosphere, some standing waves bear similarities to waves in an organ pipe, with the size of the pipe (size of the magnetosphere or distance between boundaries in the magnetosphere) dictating the frequency (e.g., Hartinger et al., 2023a).

These ULF waves can lead to space weather impacts such as geomagnetically induced currents in power grids (e.g., Hartinger et al., 2023b), upper atmosphere heating that increases satellite drag (e.g., Shi et al., 2025), and radiation belt dynamics that can potentially lead to satellite damage (Lanzerotti, 2017). Thus, understanding the factors controlling the properties of these waves (amplitude, frequency, etc.) is a high priority (Wright et al., 2024). Toroidal modes, a particular type of standing Alfvén wave, have been linked to several of these space weather impacts. Theory, modeling, and observation have shown that toroidal modes have azimuthal (east-west) magnetic field disturbances and radial (away from Earth near the equatorial plane) electric field disturbances in the magnetosphere, with amplitudes, frequencies, and other properties generally depending on the interplay between the driving conditions in the solar wind, the Alfvén speed throughout the magnetosphere (both radial and along a magnetic field line), and the ionospheric conductance (e.g., Wright et al., 2024).

Most observational ULF wave studies have relied on identification and classification schemes based on finite time windows. For example, wave activity can be identified using Discrete Fourier Transforms (DFT) or Discrete Wavelet Transforms (DWT) to obtain band-integrated wave power for a certain frequency and time range, or by detecting the occurrence of a discrete frequency peak in a power spectrum over a finite time interval. However, ULF waves are inherently non-stationary as they grow/dissipate, and their properties depend on many driving conditions and magnetosphere/ionosphere conditions that evolve dynamically. Moreover, a finite time window typically lasting a few wave cycles will contain no information about the wave activity occurring in the preceding or following hours or days, thus recurring patterns of wave dynamics during geomagnetic storms (Archer et al., 2018), high-speed streams (Engebretson et al., 1998), and other conditions relevant to space weather impacts will be missed. In some cases, this challenge can be addressed via superposed epoch analysis, but this is only possible if a fiducial time, such as an interplanetary shock arrival time, is known a priori; there may well be repeatable wave conditions without previously known fiducial times as was shown by, for example, Archer et al. (2018). Moreover, multiple ULF wave modes can occur at the same time with the same frequency, making it challenging to isolate different wave modes. Past studies have addressed some of these challenges through visual inspection of dynamic power spectrograms covering, for example, an entire satellite orbit and using qualitative classification of the wave activity seen by one or two researchers participating in the study. However, there is no way to know a priori whether a single orbit is sufficient to capture complex but recurring patterns in wave dynamics, and visual inspection itself has limitations when using standard survey plots (e.g., capturing multiple wave modes that overlap in frequency, setting appropriate time, frequency, and wave power scaling without advance knowledge of the properties of the waves of interest). For these reasons, current ULF wave event detection algorithms are not well suited to finding complex but recurring patterns in wave dynamics.

Audio analysis provides one path forward to address these challenges. Sonification is the use of non-speech audio to convey information (Kramer et al., 2010), and it is often used in the space sciences (United Nations Office of Outer Space Affairs, 2023). For example, the amateur (ham) radio community has long developed listening approaches around the identification of signals and interference related in part to space weather. The most well-known example of sonification in amateur radio is the reception by ear of Morse code (referred to in the community as “CW,” for “continuous wave”). These listening skills were leveraged for scientific purpose by the Golden Ears project (Perry et al., 2020), which recruited experienced CW operators to demodulate signals which could not be distinguished by a computer. Audification, a one-to-one mapping of data samples to audio samples (Dombois and Eckel, 2011), has been demonstrated to be useful in the analysis of space physics time series data (Alexander et al., 2014). Several well-known space physics phenomena owe their descriptive names to audification, including whistlers (Helliwell, 1965), chorus (Sazhin and Hayakawa, 1992), hiss (LaBelle and Treumann, 2002), and “lion roar” emissions (Smith and Tsurutani, 1976). More recent work driven by auditory observations has identified carbon charge states as an ideal indicator for determining solar wind source regions (Landi et al., 2011), novel long-duration proton-cyclotron wave storm activity in the interplanetary solar wind (Wicks et al., 2016), and the unexpected prevalence of poloidal field line resonances during geomagnetic storm recovery (Archer et al., 2018). This technique has now been integrated into NASA’s CDAWeb data repository¹ as a standard data product, and novel uses continue to push boundaries including auralization methods for multi-spacecraft observations (Collins et al., 2024).

Citizen science, or participatory science, is the collection and/or analysis of scientific research data by members of the general public in collaboration with professional scientists. Many of the most popular analysis-based citizen science projects involve interacting with captivating imagery, such as Galaxy Zoo, Penguin Watch, Chimp&See, and Cloudspotting on Mars (https://zooniverse.org). This reliance on imagery in citizen scientist data processing projects is also true within heliophysics, with projects principally concentrated on solar (e.g., Solar Stormwatch, Solar Jet Hunter, and Solar Active Region Spotter) and auroral (e.g., Aurorasaurus, Aurora Zoo, and AuroralZone) remote sensing (see Grandin et al., 2025, and references therein). In contrast, a large portion of heliophysics research relies on in situ time-series analysis, which unfortunately does not produce the visually arresting scientific data that can motivate citizen scientist involvement (Archer, 2021a). However, citizen science analysis has also been used successfully to identify many types of naturally occurring phenomena with sound, from bird calls (Lehikoinen et al., 2023) to urban soundscapes (Cartwright et al., 2019). Thus, sonification may provide a means of engaging citizen scientists with in situ heliophysics research.

The combination of sonification with citizen science was pioneered by Archer et al. (2018) through the MUSICS (Magnetospheric Undulations Sonified Incorporating Citizen Scientists) project. MUSICS audified magnetospheric ULF waves observed in geostationary orbit for novel exploratory analysis by high-school students as part of 6-month collaborative participatory research projects (Archer et al., 2021). Students explored and analyzed the audio files using free Audacity software (https://www.audacityteam.org/) then logged events in a provided spreadsheet containing formulae for converting time, frequency, and location. This led to the identification of numerous long-lasting poloidal mode wave events during the recovery phase of geomagnetic storms, a particular class of standing Alfvén wave that was previously thought to be rare (Archer et al., 2018). The approach also had highly positive outcomes for participating high school students and teachers on their experiences (Archer et al., 2021), accessibility/equity (Archer, 2021b), and medium-to long-term impacts (Archer and DeWitt, 2021).

Building on the MUSICS project, a follow-on project, “Heliophysics Audified: Resonances in Plasmas,” (HARP) sought to adapt the MUSICS methodology to a fully-online audience and a new dataset, enlisting the help of online volunteers to identify repeatable patterns in standing Alfvén wave activity (continuum of toroidal modes) to better understand how these waves are generated. In this paper, we discuss all phases of the HARP project, including initial design and implementation (Section 2), first science results (Section 3), and participant impacts and broader outreach (Section 4). We further discuss how information from the HARP project is being used and could be used in the development of future projects.

2 HARP project design and implementation

Many of the HARP project design elements were similar to MUSICS, though adapted to a web-based platform with tutorials and additional supporting context to train and guide volunteers while also allowing open-ended observations. The data processing tools were also similar to MUSICS, though adapted to a different satellite constellation with an elliptical orbit: the Time History of Events and Macroscale Interactions during Substorms (THEMIS) mission (Angelopoulos, 2008). In this section, we detail the HARP project design and implementation, including data processing methods, interface development and testing, institutional review board compliance and data storage, and methods for validating volunteer event identifications.

2.1 Data processing and sonification

The development of the data processing and sonification methods used for HARP from NASA THEMIS spacecraft measurements was detailed in Archer et al. (2022a) and can be accessed at https://github.com/Shirling-VT/HARP_sonification. Here we briefly summarize the final methods used.

Before sonification, the magnetospheric magnetic field variations must be extracted from the data and transformed into an appropriate coordinate system. THEMIS magnetic field and plasma data as well as satellite position are loaded from the NASA CDAWeb repository. Magnetic field data are first despiked to remove noise. All data sets (magnetometer, plasma, satellite position) are then interpolated to a uniform 3-s resolution. The magnetosheath intervals are identified and removed using the plasma and magnetic field measurements (these intervals are outside the region of interest). A magnetosheath interval is identified when the satellite is $>$8 Earth radii from the Earth and one or more of the following conditions are met: (1) electron density $>10\#/cc$, (2) perpendicular electron number flux $>2\ast 10^7\#/cc/s$ or (3) velocity is $<-200km/s$ in the GSM x direction (Hartinger et al., 2023a); we further exclude rare intervals when the plasma moments required for these tests are flagged as missing or bad data quality. Magnetosheath data gaps are filled using interpolation to produce a continuous magnetic field series. A background magnetic field is computed using a 30-min window running average. Magnetic field data are then subtracted from the background magnetic field to obtain the detrended magnetic field fluctuations. The detrended magnetic field data are rotated into field-aligned coordinates where z is along the background magnetic field direction, y is the azimuthal direction in Solar Magnetospheric (SM) coordinates, and x completes the right-hand orthogonal set. Perigee intervals are removed if the satellite geocentric distance is less than 5 Earth radii. We focus on the azimuthal magnetic field component which represents the toroidal mode ULF wave. This component is used to create a waveform plot and compute its time derivative, from which spectrograms are generated using a 1024 point ($\sim$51 min) DFT window. Both time series plots and spectrograms are used in the GUI to assist with the identification of HARP structures. Finally, since (1) toroidal modes occur most often at dawn and dusk (Takahashi et al., 2015) and (2) analysis near noon and midnight was complicated by several factors (e.g., magnetopause usually inside satellite apogee near noon), for simplicity we focused on observations collected when the satellite apogee is in the dawn (magnetic local time between 0400 and 0800) and dusk (magnetic local time between 1600 and 2000) sectors. We used data from the THEMIS-E spacecraft from 1 February 2008 through 24 December 2022, roughly one solar cycle.

