Unprecedented 19 Day Type IV Radio Burst as a Corotating Electron Reservoir

We use radio measurements from Solar Orbiter, Wind, Parker Solar Probe, and STEREO-A, which together provide complementary longitudes and heliocentric distances. The Radio and Plasma Waves (RPW) instrument on Solar Orbiter (D. Müller et al. 2013; M. Maksimovic et al. 2020) records spectral power with the Thermal Noise Receiver (TNR; 2 kHz–1 MHz) and the High Frequency Receiver (HFR; 375 kHz–16 MHz). Level-2 V² Hz⁻¹ spectra are converted to solar flux units (sfu; 1 sfu = 10⁻²² W m⁻² Hz⁻¹) and normalized to a 1 au distance using a 1/r² profile, with the calibration summarized in Appendix A.

The WAVES instrument on Wind (J.-L. Bougeret et al. 1995; L. B. Wilson et al. 2021) provides RAD1 (20–1040 kHz) and RAD2 (1075 kHz–13 MHz) spectra. We convert V² Hz⁻¹ to sfu with the instrument responses (Appendix A). For this event, Wind direction finding was not used because the available calibrated RAD1 product assumes an unpolarized source, whereas our event is strongly polarized.

Parker Solar Probe carries the FIELDS suite (S. D. Bale et al. 2016; N. J. Fox et al. 2016), which includes a radio-frequency receiver providing low-frequency spectra and polarization. We use Parker Solar Probe/FIELDS flux and degree of circular polarization V/I measurements in the 0.5–3 MHz band during 2025 September 4 00:00–September 6 14:00 (Figures 1(c), (d)), when Parker Solar Probe was at ∼0.4 au, to complement Wind and to provide an additional polarization constraint at a smaller heliocentric distance.

The STEREO/WAVES experiment on STEREO-A (J. L. Bougeret et al. 2008; M. L. Kaiser et al. 2008) measures radio power, polarization, and wavevector direction with a triad of monopole antennas. We use Level-3 data (already in physical units of sfu) from 125 kHz to 2 MHz, which provide Stokes I, V, the wavevector direction (azimuth and colatitude in RTN), and an apparent source size γ at STEREO-A (V. Krupar et al. 2022). Direction-finding products are available only up to 2 MHz; consequently, the STEREO-A polarization and direction-finding panels in Figure 1 show an apparent upper-frequency cutoff at 2 MHz.

From 2025 August 21 to September 9, a single long-lived type IV continuum was detected in three successive visibility windows by four spacecraft as solar rotation carried the source region from the farside into the Earth-facing hemisphere and onward toward STEREO-A longitudes. We refer to these three intervals as Windows 1–3 (Figure 1); they represent observer-dependent visibility of the same corotating reservoir rather than three independent bursts.

The observing geometry is summarized in Figure 2. The spacecraft heliocentric distances and heliolongitudes during the three windows (given above each panel in Figure 1) were Solar Orbiter at 120 R_⊙ and λ ≈ −174° (Window 1), Wind at 213 R_⊙ and λ ≈ 0° and Parker Solar Probe at 86 R_⊙ and λ ≈ 9° (Window 2), and STEREO-A at 207 R_⊙ and λ ≈ 47° (Window 3). The horizontal bars in Figure 2(e) indicate these visibility windows in longitude-time space, while the polar view in Figure 2(a) places the observers relative to the CME propagation wedges. A linear fit to the window centroids yields synodic rotation rates ω = 1268 ± 127 day⁻¹ (SolO → Wind/Parker Solar Probe) and ω = 1316 ± 132 day⁻¹ (Wind/Parker Solar Probe → STEREO-A), corresponding to periods P ≈ 28.38 ± 2.84 and 27.35 ± 2.74 days (Figure 2(e)).

Figure 1(a) shows Window 1: a weak continuum around 1 MHz on 2025 August 21 16:00–August 22 03:00 detected by Solar Orbiter at ∼0.6 au while the source region was on the farside as viewed from Earth. Despite spacecraft electromagnetic disturbances (M. Maksimovic et al. 2021), the spectra reveal a broad continuum characteristic of interplanetary type IV emission.

Twelve days later, as the source rotated into a more favorable geometry, the continuum intensified and persisted much longer. Window 2 was observed near Earth by Wind on 2025 September 3 09:00–September 5 22:00 (Figure 1(b)). Parker Solar Probe observed the same continuum on 2025 September 4 00:00–September 6 14:00 (Figure 1(c)) with concurrent circular polarization V/I measurements (Figure 1(d), negative values correspond to left-hand circular polarization), overlapping the Wind interval and confirming continuity of the corotating reservoir. Wind and Parker Solar Probe also observed enhanced type III activity in the days surrounding CME2 (2025 August 30), including a pronounced type III storm and complex type III bursts (Figures 3(c),(d)). This storm-like activity suggests sustained electron injections onto open magnetic field lines during CME-driven magnetic reconfiguration.

One day later, Window 3 was observed by STEREO-A on 2025 September 7 00:30–September 9 09:00 (Figure 1(e)); at that time STEREO-A was located ∼47° ahead of Earth in heliolongitude. The corresponding V/I (Figure 1(f)) is predominantly negative (nearly pure left-hand circular polarization), indicating that the continuum is dominated by a single magnetoionic mode.

