- The paper reconstructs a 2024 CME–CME interaction in which a faster CME traveling at 523 km s−1 disrupted a slower 169 km s−1 CME, producing downflows identified as its remnant material.
- Multi-instrument observations show the disrupted plasma descending from several solar radii into the low corona at approximately 30 km s−1, cooling progressively from 211 Å to 171 Å emissions and potentially triggering coronal rain.
- Drag modeling supports release heights of 2.5–5.5 solar radii and constrains the radial drag parameters to β≈1.5–2.1 and k₀≈100–280, while VLA Faraday rotation independently confirms the disrupting CME’s flux-rope structure.
Overview
This paper presents a multi-instrument empirical study of an unusual solar eruptive event on 2024 August 16–17, in which a slow coronal mass ejection (CME) is disrupted by a faster following CME, and part of the disrupted material is observed falling back into the low corona. Combining white-light coronagraphy from SOHO/LASCO and STEREO-A, EUV imaging from SDO/AIA and STEREO-A/EUVI, and radio Faraday rotation (RFR) measurements from the Karl G. Jansky Very Large Array (VLA), the authors reconstruct the 3-D kinematics and morphology of four near-simultaneous CMEs, characterize the fate of the disrupted CME, and model the descent of its remnant plasma. The central observational claim—that remnant material from a failed CME can be tracked in EUV all the way back to the low corona from heights of several solar radii—is, according to the authors, without precedent.
The eruptive sequence and the disruption of CME2
Four CMEs (CME1–CME4) erupt between UT 13:30 and 15:00 on 2024 August 16, observed from SOHO at L1 and STEREO-A, which was only 22.6∘ ahead of Earth. The pivotal interaction occurs when CME4, a large but exceptionally faint eruption moving at a constant V=523 km s−1, overtakes the slower CME2 (V=169 km s−1 average leading-edge speed) at roughly UT 18:00. The southern leg of CME4 tears CME2 apart: the northern half of CME2 is carried outward within that leg, rendering it conspicuously bright, while the southern half simply disappears from coronagraphic images and is replaced by a dense swarm of small jet-like downflows.
A notable morphological result is that CME4 provides an unusually clear view of a magnetic flux rope (MFR) leg in white-light images—the leg remains visible long after the leading edge exits the LASCO/C2 field of view. The authors note that this brightness pattern is inverted relative to typical CMEs (bright trailing structure rather than bright leading edge), which explains why the CDAW catalog misidentifies the leg as a separate eruption. The event therefore offers a rare opportunity to identify MFR legs directly in coronagraph data, a task that is normally very difficult.
Kinematic modeling shows CME1 undergoing unusually strong deceleration (from roughly 700 km s−1 to 398 km s−1), while CME3 accelerates modestly to 357 km s−1. A full 3-D forward-modeling reconstruction assuming parametrized MFR shapes yields trajectory longitudes and latitudes for all four events; all four trajectories and orientations mirror the heliospheric current sheet (HCS), whose strongly warped, locally north-south orientation at this time is shown via a PFSS model. The authors interpret the eruptions as symptoms of instability in this segment of the HCS, consistent with prior work on CME deflection toward the HCS magnetic-pressure minimum. This HCS geometry also explains the pervasive pre-existing downflow activity on both limbs: a face-on streamer belt increases contrast and spreads the narrow downflows over wider position angles.
Triangulation of the downflows
Because the small downflows following CME2's disappearance are compact features, they can be located by direct triangulation between COR2-A and LASCO/C2 image pairs rather than by forward modeling. Three prominent downflows tracked over several hours have inferred HEE coordinates closely matching the reconstructed central trajectory of CME2, (λs,βs)=(103∘,−40∘), independently confirming that the downflows consist of CME2 remnant material. There is evidence of non-radial motion—for example, one downflow's longitude shifts from approximately 105∘ to V=5230 between 6.0 and 4.5 RV=5231—suggesting the flows follow non-radial field lines funneling material sunward.
Radio Faraday rotation constraints on CME4
Between UT 16:00 and 22:00, the VLA monitored eight polarized background radio sources, cycling through pointings about every 30 minutes. Two lines of sight intercept CME3 and CME4; this paper analyzes the CME4 sight line toward a double-lobed radio galaxy with lobes separated by V=5232. The measured rotation measures show two distinct signatures: first, a positive-to-negative RM transition (peaking near V=5233 rad mV=5234) as the line of sight crosses the top of the MFR, and later a much larger negative shift as the sight line grazes the dense southern leg enriched with CME2 material—roughly a factor of 4 larger than the apex signature, implying leg densities about four times higher.