THEMIS’s eccentric orbit means that direct audification, for example, as used by Alexander et al. (2014) in the solar wind and Archer et al. (2018) at geostationary orbit, is unsuitable since the rate at which ULF waves’ properties change in the resulting audio is too fast. Time scale modification, the speeding up or slowing down audio without affecting the frequency content, is therefore applied to the extracted ULF waves to stretch the time-series. The choice of time-scale modification method, stretch factor, and spectrum normalization was decided based on a public dialogue targeted at various stakeholder communities outside of the space sciences (Archer et al., 2022a). Informed by the survey results, HARP uses a wavelet phase vocoder with time stretch factor of 6, followed by the approximately $1/f$ background noise spectrum being normalized to an equal-loudness contour. The resulting time-series is then audified by normalizing to the maximum value across the interval and writing to Waveform Audio File Format (WAV) with sample rate 44.1 kHz using the soundfile Python package.

2.2 Institutional review board review compliance and data storage

Human subjects research for HARP was reviewed by the UCLA Institutional Review Board (IRB); all web/email/print materials, information sheets, and informed consent/assent forms were approved in Fall 2021. The HARP study was conducted in two separate phases. In Fall 2021, Phase 1 recruited eleven high school students from a teen science cohort already affiliated with SSI via local libraries in Boulder, Colorado (for more details on participant impacts from this beta test, see Section 4). All parents signed a written informed consent and students under 18 signed assent to enroll in the study. About a dozen additional UCLA adult volunteers were similarly recruited several months later. Participants attended a tutorial and in-person focus groups to assess their background knowledge of space weather, beta test the HARP GUI, and provide valuable feedback to optimize instructions and procedures for later use with the general public.

Phase 2 of the study was implemented during the Heliophysics Big Year in April 2023 via the HARP website (https://listen.spacescience.org), recruiting citizen scientists globally via NASA communications, listservs, social media, YouTube, etc. Users registered for an account with an email and password to allow intermittent analysis sessions without losing progress, and password recovery if the login was forgotten. Participation was limited to adults only, requiring that they read the study information sheet and willingly agree via checkbox to a waiver of informed consent, integrated into the web interface. Participants could then voluntarily check a box to either agree or decline to be contacted by email regarding study results, publications, or future HARP-related research opportunities. All collected HARP data were immediately anonymized and encoded with a unique UserID number, stored on an encrypted, password-protected server at SSI. Access to this personally identifiable information and code link database is confidential, restricted only to the SSI IT administrators (HARP team members Evaldas Vidugiris and James Harold) and will never be distributed or shared. All researchers and staff can only view anonymized, coded data for the purposes of analysis and tracking reliability of each user’s responses.

Volunteers were notified via the website and login pages that the prime data collection phase would conclude on 22 January 2024, meaning that they would need to make any marks by that date to ensure they were included in the initial analysis. However, the website and GUI remained active after that date, and a small number of marks continue to be collected. In this study, we use volunteer marks made through September 2024, though the majority of these marks were made between 23 April 2023 and 22 January 2024 (i.e., the prime data collection phase).

2.3 Graphical User Interface

The HARP Graphical User Interface was built using Unity (https://unity.com/), a game engine used for applications such as video games and educational software. It can run in a web browser, and it has been used frequently by Space Science Institute and the National Center for Interactive Learning for online educational games (https://www.scigames.org/). It was chosen for the HARP project because of ease of use, familiarity, and built-in functionality needed for HARP.

The HARP GUI incorporates several elements based in part on MUSICS, best practices in other citizen science projects (Grandin et al., 2025), experience from past online educational games, and feedback from beta testers. After logging in, the volunteers encounter a welcome screen where they can see how many wave events they have marked, how many of the 3-orbit intervals they’ve completed, and how many more they need to complete to earn a new medal. They then have the option to enter a “Practice” (tutorial) mode or “Continue” identifying events.

The tutorial mode provides a walkthrough of the HARP interface controls and functionality, brief explanations of the science behind the waves the volunteers identify, and example events for volunteers to select. Figure 1 shows one screen from the tutorial that appears after the user has been instructed to draw a box around a wave event; this step shows the volunteer how to fill out information in a popup window and provide more details about the wave event, including their confidence in event detection. The wave activity shown in the background is synthetic and represents an ideal situation where standing Alfvén waves are present for the entire satellite orbit with a single harmonic (first third of plot), with two harmonics (second third of plot), and where no waves are present (last third of plot). After stepping through this ideal example, volunteers are then presented with two events with real satellite measurements and provided further guidance. In all cases, as the volunteers are stepping through the events they are guided on how to use both visual and audio cues (audio is played during the tutorial, see Hartinger (2025) for the corresponding audio clip for Figure 1) to identify wave events and record supporting information such as pitch direction.

Figure 1

Figure 1. The HARP Graphical User Interface for tutorial mode. The measurements used are from synthetic data meant to represent idealized conditions. See Hartinger (2025) for the corresponding audio file.

Figure 2 shows the HARP GUI during standard data collection mode [the corresponding audio clip can be found at Hartinger (2025)]. On the left part of the screen, users see the orbital position of the THEMIS satellite move as the audio advances; users can also change the perspective of the viewer relative to the satellite (e.g., view from above the north pole, above the south pole, etc.), with the button on the bottom left resetting to the default view. On the top right, the magnetic field time series (east-west component) is shown, while the corresponding spectrogram is shown on the bottom right. Several buttons on the far right provide functionality to control volume, contrast, zoom, audio playback, and more. At the most basic level, the GUI facilitates use of both visual and audio cues to identify times and frequencies when wave activity is occurring, by drawing a box around the wave activity feature on the spectra, and then recording information and text comments about the event (see Figure 1 for popup fields). Once volunteers have finished marking wave events, they are instructed to select the button “Mark event as completed” at the bottom; when they do this, a final popup window appears that allows the user to record more information about the entire 3-orbit satellite interval (e.g., were the waves particularly clear, was the interval particularly noisy). All of this information can ultimately be used by researchers to address the project science questions, as well as additional science questions outside the scope of the original project.

Figure 2

Figure 2. The HARP graphical user interface for standard data collection mode. Actual satellite magnetic field measurements are used. Taken from https://listen.spacescience.org. See Hartinger (2025) for the corresponding audio file.

The HARP GUI was designed to provide a user-friendly and encouraging experience for volunteers who would likely be unfamiliar with toroidal modes and other aspects of space plasma physics. Wherever possible, technical terms were removed from the tutorial and explanation was provided for what volunteers were seeing and hearing; these explanations and several other GUI modifications were developed in partnership with beta testers (see Section 2.4). Additionally, incentives and encouragement were provided through (1) personalized status updates that show the number of marks and events completed as well as the ability to earn medals when a certain number of events were completed, (2) explanations for how their marks would be used by scientists to better understand space weather, (3) encouragement to make marks when they observed unexpected trends (i.e., waves that were not exactly like the ideal cases in the tutorial). Finally, the length of the audio clips and overall 3-orbit analysis interval was chosen based on feedback from beta testers on the optimal clip length to maintain focus and interest.

Supporting information was also provided on the HARP Website and linked throughout the GUI. For example, a promotional video was created to explain the science behind HARP, solicit help from volunteers, and also show a brief view of the GUI. A short walkthrough of the GUI functionality to mark wave events was also shown in a separate video. Further background material on the science, space weather as a whole, and sonification was also provided, as well as a FAQ generated in response to volunteer questions.

2.4 Beta testing

Many changes and additions were made to the GUI based on feedback provided during the beta tests (Phase 1 of project, see Section 2.2) as well as from online volunteers after the project launch, ranging from additional instructions in the tutorial to modifications to the mark popup windows. For example, beta test volunteers were initially only guided through an ideal event based on modeled wave properties (Figure 1). Many volunteers commented that this was not enough guidance, and that they wanted additional guidance in marking wave activity from real satellite measurements, and to be shown how scientists would mark the wave activity. As a result, we expanded the tutorial to include real satellite measurements and comparisons with scientist marks. Conversely, some volunteers did not want to click through the entire tutorial and instead preferred to begin making marks immediately; as a result, we included an option to skip the tutorial.

These beta testers also made one of the most significant findings in the HARP project, identifying an anomalous event that we analyze in more detail in Section 3. This finding led us to choose one of our tutorial time intervals to coincide with the event that they identified so that we could highlight the need to mark a range of possible events, not only the events that look and sound like the idealized events (Figure 1).