Figures 1(g) and 1(h) show the STEREO-A wavevector direction in the RTN frame ( from the Sun to the spacecraft, prograde in the orbital plane, and ). The azimuth decreases steadily from eastward (positive) toward ≈0°, consistent with slow rotation-parallax drift, while the colatitude remains slightly <90°, indicating a modest southward offset. Finally, Figure 1(i) shows the apparent source size γ, which varies between ∼20° and ∼40°, with smaller sizes coincident with the brightest intervals.

We also note a systematic downward drift of the continuum’s characteristic frequency and a gradual narrowing of its bandwidth over each visibility window. In Window 3, the bright ridge shifts from ∼1 to 2 MHz near the start toward ∼0.5–1 MHz near the end. A similar drift is apparent in the Wind and Parker Solar Probe spectra during Window 2. Because this trend is observed from multiple heliolongitudes, it is unlikely to arise solely from a single structure propagating radially through the solar wind; instead, it may reflect corotation and changing visibility of different portions of a large loop/streamer-top trap as the system rotates.

The sequential appearance of the continuum at the three vantage points, with delays consistent with solar rotation, suggests a fixed source region that becomes detectable only within an effective longitudinal visibility window (a lighthouse effect). During the strong phase, the visibility durations are 61 hr at Wind, 62 hr at Parker Solar Probe, and 56.5 hr at STEREO-A (Window 2 and Window 3; Figure 1). For a corotating source, these durations imply an effective detectability width W_vis ≈ Ω_⊙T_vis ≃ 31°–34° (full width) for Ω_⊙ ≈ 132 day⁻¹ (consistent with the rotation rates inferred from the window-to-window handoffs). An independent constraint comes from the rapid Parker Solar Probe → STEREO-A handoff: the continuum ends at Parker Solar Probe on 2025 September 6 14:00 and begins at STEREO-A on 2025 September 7 00:30, while the spacecraft longitudes differ by Δλ ≃ 38° (Parker Solar Probe at λ ≈ 9°, STEREO-A at λ ≈ 47°; Figure 2(e)). The 10.5 hr gap corresponds to only Ω_⊙Δt ≈ 6° of solar rotation with no detection at either spacecraft; interpreting this as a blind longitudinal sector implies W_vis ≈ Δλ − Ω_⊙Δt ≈ 32° (about ±16° about central meridian). We stress that W_vis is an effective detectability window shaped by intrinsic anisotropy, occultation by dense coronal/CME structures, and propagation effects (refraction/scattering) at hectometric frequencies, rather than a literal narrow emission beam. Consequently, nondetections between the windows should not be interpreted as cessation of the reservoir; the emission can persist but fall below detectability outside the visibility window. In Section 2.3, we therefore use the more restrictive, high-confidence duration within Window 3 to estimate the geometric transverse diameter W_rot of the emitting trap, while W_vis should be regarded as an effective detectability width (upper limit).

Three fast CMEs erupted during the interval—CME1 on 2025 August 21, CME2 on August 30, and CME3 on September 4—with sheath (shock-front) speeds of ≈1375, ≈1265, and ≈1379 km s⁻¹, respectively (propagation directions and angular widths summarized in Figure 2).¹⁰ CME1 and CME2 were accompanied by hectometric type II radio emission; for CME2, the associated interplanetary shock was also sampled in situ near Earth by Wind. CME3 was followed by an unusually low solar wind proton density at Wind, n_p ∼ 0.1 cm⁻³ (the absolute value at such low densities may carry increased uncertainty, but the rarefaction itself is robust). Each of the three CMEs appears to correspond to an enhancement or continuation of the type IV burst.

We present additional observations and context for the long-duration type IV radio burst event. Figure 3(a) shows the Earth-view GOES-18 soft X-ray flux (1–8 Å and 0.5–4 Å) for context. Several moderate enhancements are visible near CME2 and CME3. Because CME1 originated on the farside of the Sun, any associated soft X-ray emission may be poorly represented in GOES; complementary farside diagnostics (e.g., Solar Orbiter/STIX) are beyond the scope of this work. Figures 3(b)–(e) compile the multispacecraft dynamic spectra and mark Windows 1–3 (magenta arrows) discussed in Section 2.2, together with flare/CME and in situ context. A pronounced type III storm is visible around CME2 (2025 August 30; orange arrow in Figures 3(c),(d)), together with complex type III activity (J. Zhang et al. 2024; M. Pulupa et al. 2025). This storm precedes the Wind/Parker Solar Probe type IV visibility Window 2 (magenta arrows) rather than immediately preceding the type IV onset. Such storm-like activity implies sustained access of the source region to open field lines and ongoing reconnection; its timing near CME2 is consistent with CME-driven magnetic field reconfiguration that intermittently modifies open field connectivity and hence the type III burst rate.

Figures 3(f)–(h) show in situ solar wind measurements from Wind (green curves), Solar Orbiter (magenta; where available), and STEREO-A (red curves) over the course of the event. Figure 3(f) gives the proton number density (n_p), Figure 3(g) the bulk solar wind speed (V_SW), and Figure 3(h) the total magnetic field strength (∣B∣). Several effects of the fast CMEs are evident in the near-Earth data (Wind, green).

CME1, which erupted on 2025 August 21, originated from a farside active region; it was associated with a hectometric type II radio burst (consistent with a CME-driven shock).