The sign reversal is the key diagnostic: in the MFR paradigm, the wrapping of azimuthal field around the flux rope axis naturally produces opposite RM signs on the leading and trailing sides of the axis crossing. The authors insert a physically motivated field model (based on the Nieves-Chinchilla et al. prescription) into their 3-D CME4 reconstruction, leaving only two free parameters: axial field V=5235 and peak azimuthal field V=5236, both referenced to 1 au under self-similar expansion. Values of V=5237 nT and V=5238 nT, combined with an internal density of V=5239 cm−10 at the encounter distance of 9.4 R−11, reproduce the observed positive-negative signature and its roughly one-hour duration well, though only after an arbitrary 45-minute time shift attributed to uncertainties in the faint leading edge reconstruction. The implied 1 au density of about 9.5 cm−12 is consistent with typical in situ CME densities.
The authors are appropriately candid about non-uniqueness: scaling −13, −14, and −15 together leaves RM unchanged, and reversing the overall MFR polarity (northern leg negative, left-handed chirality) also yields a positive-negative signature, particularly if inner-edge densities are enhanced. They further concede that the assumed tilt angle of CME4 is not constrained precisely enough to rule out the alternative polarity firmly. No attempt is made to model the leg signature itself, since the reconstruction cannot place the line of sight inside the leg—a geometric limitation of single-valued MFR edge prescriptions.
Low-coronal EUV signatures of the returning plasma
The most distinctive result concerns SDO/AIA observations of the low corona beginning roughly 6–17 hours after CME2's disruption. A broad, faint downflow stretches across much of the southwest quadrant above a loop arcade, visible successively in the 211 Å, 193 Å, and 171 Å bandpasses. Brightness measurements along −16 arcs at heliocentric distances of 1.20–1.35 R−17 show clear high-to-low height brightening sequences, most cleanly in 171 Å, yielding an estimated descent speed of −18 km s−19.
A pronounced temperature dependence emerges: the downflow signature appears first at V=1690 (211 Å) and last at V=1691 (171 Å), with the 304 Å response lagging the 211 Å brightening by about eight hours. The natural interpretation is progressive radiative cooling of the descending plasma. The 304 Å behavior is interpreted differently—not as cold material arriving from above, but as condensations triggered by the warm downflows, manifesting as coronal rain flowing down from the brightened loop arcade. This leads the authors to a suggestive hypothesis: since similar small-scale downflows are ubiquitous in coronagraphic images even absent major CMEs, downflows from the upper corona may contribute to coronal rain more generally than previously appreciated.
The 304 Å light curve additionally exhibits a statistically robust periodicity of 50 minutes (Lomb-Scargle analysis), comparable to periods reported for large prominence oscillations (e.g., the GONG HV=1692 filament survey mean of V=1693 minutes). However, the authors cannot visually identify any physical oscillation responsible, and the cause of the periodicity remains unresolved—an honest open question.
Kinematic drag modeling
The observed low-coronal speed of V=1694 km sV=1695 is far below the 618 km sV=1696 surface escape speed, ruling out ballistic descent and requiring drag. Adopting the quadratic drag formalism used previously for core fallback events, but generalizing the drag coefficient to a radial power law V=1697 with reference radius V=1698 RV=1699, the authors search for parameter combinations consistent with arrival at 1.25 R−10 within 6–17 hours at velocities between −11 and −12 km s−13, for launch heights of 2.5–5.5 R−14.
Two families satisfy these constraints: −15 and −16, reflecting a degeneracy between the power-law index and coefficient amplitude. The preferred low-end solution −17 yields arrivals at 6.4, 10.0, and 14.8 hours for material released from 3.5, 4.5, and 5.5 R−18 respectively, at −19 km s−10. The high-−11 family is disfavored because it predicts coronagraph-height downflow speeds near −12 km s−13 or faster, inconsistent with the −14 km s−15 measured in COR2-A. The acceptable range is thus −16–2.1 and −17–280, with stated preference for the lower end.
Limitations and open questions
Several caveats bear directly on the results. The 45-minute timing offset required in the RM fit reflects genuine uncertainty in the CME4 reconstruction, whose leading edge is extremely faint. The polarity and chirality of the CME4 flux rope cannot be uniquely determined from the RM data alone, and the tilt angle entering that determination is itself uncertain. The drag model parameters are degenerate and the quadratic drag form is phenomenological rather than derived from first principles; distinguishing among competing −18 combinations would require independent density or field measurements along the downflow paths. The 50-minute 304 Å periodicity lacks an identified physical mechanism. Finally, whether the ubiquitous narrow coronagraphic downflows routinely reach the low corona and seed coronal rain—as hypothesized here—remains untested beyond this single event.
Conclusion
This paper documents a complete observational chain linking a CME–CME interaction in the upper corona to measurable consequences in the low corona: stereoscopic reconstruction establishes the disruption geometry, direct triangulation ties the resulting downflows to the destroyed CME, VLA Faraday rotation confirms the flux rope structure of the disrupting CME and quantifies its fields, SDO/AIA captures the cooled remnant plasma descending at −19 km s−10 with a clear cooling sequence across EUV bandpasses, and drag models reconcile the slow descent with release heights up to 5.5 R−11. Beyond its specific findings, the event demonstrates that upper-coronal downflows can trigger coronal rain condensations, raising the question of how frequently such externally driven rain occurs relative to the classical quasi-static loop cooling scenario.