2.5 Methods of validating volunteer results

We validated the results obtained from the HARP volunteers using two methods. First, we compared the volunteer marks with expert marks in a few events. By design, the first four events (each event is three orbits of THEMIS, similar to Figure 2) that volunteers encounter are not random, presented in the same sequence for all volunteers. They include two partly guided tutorial events and two events in standard data collection mode (i.e., no guidance). These four events were divided into sub-intervals corresponding to half-orbits of the THEMIS-E spacecraft. Five scientists on the HARP team examined these time intervals for the presence or absence of standing Alfvén waves, and this information was used to construct two ratings for volunteers, a “signal” rating and a “noise” rating. The signal rating used eight half-orbits with a clear consensus among the scientists on the presence of standing Alfvén waves; volunteers received a rating of eight if they marked an event in all half-orbits, 0 if they marked none. The noise rating used 4 half-orbits with a clear consensus from scientists on the absence of standing Alfvén waves (noise/no clear events); volunteers received a rating of 4 if they did not mark any events during any of these half-orbits. In practice, the noise rating was only used to identify volunteers who marked nearly every half-orbit, regardless of the presence of clear wave activity. Although marks from all volunteers were considered in the study, we used a threshold of 4 for the signal rating type and one for the noise rating type to reduce the dataset for some analysis.

The second validation method was based on replication of statistical results from past studies. Though the HARP project’s goals were to identify new types of ULF wave dynamics, the dataset can also be used to replicate previously known wave activity trends as a validation step. Anderson et al. (1990) used magnetic field measurements from the AMPTE/CCE spacecraft, which had an orbit and sampling rate similar to THEMIS, to examine ULF wave activity primarily using dynamic power spectra of single orbits. Though their focus was only on frequencies from 0 to 80 mHz and radial distances (dipole L) from 5 to 9 Earth radii, their region of interest overlaps sufficiently with THEMIS to compare broad trends in measurements. Moreover, much like the HARP project’s use of visual inspection, Anderson et al. (1990) use visual inspection of dynamic spectrograms to categorize wave activity in half-hour segments, with their “harmonic mode” [Figure 6 in Anderson et al. (1990)] and “fundamental mode” [Figure 7 in Anderson et al. (1990)] event types similar to the ideal events that HARP users are trained to identify. The clearest result from their Figures 6, 7 is a significant preference for more dawn sector events compared to dusk, with roughly double the events observed at dawn in comparison to dusk (exact results depend on the radial distance, event type, and specific local time). Similar results have been found in other studies for the fundamental toroidal mode using different methods. For example, Takahashi et al. (2015) used an automated algorithm to identify monochromatic wave activity in THEMIS satellite electric field (velocity moment) measurements, again finding roughly double the events at dawn compared to dusk (Figures 12a, 13a in that study). These past results are consistent with HARP volunteer findings, who identified a clear preference for dawn events. Figure 3 shows one example of this dawn-dusk asymmetry. Considering only the 4 bars on the left, there are significantly more events identified at dawn (orange bars) than dusk (purple bars); the other categories shown on this plot will be discussed further in later sections. Note that in this Figure, only marks from rated volunteers are included (using thresholds mentioned above, four for signal rating and one for noise rating), and marks from the tutorial events are excluded.

Figure 3

Figure 3. A subset of the number of volunteer marks that followed expected trends for standing Alfvén waves (left 4 bars, pitch decreases with increasing distance from the Earth) and unexpected trends (right 4 bars, pitch increases with increasing distance from the Earth). “Up” and “Down” refer to whether the frequency/pitch increases or decreases, respectively, while “In” and “Out” refer to whether the spacecraft is inbound (moving towards the Earth) or outbound (moving away from the Earth). This subset does not include marks that, for example, correspond to waves/sounds with constant pitch. Only marks from rated volunteers are included in this Figure, and marks from tutorial (guided) events are excluded.

Finally, although not used formally as a validation step, we also checked heat maps for volunteer mark occurrence to assess whether volunteers were reaching a consensus in identifying events. For further details, see Section 3.1.

3 Science results from HARP: standing Alfvén waves in inverted wave speed profile

In this section, we discuss a key early finding from HARP related to unexpected trends in standing Alfvén wave properties. We first briefly introduce further background on the expected trends beyond what was discussed in Section 1.

The frequency of standing Alfvén waves depends on several factors. In realistic representations of the Earth’s distorted dipole magnetic field, the frequency would generally be expected to decrease with increasing radial distance owing to the longer magnetic field lines at larger distances. The spatially varying Alfvén speed also plays an important role. Most past ULF wave studies have assumed that the Alfvén speed decreases with distance from the Earth, with the exception of at the plasmapause in the inner magnetosphere and within discrete density structures such as plasmaspheric plumes (e.g., Elsden and Wright, 2022). This is because of expected trends in the Earth’s magnetic field and plasma mass density, which both generally decrease with increasing distance from the Earth, with the dipole magnetic field $(B)$ at the equator decreasing $\propto L^{-3}$ and the plasmaspheric mass density at the equator $(\rho )$ nominally thought to decrease $\propto L^{-4}$ or $\propto L^{-3}$, leading the Alfvén speed $(B/\sqrt{{\mu }_0\rho })$ to decrease $\propto L^{-1}$ or $\propto L^{-1.5}$ (Moore et al., 1987). However, the assumptions of a dipole magnetic field and plasmaspheric mass density trends become less reliable with increasing distance from the Earth, with distorted dipole fields and contributions from other plasma sources (e.g., Low Latitude Boundary Layer) becoming more important. Observations have shown that the $\propto L^{-1}$ trend often does not hold, with Alfvén speed increasing or remaining fixed as distance from the Earth increases (Archer et al., 2015; 2017). Moreover, much of the early theory used to describe ULF wave properties was based on idealized geometries such as boxes, cylinders, and rectangular waveguides, whereas recent theory and global simulations have revealed the importance of 3D effects in realistic geometries that better represent the Earth’s magnetosphere (Archer et al., 2022b; Wright et al., 2024).

Taken together, these facts suggest that the assumptions used for the radial Alfvén speed profile and continuum of Alfvén frequencies in the outer magnetosphere may not always hold, and an inverted Alfvén speed profile (speed increases as radial distance increases) may be present. Despite this, to our knowledge there have been no studies examining observations of ULF waves in inverted Alfvén speed profiles in the outer magnetosphere (though attention has been given to density structures in the inner magnetosphere such as plasmaspheric plumes). The volunteers in the HARP project resolved this gap almost immediately, when during the GUI beta testing they identified a wave event that had rising pitch with increasing radial distance continuing out to the THEMIS satellite apogee ($\sim$13 Earth radii). This section focuses on this case study and related statistical trends.

3.1 Case study

Figure 4 shows the east-west magnetic field time series and corresponding dynamic power spectrogram for 3-orbits of the THEMIS-E spacecraft from 2012-03-17 to 2012-03-20 using the same signal processing parameters and plotting format as the volunteers examined in the GUI. When examining this particular time range, the volunteers immediately identified the familiar U-shape on the spectrogram that they encountered in the tutorial (e.g., Figure 1) and identified several wave events consistent with the expected trend of increasing frequency with decreasing radial distance, which we hereafter refer to as “standard HARP events” (the analogy with a harp musical instrument holds here, since the pitch of harp strings increases as the string length decreases - see Section 1). However, a few volunteers also identified one event during the outbound pass of the THEMIS-E spacecraft on the third orbit shown in Figure 19 March 2012 02:46 to 14:44 UT) that did not follow the expected trend; in particular, the frequency increased as the radial distance increased. This was clear when listening to the audio from this orbit compared to other orbits, and can also be seen to some extent on the spectrogram. We hereafter refer to this as a “reverse harp event” since it followed the reverse trend for frequency that we expected for the toroidal mode. Note there are more wave event categories than standard and reverse harps which we’ll consider later in this section. The audio clip for the single orbit corresponding to the reverse harp event can be heard on the NASA SoundCloud account, and it can also be compared to an audio clip corresponding to single orbit with a standard harp event [both audio files are also in Hartinger (2025)].

Figure 4

Figure 4. The time series of the east-west magnetic field component in field-aligned coordinates, or $BFA{C}_y$ (identical to the azimuthal direction in SM coordinates), and the corresponding dynamic power spectrogram during the 3-orbits from 2012-03-17 to 2012-03-20, using the same parameters and in the same plotting format as was shown to volunteers in the HARP GUI. The audio clip for this time interval can be found in Hartinger (2025) and on https://soundcloud.com/nasa/multiple-passes-through-the-magnetospheric-harp-on-march-17-20-2012?in=nasa/sets/earths-magnetic-harp.