CME2, launched on 2025 August 30, drove a global shock that not only produced a type II burst but was also detected in situ: at Wind around September 1–2, the proton density and magnetic field show abrupt jumps, and the solar wind speed surges, signatures of the interplanetary shock.

CME3, erupting on 2025 September 4, did not have an obvious type II radio signature, but its passage had a dramatic impact on the solar wind: Wind observed an extremely low-density plasma void (with n_p dropping to ∼0.1 cm⁻³) in the days after the eruption. This rarefied interval, evident in Figure 3(f), suggests that CME3 over-expanded and generated an exceptionally rarefied trailing wake, leading to a transient “disappearance” event in the solar wind density. Similar low-density anomalies have been observed in the wake of some fast CMEs. The STEREO-A in situ measurements (red curves) show corresponding solar wind conditions from its vantage point, though the CME impacts at STEREO-A were less pronounced. Notably, STEREO-A did not experience as deep a density drop after CME3, consistent with the possibility that the bulk of CME3’s rarefied wake was directed toward Earth’s longitude. Such rarefied wakes may also enhance hectometric radio visibility by reducing propagation blurring along the line of sight, as suggested for the 2002 May long-duration type IV event ( Appendix C).

These contextual data confirm the temporal relationship between the CME eruptions and the observed radio and in situ signatures: CME1 and CME2 coincided with intense radio bursts (types II and IV), while CME3’s aftermath is captured by the anomalous drop in solar wind density.

Figure 4 provides a potential field source surface (PFSS) context for the three representative epochs used in this work (M. D. Altschuler & G. Newkirk 1969; K. H. Schatten et al. 1969; J. Worden & J. Harvey 2000; P. Riley et al. 2006). The PFSS extrapolations were computed from ADAPT photospheric boundary maps assimilating NSO/GONG synoptic magnetograms (J. W. Harvey et al. 1996; C. N. Arge et al. 2010; K. S. Hickmann et al. 2015). The source-surface polarity inversion line (PIL) outlines the streamer belt/heliospheric current sheet footprint at R_ss = 2.5 R_⊙. The observing spacecraft samples a similar Carrington sector relative to this PIL as the Sun rotates, supporting the interpretation that the long-duration hectometric continuum is associated with a long-lived, corotating large-scale structure rooted near the streamer belt. Notably, the projected observer locations in all three windows fall on the B_r > 0 side of the source-surface PIL. Moreover, in each window, the nearest segment of the source-surface PIL lies roughly ∼90° eastward of the projected sub-spacecraft longitude, placing the streamer belt near the corresponding east limb in each spacecraft’s view; this is consistent with the PFSS renderings (right panels of Figure 4) and with the streamer-like white-light context during Window 3, discussed later in Section 2.5. If the hectometric ray path remains predominantly in this magnetic sector out to 1 au, the observed persistent left-hand circular polarization at STEREO-A is consistent with ordinary-mode (O-mode) emission (Section 2.3). We emphasize that PFSS is used here only for global topological context; above R_ss, the model field is forced radial and therefore cannot be used to infer detailed magnetic geometry or mode-coupling effects at the hectometric source heights.

STEREO-A provides the most complete diagnostics during the brightest phase of the event. The geometry used below to relate the rotation duration and direction-finding angles to the physical source size and the WCRS correction is summarized in Figure 5. Figure 6 shows the time series at three representative frequencies (975, 1475, and 1925 kHz). The top row presents Stokes I (sfu) with a smoothed trend; the middle row shows the circular polarization V/I; the bottom row shows wavevector azimuth. We select high-confidence measurements with I > 5 × 10³ sfu and V/I < −0.8 to avoid weak, noisy intervals.

We estimate the source’s transverse size from the emission’s duration at STEREO-A combined with the Sun’s rotation. Durations after thresholding are T = {34,  35,  32} hr for 975, 1475, and 1925 kHz. These durations correspond to the high-confidence, strongly polarized portion of Window 3 (Figure 6); using the full Window 3 visibility duration (56.5 hr; Section 2.2) would yield a larger effective width and is best treated as an upper limit because detectability can also be shaped by beaming and propagation. With synodic rotation Ω_⊙ = 132 day⁻¹ and modeled radii r = {9.7, 7.3,  6.2} R_⊙ (E. C. Sittler & M. Guhathakurta 1999), the rotation-to-size conversion is

yielding W_rot,975 = (3.15 ± 0.95) R_⊙, W_rot,1475 = (2.44 ± 0.73) R_⊙, and W_rot,1925 = (1.90 ± 0.57) R_⊙ (1σ; dominated by a 30% height-mapping systematic). The band-averaged size is 〈W_rot〉 = (2.50 ± 0.75) R_⊙. The rotation-duration geometry used to convert T into a transverse size W_rot is sketched in Figure 5(a). This is the chord length subtending the rotation angle Δϕ = Ω_⊙T/24 at radius r, hence (and W_rot ≈ r Δϕ for Δϕ ≪ 1 in radians).

We use W for the physical full transverse diameter of the emitting trap (half-width a ≡ W/2). Two independent estimators of the same W are used: W_rot from rotation geometry (Equation 1) and from quasiperiodic pulsations (Section 2.4). When comparing to models that write the period as P ∝ L/v_A, L denotes a characteristic length scale of the oscillating structure; the appropriate geometric factor depends on the specific eigenmode (see Section 2.4, Equation 3).