When the HARP GUI was publicly released to a large number of online volunteers, we were able to check whether the volunteers achieved a consensus on identifying both the standard and reverse harp events. Figure 5 shows the time period corresponding to the third orbit in Figure 4, where the dynamic power spectrogram contains the same data as in the bottom panel of Figure 4 (third orbit). The second panel of Figure 5 shows the number of marks made in the reverse harp category as a function of frequency and time (i.e., volunteers heard the pitch increase when the satellite was moving away from the Earth or decrease when the satellite was moving toward the Earth). Though there is spread in the marks (note the logarithmic scale), there is a clear clustering around the wave power enhancement corresponding to the reverse harp event in the first part of the orbit, with far fewer reverse harps and no clear clustering in the second part of the orbit. The bottom panel is for the number of marks in the standard harp category (frequency decreases as the satellite moves away from the Earth or increases as the satellite moves towards the Earth); there are significantly more marks and clustering in the second part of the orbit where a multi-harmonic rising tone feature was heard and is clearly seen in the dynamic power spectrogram. These results indicate that while there is variability in where users make their marks, they tend to cluster on wave events. Note that some of the variability seen in the heat maps corresponds to volunteers marking other types of events, such as the wave events discussed in Section 3.4.

Figure 5

Figure 5. The top panel is for a subset of data from Figure 4 corresponding to the third satellite orbit. The second panel shows the number of marks volunteers made in the reverse harp category as a function of frequency and time. The third panel is the same but for standard harp events. The audio clip for this time interval can be found in Hartinger (2025) and on https://soundcloud.com/nasa/earths-reverse-magnetospheric-harp?in=nasa/sets/earths-magnetic-harp.

We next explore why this unexpected reverse harp event occurred. As noted at the beginning of Section 3, the radial Alfvén speed profile is a major controlling factor for standing Alfvén wave frequency. Figure 6 shows, from top to bottom, electron density, magnetic field strength, Alfvén speed, and standing Alfvén frequency for 5 orbits of the THEMIS-E spacecraft, with the first three orbits corresponding to the observations in Figure 4. The methods used to obtain the quantities are the same as in Archer et al. (2015); (2017), with the density obtained from satellite measurements and the magnetic field from the Tsyganenko and Stern (1996) empirical model with median conditions over all orbits (time varying satellite measurements cannot be used for magnetic field since the Alfvén frequency calculation requires self-consistent magnetic fields). Based on comparisons with satellite measurements, the Tsyganenko and Stern (1996) model reproduces the magnetic field during the period of interest fairly well, apart from a systematic (i.e., same across multiple orbits) overestimate of absolute magnitude near apogee which would not be expected to affect trends in standing Alfvén wave frequency estimates over several hours that are of interest for this study, though it may affect absolute frequencies.

Figure 6

Figure 6. Time series of electron density, magnetic field from Tsyganenko and Stern (1996), Alfvén speed, and Alfvén frequency on subsequent orbits before, during, and after the reverse harp event on 19 March 2012. In the top panel, the gray lines are for the original density measurements, while blue lines in all panels represent results from the methods of Archer et al. (2015); (2017). To make overall trends clearer, a 5 h median filter was also applied with output smoothed using a 3 h moving average (orange).

From the top panel of Figure 6, it is clear there is variability in the density, and this translates to variability in the Alfvén speed (third panel). The density would normally be expected to monotonically decrease when outbound on 19 March until the satellite reaches apogee at 15:10 UT, then monotonically increase until perigee at 03:08 UT on 20 March. Instead, as seen in the middle part of the top panel (i.e., third orbit), after 15:10 UT when the satellite is inbound it first observed the expected increasing trend, then an unexpected decreasing trend, before eventually returning to the expected increasing trend on the remainder of its inbound pass. The bottom panel is for the standing Alfvén frequency, where the 3 hour average (orange line) of the Alfvén speed shows significant variation from orbit to orbit. Notably, the first two orbits have the characteristic U-shape for the frequency that’s also seen in the spectrogram (Figure 4), while the third orbit has a notable break in the U-shape trend near apogee, with the frequency decreasing, increasing, flattening, and then increasing rapidly towards perigee; this is also consistent with the features in the spectrogram, and it strongly suggests that an inverted Alfvén continuum led to the reverse harp event that was seen and heard by the volunteers.

Further analysis (not shown) of electron density, Alfvén speed, and standing Alfvén frequency as a function of radial distance from the Earth indicates that the outbound passes of the THEMIS-E spacecraft were in a local time sector with different density structure than the inbound passes, leading to the situation where a reverse harp event occurred on the outbound pass whereas a standard harp event occurred on the inbound pass.

This reverse harp event occurred during the recovery phase of a minor geomagnetic storm, with a minimum Dst of −69 nT roughly 2 days before the event. During and just prior to the event, there were dynamic pressure variations in the solar wind, and the solar wind speed was $>$500 km/s. The specific wave properties observed likely related to (1) driving conditions in the previous 1–2 days that caused the specific plasma mass density and magnetic field conditions during the event (thus wave speeds and presence of inverted profile, resonance conditions, etc.) and (2) the driving conditions during the event affecting the wave amplitudes and other properties. Future studies should examine what driving conditions lead to inverted Alfvén speed profiles more generally, as such conditions can affect many different wave modes beyond just causing the reverse harp events considered here. We finally note that we explored whether variations in magnetic field may also have contributed to the inverted profile by comparing fractional changes in density and magnetic pressure during the periods of interest. In this particular event the density changes were larger, indicating that these changes played the dominant role in creating the inverted profile. In other events, however, this may well not be the case, and this should be explored further in future work.

3.2 Statistical analysis

Returning to Figure 3, we next examine statistical results for events in the standard harp (expected trend, left 4 bars) and reverse harp (unexpected trend, right 4 bars) categories. First, there are far more events in the standard harp category compared to the reverse harp category. However, there are a non-negligible number of reverse harp events (note the y-axis range), suggesting these events, while more rare than standard harp events, are not exceedingly rare. Second, there were more events identified at dawn when compared to dusk for both categories of events, though the standard harp events had a significantly larger asymmetry (note the size difference between the orange and purple bars on the left).

Figure 3 shows another trend related to frequency. For the standard harp events in the dawn sector (two orange bars in the left part of the chart), more marks were made for inbound versus outbound satellite passes. This trend is not seen for dusk events (two purple bars in the left part of the chart) or for the reverse harp events (compare the two orange bars in the right part of the chart, also compare the two purple bars in the right part of the chart). This trend is unexpected since (1) inbound and outbound satellite passes were presented to the volunteers in equal number (each 3-orbit interval has 3 inbound and 3 outbound passes), (2) there is no physical reason why the occurrence of the standard harp events should depend on whether the satellite is moving towards or away from the Earth. Thus, some other explanation is needed. We propose three possibilities that should be investigated in future work:

1. Human audio perception of rising versus falling tones: Past studies have found that humans perceive rising and falling tones differently, and they tend to overestimate pitch changes for rising tones versus falling tones Neuhoff (1998). This may translate to a tendency to pick out more rising tones for certain harp events. However, since the trend was only found for dawn-sector standard harp events (the category with the most marks), some other factor is likely in play.

2. Unexpectedly strong frequency shifts in spacecraft reference frame during inbound passes: (Anderson et al., 1989): showed that in the spacecraft frame, the frequency of standing Alfvén waves is often distorted due to the convergence of both the spacecraft motion, spatial gradients in Alfvén wave frequency and phase velocity, and the temporal variability of the waves. They found that for typical gradients in Alfvén frequency (i.e., the standard harp events), “the measured frequency is lower than the true frequency on the outbound leg of the orbit while it is anomalously high on the inbound pass.” For a satellite orbit somewhat similar to THEMIS, the frequency error was found to be on the order of 15% at a radial distance of roughly 5 Earth radii and 4% at 8 Earth radii. Though these effects are rarely considered in case studies or statistical analyses of standing Alfvén waves, they may have affected the results in this study.

3. Subtle sampling bias: Though the same number of inbound and outbound passes are shown to volunteers, and they are also present in equal numbers in the database, it is possible that the number of inbound and outbound passes at specific local times on the edges of our considered ranges for dawn and dusk may slightly differ due to the precession of the satellite apogee. This fact, combined with the known differences in local time dependence in the occurrence rates of standing Alfvén waves, could explain some of the differences seen in Figure 3; however, it seems implausible that such a large difference would be found unless there are local time variations in standard Alfén wave occurrence rates that are more extreme than suggested in past studies (e.g., if the occurrence rate at 17 MLT is 50% lower than at 16 MLT).

When marking wave events, volunteers were able to record text comments that can be used to gain more insight into the trends they saw and heard. Figure 7 shows word clouds for a summary of unique words for events in the standard and reverse harp categories. Note that the size of each word is proportional to their occurrence in the volunteer comments. While linking particular words with aspects of the wave phenomena is out of scope for this paper, we note that there are particular words that occur more often in each category. For example, reverse harp events are more likely to be associated with “chirping,” “crackling,” and “drumbeat.” Further analysis is needed to, for example, account for the fact that some volunteers were far more verbose than others, to better understand which groups of words were common across a broad swath of volunteers.

Figure 7

Figure 7. Word clouds for the “standard harp” and “reverse harp” event subsets. The size of each word is proportional to its occurrence in the volunteer comments. Note that comments from all volunteers (not just rated volunteers) were included, and that this includes comments from all events.

Finally, Figure 8 shows additional statistical results for the pitch change volunteers recorded when marking all wave events, not just standard and reverse harp events (i.e., the “up” and “down” pitch change categories). First, the Figure shows there are significant differences between the dawn and dusk distributions. At dusk, the “same” and “up-down” categories corresponding to events with no pitch/frequency change or non-monotonic frequency changes, had nearly the same number of events as the “up” and “down” categories. This could be consistent with the fact that plasma density structures such as plumes tend to occur more often in the dusk sector, and perhaps make it easier to support wave events with constant frequency or non-monotonic frequency changes. It could also be consistent with the occurrence of other categories of wave events, such as ion cyclotron waves (see Section 3.4). In contrast, at dawn the categories with by far the most events are “up” and “down,” consistent with standard and reverse harp events. As previously noted, there is a strong preference at dawn for events with increasing pitch (the “up” category), which could be consistent with any of the explanations listed above. However, no such preference for events with increasing pitch occurs at dusk, where events in the “up” and “down” category occur in roughly equal numbers. As noted above, future work is needed to determine what caused this preference for events with increasing pitch at dawn.

Figure 8

Figure 8. Number of marks in different categories for pitch direction. Only marks from rated volunteers are included in this Figure, and marks from tutorial (guided) events are excluded.

3.3 Implications for future standing Alfvén wave studies

In this section, we showed results indicating that inverted radial dependencies for Alfvén frequencies (frequency increases with increasing radial distance) can occur in the outer magnetosphere with 1283 marks (12.6%) following this unexpected trend and 8,932 marks (87.4%) following the expected trend (Figure 3). It is important to note that these dependencies are happening in the outer magnetosphere, close to the magnetopause (the THEMIS apogee is roughly 13 Earth radii), thus are unlikely to be attributed to plasmaspheric plume (density) structures in the inner magnetosphere, though may be associated with plumes in regions where the structure extends all the way to the magnetopause. During a case study investigation (Section 3.1), we found that the frequency dependence was very likely due to an Alfvén speed profile that increased with increasing radial distance near the THEMIS apogee.

To our knowledge, no theory or modeling investigations have explored toroidal mode properties during conditions where the Alfvén speed increases with radial distance all the way to the magnetopause, so we do not currently have a benchmark expectation for wave properties in the same way as for Alfvén speed profiles that decrease with increasing radial distance or for profiles with inner magnetospheric structures such as plumes (Hartinger et al., 2023b). Future theory and modeling work is needed to explore wave dynamics in radial density and Alfvén speed profiles that incorporate flatter density profiles, density contributions from the Low Latitude Boundary Layer, and other factors that may lead to these unusual Alfvén speed profiles. Future observational work with the HARP dataset and other datasets should explore how often these unusual radial Alfvén speed profiles can explain the inverted Alfvén frequency continua, what factors lead to the inverted Alfvén frequency continua, and whether these events may occur during geomagnetic storms and other conditions where space weather impacts are a concern. These studies could also leverage results from past investigations of standing Alfvén waves using ground-based measurements. For example, Lanzerotti et al. (1999) found “arch” structures in visual inspection of daily dynamic power spectrograms from high-latitude ground based magnetometers, and they attributed these structures to local time frequency variations in standing Alfvén waves just inside the magnetopause. Though their focus was on local time frequency variations in contrast to the radial variations explored in this study, the same ground magnetometer measurements could be combined with satellite measurements to, for example, extract recurring patterns in wave properties and determine how often the arch structure is related to standard versus inverted Alfén continua.

We also showed in Figure 8 that the HARP volunteers identified many events where the frequency stayed the same or changed non-monotonically (i.e., did not just increase or just decrease), and that these events tended to occur more often in the dusk sector. This is consistent with the findings of Archer et al. (2015) and Archer et al. (2017) that dusk sector electron density and Alfvén speed profiles tend to be more variable than in the dawn sector. In addition to the “reverse harp” events, more work is needed to investigate whether these wave events correspond to toroidal modes or some other wave mode that is also associated with east-west magnetic signatures. We already know that HARP volunteers identified at least one other type of wave mode, which will be discussed in Section 3.4.

Finally, we found that HARP volunteers tended to identify rising tone HARP events more often than falling tone events, and we discussed several possible explanations for this trend that need to be explored in future work. If the explanation is related to the human auditory system (Neuhoff, 1998), the results would suggest that future studies using audio analysis should account for sampling biases favoring rising tones. If the explanation is due to frequency distortions related to both satellite motion and the Alfvén continuum (Anderson et al., 1989), the results would suggest that future satellite observational studies, whether using automated detection algorithms or manual methods with visual or audio analysis, should quantify and account for frequency shifts (relative to expectation in geocentric inertial frame). This could be done by separately analyzing inbound and outbound satellite passes (prior to obtaining wave event occurrence rates, etc.) and/or transforming wave fields to a geocentric inertial frame prior to analysis, though the latter approach is often not feasible without a priori knowledge of spatial gradients in Alfvén frequencies and temporal variation of the wave activity.

3.4 Identification of other plasma waves in HARP dataset

While the primary goal of the HARP project was to identify and study standing Alfvén waves, other plasma waves falling within the frequency range of study ($\sim$1–100 mHz) were also detected by volunteers. Figure 9 shows an example period where THEMIS measured electromagnetic ion cyclotron (EMIC) waves - appearing as the bright blobs/bursts just below 100 mHz. EMIC waves are electromagnetic oscillations at frequencies below and approaching ion gyrofrequencies (see white and blue lines in Figure 9 for the local O+ and He+ gyrofrequency, respectively). These waves are generated by unstable ion distributions within Earth’s magnetosphere and are more often observed during periods of geomagnetic activity (Cornwall, 1965; Horne and Thorne, 1994; Saikin et al., 2015; Usanova et al., 2012). As they can resonate with multiple particle populations in the magnetosphere, including ring current ions and radiation belt electrons (Thorne and Kennel, 1971; Jordanova et al., 2007), characterizing their spatio-temporal structure is critical for understanding their impacts on magnetospheric dynamics. The waves observed by THEMIS appear to be in the He+ frequency band and persist for multiple spacecraft orbits, lasting from at least June 19–21, 2009. While their exact structure varies from orbit to orbit, their multi-day persistence throughout this period suggests the magnetospheric impacts of long-duration regions of EMIC activity should be considered. Sonification offers an opportunity to study EMIC waves, in addition to the ULF waves of the HARP project, and potentially identify elusive and/or unexpected patterns in wave dynamics similar to the reverse harp events in this study and the storm-time poloidal mode waves in Archer et al. (2018). A follow-on study currently underway and supported by NSF (Award Number:2342095) aims to adapt the HARP sonification techniques to specifically explore EMIC wave properties and patterns in Earth’s magnetosphere.

Figure 9

Figure 9. An example EMIC wave event identified by HARP volunteers. The top and bottom panels contain the same information on the east-west magnetic field observed by THEMIS-E as would have been seen in the GUI (e.g., Figure 2), though with additional information added: the oxygen (white line) and helium (cyan line) gyrofrequencies are overlaid on the spectrogram to indicate the type of EMIC waves observed (helium band). The audio clip corresponding to this event can be found in Hartinger (2025).

4 Participant impacts, broader impacts, and lessons learned from HARP

In this section, we discuss preliminary statistics on HARP participant impacts and lessons learned from HARP, followed by examples of broader impacts from public outreach related to HARP during the Heliophysics Big Year.

4.1 Participant impacts and lessons learned

To quantitatively assess how active different volunteers were in identifying wave events, thus indirectly assess their interest in the project, we recorded how many marks each volunteer made. Figure 10 shows how many HARP volunteers made 0–10, 10–20, 20–50, 50–100, and $>$100 marks to assess volunteer engagement at different levels. There is a significant dropoff from 1487 users making 0–10 marks to 179 users making 10–20 marks. However, after that drop, the decline is much less steep, with significant numbers of users remaining in the categories with $>$20 marks. The fact that 23 users made $>$100 marks is remarkable given the time commitment this represents (several minutes per mark when accounting for the time needed to analyze the 3-orbit interval and record information in the popup window).

Figure 10

Figure 10. The distribution of the number of marks each user made.

One of the lessons learned from HARP relates to the difficulty sustaining volunteer engagement with grant funds that were only meant for building and beta testing the interface, particularly after the funded portion of the project ends. This relates to well known challenges in citizen/participatory science (Grandin et al., 2025) with maintaining long-term funding and managing public communications with a small team of scientists. For HARP, we focused most of our efforts related to volunteer communication on the initial project launch in April-June 2023, fielding questions from volunteers via an online form and during a webinar. After our project funding ended, we shifted to managing data collection, writing up the project results, and developing ideas for new projects. Future similar projects should ideally include a framework and funding for sustained volunteer interactions. Another lesson learned from HARP relates to the challenge in maintaining the interest of volunteers with a rich variety of backgrounds. For example, several volunteers requested more detailed instructions for wave event detection with the desire to work with a much narrower definition of wave activity (e.g., a binary “yes” or “no” for the presence of waves on each orbit), while others wanted the freedom to bypass the GUI entirely and build their own event detection framework with more advanced audio analysis tools. Future similar projects should ideally support contributions from volunteers with these varying interests.