We denote the goniopolarimetric apparent half-width at the spacecraft by γ and the rotation-based geometric half-width by γ_rot. Here, B ≡ γ/γ_rot denotes the angular-broadening factor and is distinct from the magnetic field model B_cor(r) used later in the Alfvén speed estimate (Figure 5(b)). At D_SC = 207 R_⊙ (the Sun–spacecraft distance during Window 3), the above W_rot correspond to γ_rot = {044 ± 013, 034 ± 010, 026 ± 008}. Here, γ_rot is the geometric angular half-width subtended by the corotation chord at the spacecraft (i.e., for small angles, γ_rot ≈ W_rot/(2D_SC)). With γ ≃ 20° in all three bands, the apparent-to-geometric factors are ; , 59 ± 18, and 76 ± 23 for 975, 1475, and 1925 kHz, respectively; combining the bands we adopt (1σ; systematic dominated; Figure 5(b)).

For context, the modeled source heights span Δr ≈ 3.5 R_⊙ across the band, comparable to the band-averaged transverse diameter 〈W_rot〉 = (2.50 ± 0.75) R_⊙. These radii are obtained from the E. C. Sittler & M. Guhathakurta (1999) density model under the fundamental plasma emission assumption used for the numerical estimates in this work. We defer the physical interpretation of both Δr and the large broadening factor to Sections 3.2 and 3.5. This comparable radial and transverse extent suggests a large, streamer-top/flux-rope scale trap (i.e., a large closed coronal structure near the helmet streamer cusp and/or a CME flux-rope cavity) whose transverse and radial dimensions are of the same order at these heights.

The very large points to strong angular broadening—plausibly multipath scattering by solar wind density inhomogeneities—while the near-constant γ across 1–2 MHz hints that propagation effects, rather than intrinsic source expansion, dominate the observed apparent size in this instance.

Direction-finding products are not available for Solar Orbiter and Parker Solar Probe, and we do not use Wind/WAVES direction finding because the available calibrated RAD1 direction-finding product assumes an unpolarized source (Section 2.1). However, Wind observed a comparably long type IV interval (over 2 days), suggesting a similar source size was present from Earth’s viewpoint. In contrast, Solar Orbiter saw only a much shorter continuum burst.

Two broad intensity maxima are evident at each frequency (vertical markers P1 and P2). For 975 kHz, P1 and P2 occur near 2025 September 7 21:36 and September 8 10:59; for 1475 kHz at 21:15 and 09:09; and for 1925 kHz at 21:12 and 08:39. From the offsets between the 1.925 and 0.975 MHz peak times, the P1 envelope implies an apparent pattern speed of ∼1600 km s⁻¹ across the modeled height range—comparable to the CME speeds listed in Figure 2(a) and plausibly reflecting the passage of the CME-driven disturbance across the relevant density shells (E. C. Sittler & M. Guhathakurta1999). By contrast, the P2 offsets imply ∼280 km s⁻¹, suggestive of a slower reconfiguration (e.g., expansion of the post-CME cavity or sheath, rotation-parallax-modulated visibility, or a low-phase-speed MHD disturbance within the trap); we therefore treat this second value as indicative rather than a direct measure of a propagating front. The uncertainty is dominated by peak-time selection and the adopted density–height mapping (E. C. Sittler & M. Guhathakurta 1999).

For the STEREO/WAVES polarization product used below, negative V/I denotes left-hand circular polarization in the adopted Level-3 sign convention. Any magnetoionic mode inference in this Letter is therefore referenced to that convention and should be regarded as conditional on it. Stokes V remains negative throughout, with ∣V/I∣  ≳  0.9 near the peaks, indicating nearly pure left-hand circular polarization in the STEREO/WAVES sign convention. In the magnetoionic framework, such near-unity V/I suggests emission dominated by a single mode; for plasma emission at (or near) the fundamental frequency, the escaping mode is often expected to be the ordinary (O) mode (D. B. Melrose 1980; D. J. McLean & N. R. Labrum 1985). The PFSS sector context (Figure 4) indicates that the relevant lines of sight lie on the B_r > 0 side of the source-surface PIL, suggesting that the observed left-hand sense could be compatible with O-mode emission under standard quasi-longitudinal assumptions. This mode inference, however, depends on the adopted Stokes-V convention and could be modified by mode coupling (e.g., quasi-transverse layers) or heliospheric current sheet/sector crossings along the ray path.

We measure the azimuth ϕ(t) of the incoming wavevector in the spacecraft RTN frame. Even if the source corotates at fixed Carrington longitude, a finite heliocentric distance r produces a slow parallax sweep of ϕ(t) as the Sun turns; only r  →  0 would keep ϕ constant. This rotation-parallax effect has long been used to infer interplanetary source distances from the drift of the apparent direction (J.-L. Bougeret et al. 1984).