Finally, the first beta test involved high school student citizen scientists at the Lafayette Public Library who, in addition to providing valuable feedback on the GUI, also received additional instruction on space plasma physics via in person sessions and a Slack channel. These 11 students were recruited through the library’s participation in the STAR Library Education Network, a network of over 12,000 public libraries who receive free resources, activities and training from the Space Science Institute’s Education and Learning Research Group. Participating students were the sole-creators of a 2021 AGU Up-Goer presentation Holland et al. (2021), using the knowledge and skills they gained to craft an engaging presentation that was presented with no revision by SSI Education staff. One student said of the project “I only signed up for this group because my friend was participating, but now I think I’m going to change my major to physics, this was just too cool.” The library staff coordinator said, “I thought this project would be too complicated for the students, it was certainly daunting for me, but they took it and ran. I’ve heard them discussing resonance events with other kids in the teen room. It is amazing how engaged they are in such a niche topic.”

4.2 Public collaborations and broader impacts

The HARP project sparked a series of high-profile collaborations that demonstrated the potential for citizen science to transcend traditional academic boundaries. These collaborations spanned commercial partnerships, media features, and artistic installations. Notably, Magnum Ice Cream commissioned a remix of JVKE’s “Golden Hour” to anchor the launch of their Sunlovers and Starchasers campaign (Rolling Stone UK, 2023). This remix, produced by Alex Metric in collaboration with Robert Alexander from the HARP team, incorporated the rising sweep of a HARP event as the vocalist sings “shine”–creating a moment where the data’s natural crescendo aligned with the song’s emotional peak. The collaboration achieved remarkable reach, generating 18.6 million impressions across a variety of platforms and capturing a 64% Gen Z audience, demonstrating how the sonification of scientific data can serve as a bridge between research communities and younger inquisitive audiences.

NASA’s Curious Universe podcast featured HARP in “Hum of the Sun” (Season 6, Episode 2), with interviews of HARP Principal Investigator Michael Hartinger and team member Robert Alexander (NASA’s Curious Universe, 2023). The episode interweaves sonified space weather audio with the scientific narrative, showcasing the ability of audification to serve as an analytical tool and a medium for communication. Host Padi Boyd succinctly noted the benefits of the approach when she stated “Heliophysicists are used to reading charts and looking at stunning images from spacecraft. But more recently they’ve discovered that by closing your eyes and trusting your ears, you can discover things you never could have seen.”

The team also collaborated with ARTECHOUSE on the “Beyond the Light” exhibit (https://www.artechouse.com/program/beyond-the-light-nyc/), working with curators to craft auditory landscapes from solar data. This installation allowed visitors to experience the sound of the solar wind as an enveloping, physical presence, transforming the gallery into a resonant chamber where electromagnetic waves became spatially embodied audio.

These collaborations demonstrate how citizen science projects can transcend traditional academic boundaries, generating cultural artifacts and media partnerships that significantly amplify research impact and create novel pathways for public engagement with space physics research.

5 Summary and future projects

In this paper, we described the design, implementation, science results, and participant impacts from the “Heliophysics Audified: Resonances in Plasmas” (HARP) project. In particular, we discussed the design and implementation of the HARP project, including the data processing and rationale, IRB review and compliance, GUI and data collection procedure, beta testing, and methods for validating volunteer results in Section 2. We further discussed the initial science results from HARP in Section 3:

1. Overall trends in volunteer wave event identifications are consistent with past studies of dawn-dusk asymmetries in multi-harmonic toroidal mode waves.

2. Volunteers detected several categories of toroidal mode events, including events with unexpected radial frequency dependence trends. These “reverse harp” events have frequencies that increase with increasing radial distance towards the magnetopause and were associated with radially increasing Alfvén speeds near the THEMIS satellite apogee (13 Earth radii) in at least one case study.

3. Volunteers detected several types of dawn-dusk asymmetries; for example, the dusk sector tended to have more events with frequencies that did not change significantly with radial distance, or changed non-monotonically (i.e., did not just increase or just decrease).

4. Volunteers detected more events with increasing frequency than decreasing frequency in the dawn sector; this was unexpected as the same number of inbound and outbound passes were presented to volunteers, thus the same number of increasing and decreasing frequency events should be present. There are several explanations for this trend discussed in Section 3 that should be explored in future work.

5. Although the HARP project was focused on the identification of standing Alfvén waves, volunteers also detected other events including EMIC waves. A follow-on project is currently underway to adapt the HARP sonification techniques to explore EMIC wave properties in the Earth’s magnetosphere.

As noted in Section 3, these are initial results and more work is needed to mine the HARP volunteer dataset and explore the source(s) of these trends further.

In Section 4, we discussed participant impacts and lessons learned from HARP, as well as broader impacts and collaborations related to HARP:

1. 1. 332 volunteers made 10 or more event markings, 50 made 50 or more event markings, and 23 made more than 100 marks (Figure 10). The latter categories represent a substantial time commitment by volunteers and suggests the GUI and related material were able to sustain the interest of a large number of volunteers.

2. 2. Volunteers had widely varying interests, from some preferring the GUI have narrower event definitions with a binary “yes/no” choice for wave activities, while others wanted the freedom to bypass the GUI entirely to build their own detection framework. Future similar projects should ideally support these varying interests and include a framework and funding for sustained volunteer interactions.

3. 3. HARP sparked a series of collaborations and public outreach activities including a remix of JVKE’s “Golden Hour,” a NASA Curious Universe podcast “Hum of the Sun,” and a collaboration with ARTECHOUSE for their “Beyond the Light” exhibit.

The early results presented here suggest many avenues for future work in addition to those listed above. As discussed in Section 1, audio analysis techniques complement visual techniques and facilitate the identification of recurring patterns in wave activity that elude standard detection algorithms. HARP follows in a series of studies developing standardized sonification techniques that target specific wave modes and/or maximize the detection of ULF waves in challenging conditions (Archer et al., 2022a), and it is hoped that this framework can continue to be used in future studies related to other wave modes using both satellite and ground-based measurements. As noted in Section 3.4, studies of EMIC waves in the Earth’s magnetosphere are already underway.

Archer et al. (2022a) note that sonification techniques make scientific data more accessible to a wider audience. This has the advantage of both (1) lowering the barrier to entry to contribute to space science research and (2) mobilizing volunteers in large numbers to arrive at robust event detections via consensus. The latter reduces concerns about biases and reliability for event detections from a single researcher and/or the use of automated detection algorithms designed by a single researcher (Barnard et al., 2014). The HARP dataset contains detailed information about HARP volunteer event identifications that permit analysis like shown in Figure 5 that can be used to refine future automated ULF wave detection algorithms, including metadata from event markings such as the volunteers’ self-assessment of confidence in the event detection. It could also be used to develop more data-driven classifications of ULF waves beyond the usual schemes based on frequency and event duration (Jacobs et al., 1964). Finally, many other citizen science investigations have demonstrated the value of applying machine learning algorithms to datasets labeled by volunteers (Beaumont et al., 2014; Sullivan et al., 2014), and future projects like HARP would ideally incorporate a process for training machine learning algorithms for wave event detection. In general, there are many paths forward to explore the full range of uses of both sonfication and citizen science in the detection of ULF waves and analysis of the wide variety of wave dynamics possible in the Earth’s magnetosphere.

Public outreach efforts and collaborations with musicians and artists related to the HARP project and past projects (Archer, 2021a) indicate that the public continues to be fascinated by space audio. Future efforts could leverage HARP audio processing tools and outreach materials to expand and sustain public interest in space weather research via space audio, including new collaborations with artists and via informal learning activities at museums and libraries. Additionally, though social media was not extensively used during the HARP project, it is being successfully utilized by other projects to introduce sonification and space audio to a broader audience via platforms like TikTok, Instagram, and YouTube. For example, HARP team member Lauren Blum is Principal Investigator of a separate NSF project to study plasma waves using sonification, and as part of that project undergraduate researcher Tiffiny Costello is creating blogs, videos, and content for social media to share findings in an exciting and relevant way (Example: Listen to a CME).

Finally, the HARP project presents many opportunities for future collaborations with the blind and low-vision community using audio analysis. Indeed, this was one of the most common suggestions to the HARP team during the project, though due to the limited duration and budget of our seed funding grant we were not able to pursue it. In future projects, we aim to partner with researchers and others in the blind/low-vision community to create a new platform for event detection optimized for blind/low-vision volunteers.

Data availability statement

The datasets presented in this study can be found in Hartinger (2025). The data processing software can be found at github https://github.com/Shirling-VT/HARP_sonification.

Ethics statement

The studies involving human participants were reviewed and approved by the UCLA Institutional Review Board; the participants provided their written informed consent to participate in this study as discussed in Section 2.2.

Author contributions

MH: Writing – original draft, Writing – review and editing. MA: Writing – review and editing, Writing – original draft. EM: Writing – review and editing, Writing – original draft. XS: Writing – original draft, Writing – review and editing. RA: Writing – review and editing, Writing – original draft. EV: Writing – review and editing. AH: Writing – original draft, Writing – review and editing. JH: Writing – review and editing. JL: Writing – review and editing. LB: Writing – review and editing, Writing – original draft. SC: Writing – review and editing. RC: Writing – review and editing. KC: Writing – review and editing, Writing – original draft. VA: Writing – review and editing. TC: Writing – original draft, Writing – review and editing. LW: Writing – review and editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. MH, AH, JH, and EV were supported by NASA grant 80NSSC21K0796. LB, XS, MH, TC, and LW were supported by NSF AGS grant 2342095. MH was supported by NASA grant 80NSSC23K0903. MA was supported by UKRI (STFC/EPSRC) Stephen Hawking Fellowship EP/T01735X/1 and UKRI Future Leaders Fellowship MR/X034704/1. KC was supported by NSF OPP grant 2218996. EM and VA were supported by NASA grant NAS5-02099. XS was supported by NASA grant 80NSSC21K1677.

Acknowledgements

We thank all the volunteers and beta testers that contributed to all phases of the HARP project, regardless of whether they marked wave events. We thank Alessandra Pacini for feedback on the HARP GUI design, expert wave identification, and leading the space weather portion of the HARP webinar. We thank Joseph B.H. Baker for feedback on the HARP GUI design. We acknowledge NASA contract NAS5-02099 for use of data from the THEMIS Mission. Specifically: C. W. Carlson and J. P. McFadden for use of ESA data; K. H. Glassmeier, U. Auster and W. Baumjohann for the use of FGM data provided under the lead of the Technical University of Braunschweig and with financial support through the German Ministry for Economy and Technology and the German Center for Aviation and Space (DLR) under contract 50 OC 0302.

Conflict of interest

Author RA was employed by Auralab Technologies, Inc.

The remaining author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Claude Sonnet 4 was used in only two parts of the manuscript: (1) to adapt the publicly available HARP software and datasets to generate the heat maps in Figure 5 and (2) to assist in writing a portion of the text in Section 1 (some of the text in the paragraph beginning “Audio analysis provides”) and the text in Section 4.2. It was not used in any other Figures or text in the manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

¹https://cdaweb.gsfc.nasa.gov/audification_readme.html

References

Alexander, R. L., O‘Modhrain, S., Roberts, D. A., Gilbert, J. A., and Zurbuchen, T. H. (2014). The bird‘s ear view of space physics: audification as a tool for the spectral analysis of time series data. J. Geophys. Res. (Space Phys.) 119, 5259–5271. doi:10.1002/2014JA020025

CrossRef Full Text | Google Scholar

Anderson, B. J., Engebretson, M. J., and Zanetti, L. J. (1989). Distortion effects in spacecraft observations of MHD toroidal standing waves: theory and observations. J. Geophys. Res. 94, 13425–13445. doi:10.1029/JA094iA10p13425

CrossRef Full Text | Google Scholar

Anderson, B. J., Engebretson, M. J., Rounds, S. P., Zanetti, L. J., and Potemra, T. A. (1990). A statistical study of Pc 3-5 pulsations observed by the AmMPTE/CCE magnetic fields experiment 1. Occurrence distributions. J. Geophys. Res. 95, 10495–10523. doi:10.1029/JA095iA07p10495

CrossRef Full Text | Google Scholar

Angelopoulos, V. (2008). The THEMIS Mission. Space Sci. Rev. 141, 5–34. doi:10.1007/s11214-008-9336-1

CrossRef Full Text | Google Scholar

Archer, M. O. (2021a). “Hearing the invisible: the versatilty of sound in engaging,” in Communicating Astronomy with the Public Conference 2021 (Paris: International Astronomical Union), 31–32.

Google Scholar

Archer, M. O. (2021b). Schools of all backgrounds can do physics research – on the accessibility and equity of the Physics Research in School Environments (PRiSE) approach to independent research projects. Geosci. Commun. 4, 189–208. doi:10.5194/gc-4-189-2021

CrossRef Full Text | Google Scholar

Archer, M. O., and DeWitt, J. (2021). “Thanks for helping me find my enthusiasm for physics”: the lasting impacts ”research in schools” projects can have on students, teachers, and schools. Geosci. Commun. 4, 169–188. doi:10.5194/gc-4-169-2021

CrossRef Full Text | Google Scholar

Archer, M. O., Hartinger, M. D., Walsh, B. M., Plaschke, F., and Angelopoulos, V. (2015). Frequency variability of standing Alfvén waves excited by fast mode resonances in the outer magnetosphere. Geophys. Res. Lett. 42 (10), 150–159. doi:10.1002/2015GL066683

CrossRef Full Text | Google Scholar

Archer, M. O., Hartinger, M. D., Walsh, B. M., and Angelopoulos, V. (2017). Magnetospheric and solar wind dependences of coupled fast-mode resonances outside the plasmasphere. J. Geophys. Res. (Space Phys.) 122, 212–226. doi:10.1002/2016JA023428

CrossRef Full Text | Google Scholar

Archer, M. O., Hartinger, M. D., Redmon, R., Angelopoulos, V., and Walsh, B. M. (2018). First results from sonification and exploratory citizen science of magnetospheric ULF waves: long-lasting decreasing-frequency poloidal field line resonances following geomagnetic storms. Space Weather. 16, 1753–1769. doi:10.1029/2018SW001988

CrossRef Full Text | Google Scholar

Archer, M. O., Hietala, H., Hartinger, M. D., Plaschke, F., and Angelopoulos, V. (2019). Direct observations of a surface eigenmode of the dayside magnetopause. Nat. Commun. 10, 615. doi:10.1038/s41467-018-08134-5

PubMed Abstract | CrossRef Full Text | Google Scholar

Archer, M. O., DeWitt, J., Thorley, C., and Keenan, O. (2021). Evaluating participants’ experience of extended interaction with cutting-edge physics research through the PRiSE “research in schools” programme. Geosci. Commun. 4, 147–168. doi:10.5194/gc-4-147-2021

CrossRef Full Text | Google Scholar

Archer, M. O., Cottingham, M., Hartinger, M. D., Shi, X., Coyle, S., Hill, E., et al. (2022a). Listening to the magnetosphere: how best to make ULF waves audible. Front. Astronomy Space Sci. 9, 877172. doi:10.3389/fspas.2022.877172

CrossRef Full Text | Google Scholar

Archer, M. O., Southwood, D. J., Hartinger, M. D., Rastaetter, L., and Wright, A. N. (2022b). How a realistic magnetosphere alters the polarizations of surface, fast magnetosonic, and Alfvén waves. J. Geophys. Res. (Space Phys.) 127, e2021JA030032. doi:10.1029/2021JA030032

PubMed Abstract | CrossRef Full Text | Google Scholar

Barnard, L., Scott, C., Owens, M., Lockwood, M., Tucker-Hood, K., Thomas, S., et al. (2014). The solar stormwatch CME catalogue: results from the first space weather citizen science project. Space Weather. 12, 657–674. doi:10.1002/2014SW001119

CrossRef Full Text | Google Scholar

Beaumont, C. N., Goodman, A. A., Kendrew, S., Williams, J. P., and Simpson, R. (2014). The Milky Way Project: leveraging citizen science and machine learning to detect interstellar bubbles. Astrophysical J. Suppl. Ser. 214, 3. doi:10.1088/0067-0049/214/1/3

CrossRef Full Text | Google Scholar

Cartwright, M., Dove, G., Méndez Méndez, A. E., Bello, J. P., and Nov, O. (2019). “Crowdsourcing multi-label audio annotation tasks with citizen scientists,” in Proceedings of the 2019 CHI conference on human factors in computing systems (Glasgow: Association for Computing Machinery), 1–11. doi:10.1145/3290605.330052

CrossRef Full Text | Google Scholar

Collins, K., Alexander, R., Verniero, J., and Candey, R. (2024). “Auralization of magnetic multiscale satellite data: toward integrated audification in space science,” in Proceedings of the international conference on auditory display (Atlanta: Georgia Tech Library). doi:10.21785/icad2024.000

CrossRef Full Text | Google Scholar

Cornwall, J. M. (1965). Cyclotron instabilities and electromagnetic emission in the ultra low frequency and very low frequency ranges. J. Geophys. Res. 70, 61–69. doi:10.1029/JZ070i001p00061

CrossRef Full Text | Google Scholar

Dombois, F., and Eckel, G. (2011). “Audification,” in The Sonification Handbook (Berlin: Logos Publishing House).