Between P1 and P2 the azimuth drifts steadily (negative denotes west-to-east motion): −4.63, −2.44, and −1.01 deg day⁻¹ at 975, 1475, and 1925 kHz, respectively (Figures 6(g)–(i)). Under the corotation hypothesis, this drift provides a purely geometric constraint on the apparent rotation radius,

where D_SC is the Sun–spacecraft distance and Ω_⊙ is the synodic rotation rate as seen from D_SC (we adopt 13.2 deg day⁻¹ at 1 au). Using D_SC  ≈  207 R_⊙ yields R_app  ≈  53.8, 32.3, and 14.7 R_⊙ at 975, 1475, and 1925 kHz, respectively. Comparing these values with density model heights R_model(f) (Section 2.4) gives Ξ_F ≡ R_app/R_model  ≈  5.55, 4.42, 2.38. R_app is the effective heliocentric distance of a strictly corotating point source that would reproduce the measured azimuth sweep; in practice R_app  ≳  R_model because scattering/refraction displaces the apparent arrival direction away from the true source. The increase of Ξ_F toward lower frequency reflects stronger scattering at lower f_pe, and motivates the frequency-dependent correction adopted in our WCRS localization.

To isolate quasiperiodic pulsations, we remove slow trends from Stokes I by subtracting a 256 minute running median, yielding residuals ΔI(t) (Figures 7(a),(c),(e)). The quasiperiodic pulsation periodograms are computed over the same time intervals shown in Figures 7(a)–(e), selected by the high-confidence thresholds I > 5 × 10³ sfu and V/I < −0.8 (Section 2.3). The P1/P2 times marked in Figure 6 are retained only as visual guides to the broad envelope and are not used as inputs to the periodogram analysis. Lomb–Scargle periodograms (J. D. Scargle 1982) over 5–60 minutes show the highest-power peaks at 57.1 minutes (975 kHz), 50.3 minutes (1475 kHz), and 44.8 minutes (1925 kHz) (Figures 7(b), (d), (f)); the modulation persists for multiple cycles (at least 3–4) over this high-confidence interval. Given potential red noise (i.e., excess low-frequency/background power that can bias periodograms at long periods), we treat these as characteristic and use them to derive a lower bound on the trap size (Equation 4).

We interpret the quasiperiodic pulsations as MHD eigenmodes of a long-lived magnetic trap (H. Rosenberg 1970; V. V. Zaitsev & A. V. Stepanov 1975; V. M. Nakariakov & V. F. Melnikov 2009; I. V. Zimovets et al. 2021). For a cylindrical trap supporting the fundamental fast sausage mode (azimuthal number m = 0), the longest allowed period occurs near the long-wavelength cutoff of the sausage dispersion relation and equals

with j_0,1  ≈  2.4048 the first zero of J₀ (P. M. Edwin & B. Roberts 1983; B. Roberts et al. 1984; V. M. Nakariakov et al. 2003; M. J. Aschwanden et al. 2004). Inverting Equation (3) for a measured P yields a conservative lower bound on the transverse size of a sausage-mode trap,

We evaluate v_A(r) at the band-specific heights using the E. C. Sittler & M. Guhathakurta (1999) density model and a radial field of B(r) ∝ r⁻² normalized to 6 nT at 1 au. The in situ n_p and ∣B∣ at 1 au in Figures 3((f)–(h)) provide heliospheric context but are not a direct validation of the coronal v_A(r) used here, which depends on the adopted coronal density model and field scaling.

Using Equation (4) with the measured periods (57.1, 50.3, and 44.8 minutes at 975, 1475, and 1925 kHz), we obtain , , and , and an ensemble mean (1σ; uncertainty dominated by field/density systematics). These lower bounds are in good agreement with the rotation-geometry diameter W_rot = (2.50 ± 0.75) R_⊙ (Section 2.3), supporting a trap of streamer-top/flux-rope scale (Figure 7(g)).

We note that global trapped sausage modes require loops that are sufficiently short and dense to exceed the long-wavelength cutoff; in long, slender streamer-top structures, the fundamental can be leaky or suppressed, in which case Equation (4) remains a useful lower bound, and the observed oscillation may instead correspond to a higher harmonic or a localized portion of the structure rather than the global fundamental (V. M. Nakariakov et al. 2003; M. J. Aschwanden et al. 2004). The persistence of these oscillations over several cycles suggests they may be continuously driven or reinforced—for example, by energetic particles bouncing within the trap in resonance with the MHD mode (I. V. Zimovets et al. 2021).

Localizing low-frequency sources from one vantage point is challenging because refraction and scattering bend ray paths and because direction-finding solutions are strongly affected by angular broadening. We therefore introduce the WCRS method, which combines (i) the empirically determined scattering factors Ξ_F(f) from the azimuth sweep (Section 2.3), (ii) an assumed coronal density model to map frequency to radius, and (iii) simple ray-sphere geometry. For readability, we move the full WCRS derivation (coordinate transforms and the ray-sphere solution) to Appendix B; here, we summarize the procedure and the resulting source locations.

Briefly, we correct the STEREO/WAVES direction-finding angles by shrinking their deviation from the Sun–spacecraft line according to Ξ_F(f), and intersect the resulting back-traced line of sight with the density model shell R_model(f) (Appendix B). This wavevector correction is intended to undo the dominant displacement of the apparent arrival direction caused by wave propagation (scattering/refraction) and thereby recover a physically interpretable trajectory. The resulting close-side solutions for 975, 1475, and 1925 kHz are shown in Figures 2(b)–(d).

We do not show the farside intersection branch: in this event, it would place the source in the sector of CME1, which erupted about 2 weeks before the STEREO-A type IV window; solar rotation would carry such a structure far from the observed longitudes, making that association implausible. We therefore interpret the close-side solution as the physically plausible branch for this event.

The point-to-point trajectories at different frequencies are not strictly continuous in time because each location is derived independently from single-spacecraft direction finding at discrete frequencies and times, and residual propagation effects and density model systematics introduce scatter; nevertheless, the ensemble consistently traces the streamer belt.

For the white-light context in Figure 8 we show only the wavevector-correction step: we apply the angular shrinking of V. Krupar et al. (2024a) to ϕ^RTN and θ^RTN (Equations (B1)–(B2)) and project the resulting corrected arrival directions onto the SECCHI frames. This provides a qualitative check of the association with the helmet streamer; by itself, it does not determine the heliocentric distance (set by R_model(f)) or discriminate between the close- and farside intersections along a given line of sight. Small apparent reorderings of the three frequencies in Figure 8 should therefore not be interpreted as true height inversions. They reflect sky-plane projection plus residual propagation scatter within a very broad apparent source (γ ∼ 20°; V. Krupar et al. 2012), so the offset between the plotted centroids is much smaller than the propagation-broadened source extent. The WCRS solutions used for the physical interpretation retain the expected ordering, with the 975 kHz source farther from the Sun than the 1475 and 1925 kHz sources.

Download figure:

Video Standard image High-resolution image

We note that (i) the factors Ξ_F (and Ξ_H, if harmonics are considered) are taken from the previous subsection for this event; constant fallbacks Ξ_F  ≈  3.72 and Ξ_H  ≈  2.10 over 350 kHz–1 MHz are supported by the STEREO survey (V. Krupar et al. 2024a). (ii) All trigonometric functions above take degree arguments; the data products are in degrees throughout. (iii) The WCRS localization used for Figure 2 uses both corrected angles and yields a 3D intersection point on ∣r∣ = R_model(f) (Equations (B6)–(B11)); Figure 2 shows the (x, y) projection of the close-side solution, whereas Figure 8 only illustrates the corrected direction of arrival on the plane of the sky.

The observations reveal a hectometric type IV continuum observed over ∼19 days in three successive visibility windows and consistent with corotation with the Sun, becoming visible in sequence to Solar Orbiter, Wind, and STEREO-A as the viewing geometry turns favorable. The strong, nearly pure left-hand circular polarization (∣V/I∣  ≳ 0.9) indicates a single dominant magnetoionic mode in a coherent magnetic environment. This is consistent with coherent emission escaping predominantly in a single magnetoionic mode. Fundamental plasma emission is a natural working assumption (D. J. McLean & N. R. Labrum 1985), but an ECME contribution in an unusually low-density cavity (potentially involving Z-mode waves and mode conversion) also remains plausible (R. M. Winglee & G. A. Dulk 1986a, 1986b; T. Formánek et al. 2025). WCRS localizes the source near the helmet streamer belt at physical heights of order 6–10 R_⊙ after propagation is accounted for, a region where large flux ropes and streamer arcades can confine electrons over long intervals. As an order-of-magnitude check, if the emission is plasma radiation at 0.5–3 MHz, the implied densities are n_e ∼ 3 × 10³–10⁵ cm⁻³. Standard Coulomb slowing-down and pitch-angle deflection estimates for ∼20–50 keV electrons then give characteristic collisional times ranging from days at the dense end to months at the tenuous end (J. C. Brown 1972; A. G. Emslie 1978). Coulomb collisions alone therefore do not obviously preclude multiday confinement, although continued replenishment and escape/transport losses may still be important. We do not infer that the reservoir was already fully established at CME1; rather, CME1–CME3 are treated as plausible contributors to sustaining or replenishing the trapped population over the full 19 day interval.

Our WCRS height inference uses a plasma-frequency mapping (fundamental/harmonic); if the emission instead tracked the cyclotron frequency, the absolute heights would shift, but the central observational requirements—corotation, a limited visibility window, and a long-lived reservoir repeatedly replenished by CMEs—remain unchanged. Operationally, the same procedure provides (i) shock height–time profiles from type II bursts suitable for CME nowcasting, and (ii) maps of open field connectivity from type III bursts, both from a single vantage point in near real time.

Two independent size diagnostics are mutually consistent: the rotation-based diameter W_rot (Section 2.3) and the magnetoseismic lower bound (Section 2.5). We therefore adopt a characteristic trap diameter of W ≈ 2.5–3.0 R_⊙ (half-width of a ≈ 1.3–1.5 R_⊙), with uncertainties dominated by the density model and B(r) assumptions.

Three fast CMEs originated from the same solar sector during the 19 day span of the event. Their timing relative to the continuum suggests that each CME could contribute to sustaining the reservoir by reorganizing the large-scale magnetic environment and either (i) injecting/reaccelerating electrons within the evolving CME/streamer system, and/or (ii) modulating magnetic connectivity so that the source region at the base of the large corotating loop continues to supply electrons over many days. This interpretation aligns with prior associations of type IV emission with CMEs and shocks (M. Pick & N. Vilmer 2008; R. Miteva et al. 2017; A. Mohan et al. 2024). The pronounced type III storm activity near CME2 (Figures 3(c),(d)) further supports sustained electron release from the same sector. Distinguishing between replenishment of a trapped population and quasi-continuous injection from the low corona will require dedicated modeling and/or energetic electron measurements, and is left for future work.