Google Scholar

Elsden, T., and Wright, A. N. (2022). Polarization properties of 3-D field line resonances. J. Geophys. Res. (Space Phys.) 127, e2021JA030080. doi:10.1029/2021JA030080

CrossRef Full Text | Google Scholar

Engebretson, M., Glassmeier, K.-H., Stellmacher, M., Hughes, W. J., and Lühr, H. (1998). The dependence of high-latitude Pc5 wave power on solar wind velocity and on the phase of high-speed solar wind streams. J. Geophys. Res. 103, 26271–26384. doi:10.1029/97JA03143

CrossRef Full Text | Google Scholar

Grandin, M., Ledvina, V. E., Musset, S., Partamies, N., Frissell, N. A., Bruus, E., et al. (2025). Citizen science in space and atmospheric sciences: opportunities and challenges. Surv. Geophys. doi:10.1007/s10712-025-09888-6

CrossRef Full Text | Google Scholar

Hartinger, M. D. (2025). “Heliophysics Audified: resonances in Plasmas” datasets and audio files [Data set]. Geneva: Zenodo. doi:10.5281/zenodo.17064808

CrossRef Full Text | Google Scholar

Hartinger, M. D., Elsden, T., Archer, M. O., Takahashi, K., Wright, A. N., Artemyev, A., et al. (2023a). Properties of magnetohydrodynamic normal modes in the Earth’s magnetosphere. J. Geophys. Res. (Space Phys.) 128, e2023JA031987. doi:10.1029/2023JA031987

CrossRef Full Text | Google Scholar

Hartinger, M. D., Shi, X., Rodger, C. J., Fujii, I., Rigler, E. J., Kappler, K., et al. (2023b). Determining ULF wave contributions to geomagnetically induced currents: the important role of sampling rate. Space Weather. 21, e2022SW003340. doi:10.1029/2022SW003340

CrossRef Full Text | Google Scholar

Helliwell, R. A. (1965). Whistlers and related ionospheric phenomena. Stanford, California: Stanford University Press.

Google Scholar

Holland, A., Hartinger, M., Harold, J., and Masongsong, E. (2021). “Working WITH teen stakeholders to make sense of Heliophysics Science (sun learning, as it were!),” in AGU fall meeting abstracts. vol. 2021 of AGU fall meeting abstracts (New Orleans: American Geophysical Union). ED23A–11.

Google Scholar

Horne, R. B., and Thorne, R. M. (1994). Convective instabilities of electromagnetic ion cyclotron waves in the outer magnetosphere. J. Geophys. Res. 99, 17259–17274. doi:10.1029/94JA01259

CrossRef Full Text | Google Scholar

Jacobs, J. A., Kato, Y., Matsushita, S., and Troitskaya, V. A. (1964). Classification of geomagnetic micropulsations. J. Geophys. Res. 69, 180–181. doi:10.1029/JZ069i001p00180

CrossRef Full Text | Google Scholar

Jordanova, V. K., Spasojevic, M., and Thomsen, M. F. (2007). Modeling the electromagnetic ion cyclotron wave-induced formation of detached subauroral proton arcs. J. Geophys. Res. (Space Phys.) 112, A08209. doi:10.1029/2006JA012215

CrossRef Full Text | Google Scholar

Kramer, G., Walker, B., Bonebright, T., Cook, P., Flowers, J. H., Miner, N., et al. (2010). Sonification report: status of the field and research agenda. Lincoln: University of Nebraska.

Google Scholar

LaBelle, J., and Treumann, R. A. (2002). Auroral radio emissions, 1. hisses, roars, and bursts. Space Sci. Rev. 101, 295–440. doi:10.1023/A:1020850022070

CrossRef Full Text | Google Scholar

Landi, E., Alexander, R. L., Gruesbeck, J. R., Gilbert, J. A., Lepri, S. T., Manchester, W. B., et al. (2011). Carbon ionization stages as a diagnostic of the solar wind. Astrophysical J. 744, 100. doi:10.1088/0004-637X/744/2/100

CrossRef Full Text | Google Scholar

Lanzerotti, L. J. (2017). Space weather: historical and contemporary perspectives. Space Sci. Rev. 212, 1253–1270. doi:10.1007/s11214-017-0408-y

CrossRef Full Text | Google Scholar

Lanzerotti, L. J., Shono, A., Fukunishi, H., and Maclennan, C. G. (1999). Long-period hydromagnetic waves at very high geomagnetic latitudes. J. Geophys. Res. 104, 28423–28436. doi:10.1029/1999JA900325

CrossRef Full Text | Google Scholar

Lehikoinen, P., Rannisto, M., Camargo, U., Aintila, A., Lauha, P., Piirainen, E., et al. (2023). A successful crowdsourcing approach for bird sound classification. Citiz. Sci. Theory Pract. 8, 16. doi:10.5334/cstp.556

CrossRef Full Text | Google Scholar

Moore, T. E., Gallagher, D. L., Horwitz, J. L., and Comfort, R. H. (1987). MHD wave breaking in the outer plasmasphere. Geophys. Res. Lett. 14, 1007–1010. doi:10.1029/GL014i010p01007

CrossRef Full Text | Google Scholar

NASA’s Curious Universe (2023). Hum of the sun. Podcast. Season 6, Episode 2. Available online at: https://www.nasa.gov/podcasts/curious-universe/the-hum-of-the-sun/.

Google Scholar

Neuhoff, J. (1998). Perceptual bias for rising tones. Nature 395, 123–124. doi:10.1038/25862

PubMed Abstract | CrossRef Full Text | Google Scholar

Perry, G., Erickson, P. J., Blain, B. D., Reif, R., and Frissell, N. A. (2020). “Update on the golden ears project,” in HamSCI Workshop 2020 (Scranton, PA: HamSCI: HamSCI).

Google Scholar

Rolling Stone UK (2023). Tiktok star jvke announces new version of ’golden hour’ featuring sounds from nasa library. Rolling Stone UK. Available online at: https://www.rollingstone.co.uk/culture/tiktok-star-jvke-announces-new-version-of-golden-hour-featuring-sounds-from-nasa-library-29304/.

Google Scholar

Saikin, A. A., Zhang, J. C., Allen, R. C., Smith, C. W., Kistler, L. M., Spence, H. E., et al. (2015). The occurrence and wave properties of H⁺-He⁺-and O⁺-band EMIC waves observed by the Van Allen Probes. J. Geophys. Res. (Space Phys.) 120, 7477–7492. doi:10.1002/2015JA021358

CrossRef Full Text | Google Scholar

Sazhin, S. S., and Hayakawa, M. (1992). Magnetospheric chorus emissions: a review. Planet. Space Sci. 40, 681–697. doi:10.1016/0032-0633(92)90009-d

CrossRef Full Text | Google Scholar

Shi, X., Chakraborty, S., Baker, J. B. H., Hartinger, M. D., Wang, W., Ruohoniemi, J. M., et al. (2025). Statistical characterization of Joule heating associated with ionospheric ULF perturbations using SuperDARN data. J. Geophys. Res. Space Phys. 130, e2024JA033452. doi:10.1029/2024JA033452

CrossRef Full Text | Google Scholar

Smith, E. J., and Tsurutani, B. T. (1976). Magnetosheath lion roars. J. Geophys. Res. 81, 2261–2266. doi:10.1029/JA081i013p02261

CrossRef Full Text | Google Scholar

Sugiura, M., and Wilson, C. R. (1964). Oscillation of the geomagnetic field lines and associated magnetic perturbations at conjugate points. J. Geophys. Res. 69, 1211–1216. doi:10.1029/JZ069i007p01211

CrossRef Full Text | Google Scholar

Sullivan, B. L., Aycrigg, J. L., Barry, J. H., Bonney, R. E., Bruns, N., Cooper, C. B., et al. (2014). The ebird enterprise: an integrated approach to development and application of citizen science. Biol. Conserv. 169, 31–40. doi:10.1016/j.biocon.2013.11.003

CrossRef Full Text | Google Scholar

Takahashi, K., Hartinger, M. D., Angelopoulos, V., and Glassmeier, K.-H. (2015). A statistical study of fundamental toroidal mode standing Alfvén waves using THEMIS ion bulk velocity data. J. Geophys. Res. (Space Phys.) 120, 6474–6495. doi:10.1002/2015JA021207

CrossRef Full Text | Google Scholar

Thorne, R. M., and Kennel, C. F. (1971). Relativistic electron precipitation during magnetic storm main phase. J. Geophys. Res. 76, 4446–4453. doi:10.1029/JA076i019p04446

CrossRef Full Text | Google Scholar

Tsyganenko, N. A., and Stern, D. P. (1996). Modeling the global magnetic field of the large-scale Birkeland current systems. J. Geophys. Res. 101, 27187–27198. doi:10.1029/96JA02735

CrossRef Full Text | Google Scholar

United Nations Office of Outer Space Affairs (2023). Sonification: a tool for research, outreach and inclusion in space sciences. Vienna: United Nations.

Google Scholar

Usanova, M. E., Mann, I. R., Bortnik, J., Shao, L., and Angelopoulos, V. (2012). THEMIS observations of electromagnetic ion cyclotron wave occurrence: dependence on AE, SYMH, and solar wind dynamic pressure. J. Geophys. Res. (Space Phys.) 117, A10218. doi:10.1029/2012JA018049

CrossRef Full Text | Google Scholar

Wicks, R. T., Alexander, R. L., Stevens, M., Wilson III, L. B., Moya, P. S., Viñas, A., et al. (2016). A proton-cyclotron wave storm generated by unstable proton distribution functions in the solar wind. Astrophysical J. 819, 6. doi:10.3847/0004-637X/819/1/6

CrossRef Full Text | Google Scholar

Wright, A. N., Hartinger, M. D., Takahashi, K., and Elsden, T. (2024). “Alfvén waves in the Earth’s magnetosphere,”. Alfvén waves across heliophysics: progress, challenges, and opportunities. Editor A. Keiling (Hoboken: American Geophysical Union), 285, 215–247. doi:10.1002/9781394195985.ch10

CrossRef Full Text | Google Scholar