The 19 day corotating continuum reported here echoes the ∼6 days, nearly pure LHC hectometric type IV of 2002 May 17–23 (M. J. Reiner et al. 2006). In both cases, fast/wide CMEs preceded or accompanied the continuum, and the peak phase featured hour-scale quasiperiodic pulsations consistent with standing MHD eigen-oscillations in a large coronal resonant cavity (the electron trap). Importantly, in both epochs the continuum was followed at Earth by an extremely low-density solar wind interval—for 2025 after CME 3 (Figure 3(f)), and for 2002 within ∼1–2 days after the radio maximum ( Appendix C). In the 2002 case, the deepest density dropout occurs after the main ICME signatures (enhanced ∣B∣ and V_SW; Figure 9), suggesting a trailing rarefaction/wake rather than the ICME body itself. This repeated ordering suggests that the radio reservoir and the subsequent “disappearing wind” interval may be linked to an over-expanded ICME cavity together with its rarefied wake: the cavity/streamer-top structure can confine and reenergize electrons as it corotates, while the wake reduces scattering/absorption and enhances detectability. The large apparent-to-true size ratio () we measure here further implies that such events are most conspicuous when propagation through the rarefied wake is unusually transparent; under typical, more turbulent conditions, scatter broadening and depolarization may render analogous reservoirs harder to detect from a single vantage point.

A key result is the large apparent-to-geometric broadening factor of measured by STEREO-A, with across 1–2 MHz (Section 2.3). We denote the apparent half-width (angular radius) of the source image at the spacecraft by γ (from goniopolarimetric size measurements; V. Krupar et al. 2012), whereas γ_rot is the geometric half-width inferred from the rotation-based size. The very large demonstrates that the apparent extent is dominated by propagation—multipath scattering by solar wind density inhomogeneities—rather than by the intrinsic source size. The near-constancy of γ with frequency, despite changes in modeled source height, further supports a scattering-controlled point-spread function at these frequencies.

The two scattering diagnostics in this work, Ξ_F(f) from direction-of-arrival deflection and from broadening, are physically distinct but complementary. Ξ_F quantifies how far the incoming wavevector is tilted from the true source direction, which must be corrected to recover location and height. quantifies how much the apparent image is smeared relative to the true source size. Taken together, the large Ξ_F at lower frequencies and the event-averaged explain why the measured angular sizes (γ ∼ 20°–40°) greatly exceed the physical trap diameter (W ∼ 2.5–3.0 R_⊙): the former are propagation-broadened images, whereas the latter reflect the intrinsic geometry constrained by rotation and magnetoseismology.

When an independent geometric width is unavailable (as in many type II/III cases), one may use an empirical (or the event average of ) as a propagation debroadening factor. If γ is the measured goniopolarimetric half-width at the spacecraft, an estimate of the intrinsic transverse diameter is

where D_SC is the Sun–spacecraft distance and R_model(f) is the modeled heliocentric height of the source at frequency f. Thus, provides a simple, instrument-agnostic correction from apparent to physical source size for type II/III bursts at hectometric frequencies where scattering conditions are comparable.

Our analysis has some limitations. It relies on single-spacecraft direction-finding and an empirical scattering correction. The absolute source heights inherit uncertainties from the coronal density model (and our assumed B(r) ∝ r⁻² scaling). We also assumed a constant Ξ_F(f) over the multiday interval, and any uncorrected polarization calibration offsets could introduce additional systematic error. Nevertheless, multiple independent observations—the multispacecraft visibility sequence, the strong circular polarization in one mode, the agreement between rotation-based and magnetoseismic sizes, and the consistency of WCRS-derived positions with helmet streamer geometry—all converge to the same interpretation: a long-lived coronal electron reservoir.

Methodologically, our WCRS approach enables single-spacecraft radio source localization today, and it will synergize with radio imaging by upcoming missions like SunRISE (J. Lazio et al. 2017). By combining WCRS with imaging data and in situ particle measurements, future studies can more precisely quantify how CMEs seed and maintain such heliospheric electron reservoirs, improving both our physical understanding and space-weather monitoring capabilities.

The observations and analysis link, to our knowledge, the longest reported hectometric type IV continuum to a corotating electron reservoir, and show how frequency-aware propagation corrections can yield a physically interpretable single-spacecraft localization estimate.

Event-specific observational/physical conclusions:

1. 1.  

   Record-long hectometric type IV. To our knowledge, we observed the longest reported hectometric type IV continuum, spanning 0.5–3 MHz across three successive visibility windows over nearly 19 days at four spacecraft spread across 220° longitudinally.

2. 2.  

   Sustained reservoir. Three fast CMEs from the same solar sector plausibly helped maintain the reservoir by restructuring the large-scale trap and enabling continued electron injection from the low corona.

3. 3.  

   Recurring pattern across cycles. A possible recurring pattern across solar cycles may exist: the 2025 event shares strong polarization, hour-scale quasiperiodic pulsations, and a subsequent rarefied solar wind interval with the 2002 May 17–23 event (M. J. Reiner et al. 2006). This suggests (but does not prove) that over-expanded CME/ICME cavities and their rarefied wakes could both confine/reenergize electrons and reduce propagation blurring, enhancing hectometric type IV visibility; testing this hypothesis will require a larger event sample.

Methodological contribution (WCRS) and scope/implications:

1. 1.  

   Single-spacecraft localization. Using a frequency-dependent scattering correction (WCRS), a single spacecraft can provide an event-specific localization estimate for low-frequency sources. In this event, WCRS places the source near a helmet streamer at a physical height of ∼6–10 R_⊙.

2. 2.  

   Intrinsic size from two diagnostics. Solar rotation geometry and sausage-mode magnetoseismology (P = (π/j_0,1)W/v_A) are consistent with an intrinsic transverse width of only a few solar radii, W ≈ 2.5–3.0 R_⊙ (with the seismology providing a lower bound).

3. 3.  

   Propagation-dominated image. The apparent goniopolarimetric half-width is much larger than the inferred intrinsic width (B ∼ 60 ± 18), showing that interplanetary scattering strongly affects the observed source extent in the 1–2 MHz band.

4. 4.  

   Outlook. WCRS and MHD seismology provide powerful single-spacecraft diagnostics. Coupling these with forthcoming low-frequency imaging (e.g., SunRISE) and in situ particle data will sharpen constraints on source geometry and energetics, including shock height–time profiles from type II lanes and connectivity maps from type III ensembles, enhancing our space-weather forecasting toolkit.

RPW acquires electric spectra with the TNR (∼4 kHz–1 MHz) and the HFR (0.375–16 MHz) operating in dipole mode using the V1, V2, and V3 booms (M. Maksimovic et al. 2020). Narrowband spacecraft disturbances—most prominently at ∼80 kHz (reaction wheels) and ∼120 kHz (PCDU) with harmonics—contaminate parts of the band; therefore, all flux conversions are restricted to the vetted clean channel lists compiled for Solar Orbiter operations (M. Maksimovic et al. 2021).

Our calibration strategy deliberately builds on and extends the approach of A. Vecchio et al. (2021): that study established an in-flight cross-calibration from only three relatively weak type III bursts, using Wind/WAVES RAD1 flux as the reference and deriving reduced effective lengths ΓL_eff. Here, we adopt STEREO-A/WAVES (J. L. Bougeret et al. 2008; V. Krupar et al. 2022) as the absolute flux etalon and enlarge the calibration set to 12 near-alignment type III intervals between 2022 November 22 and 2022 December 3. For each clean frequency, we solve the short-dipole inversion frequency by frequency so that the 1 au renormalized RPW flux matches STEREO-A. Combining all events yields more stable, higher-fidelity ΓL_eff(f) estimates across the TNR/HFR bands (including the mild high-frequency roll-off), with uncertainties taken as the event-to-event scatter. The adopted ΓL_eff(f) values used throughout this Letter are summarized in Table 1.

Table 1. Solar Orbiter/RPW Dipole Gains Used in This Work: ΓL_eff (m)

RPW Dipole	ΓLeff for f ≤ 2000 kHz	ΓLeff for f  >  2000 kHz
V1–V2	2.903 ± 0.184
V2–V3	2.410 ± 0.186
V3–V1	2.925 ± 0.175

Note. The left column applies below 2 MHz; the right column gives the frequency dependence above 2 MHz (f in kHz).

Download table as:  ASCIITypeset image

Table 2. Wind/WAVES Reduced Effective Lengths ΓL_eff (m) from STEREO-A Cross-calibration of 19 Type III Bursts (2023 August)

Channel	RAD1: 20–1040 kHz	RAD2: f ≤ 2500 kHz	RAD2: f  >  2500 kHz (with f in kHz)
S	30.510 ± 1.526	6.272 ± 0.314
SP	30.620 ± 1.531	6.529 ± 0.326
Z	3.163 ± 0.158	2.706 ± 0.135

Download table as:  ASCIITypeset image

Wind/WAVES acquires radio autospectra with two swept receivers (J.-L. Bougeret et al. 1995), RAD1 (20–1040 kHz) and RAD2 (1.075–13 MHz), using three orthogonal electric sensors (X, Y, Z). A front-end switching/matrix selects S, SP, and Z channels with different antenna couplings in the two bands: in RAD1, the S/SP chains sample the spin-plane path (SEP mode; ≈X), whereas in RAD2, the S/SP chains sample a summed path (SUM mode; ≈X + Z). SP is the 90° quadrature replica of S for goniopolarimetry; Z is the spin-axis monopole in both bands (J.-L. Bougeret et al. 1995). These mode differences (SEP versus SUM) and their distinct impedance/matching paths lead to different reduced effective lengths ΓL_eff for RAD1 and RAD2, even for the same labeled channel.

As for Solar Orbiter, we use STEREO-A/WAVES as the absolute flux etalon. We selected 19 type III bursts occurring during 2023 August 2 to 2023 August 20 with longitudinal separations of <1°. For each event, we distance normalize both instruments by (1 au/r)², and determine ΓL_eff(f) so that the WAVES flux reproduced via Equations (A1) and (A2) matches the STEREO-A flux at the nearest channel. We use Z for the science analysis in Figures 1 and 3 (least affected by spin/mode artifacts); S/SP are reported here for completeness (Table 2).