It is well known that explosive energy releases in the form of solar flares and coronal mass ejections (CMEs) are driven magnetically. In this process, magnetic energy is converted to the kinetic energy of particles and the heat and bulk motion of plasmas (Thompson et al., 2021). CMEs open up a portion of the solar corona, along which energetic particles are released into interplanetary space and can impact the space weather. At present, routine measurements of the full disk magnetic field are available only in the lower solar atmosphere, the photosphere. Therefore, the capability to identify the solar surface signatures of an open magnetic structure and track magnetic field evolution prior to eruption will help us explore mechanisms governing the onset of the eruption and predict space weather.

Traditionally, coronal holes (Cranmer, 2009), or regions of persistent lack of emission at soft X-ray (SXR) and extreme ultraviolet (EUV) wavelengths, are considered to map the feet of open field lines on the solar surface. The lifetimes of coronal holes range from many days to multiple weeks (Lowder et al., 2017, and references therein). The temporary, and often abrupt, opening up of the magnetic structure associated with a CME produces transient coronal holes, also called coronal dimmings (Sterling and Hudson, 1997). A comprehensive review of observational signatures and an interpretation of coronal dimmings has been provided by Veronig et al. (2024, in press). Mechanisms causing the explosive “opening up” of the corona include the onset of an ideal instability, in which the global force balance is lost during the quasi-equilibrium evolution of the system (Forbes and Isenberg, 1991; Török and Kliem, 2005; Kliem and Török, 2006; Isenberg and Forbes, 2007), or the onset of a non-ideal instability, often referred to as magnetic reconnection, which abruptly changes the connectivity of the field lines and, in this way, removes or weakens the constraints that would maintain the force balance of a coronal structure. The “tether-cutting” (Moore et al., 2001) and “break-out” (Antiochos et al., 1999) configurations have been the most well-known reconnection geometry, which occurs either below or above the erupting structure, be it a magnetic flux rope or a sheared arcade (Patsourakos et al., 2020).

In the case of major solar eruptive events, the onset of the eruption, either due to ideal or non-ideal instabilities, takes place in the corona, but perturbations propagate along magnetic field lines to reach the lower atmosphere on Alfvénic timescales. In this paper, we focus on dimming and brightening signatures observed at the foot or base of magnetic structures that are undergoing dynamic evolution, such as magnetic reconnection or plasma expansion/eruption. As a direct consequence of magnetic reconnection, energy flux, via particle beams (Fisher et al., 1985), thermal conduction (Longcope, 2014), or Alfvén waves (Fletcher and Hudson, 2008), is transported along newly reconnected field lines to be deposited in the denser lower atmosphere, producing enhanced brightening there. On the other hand, dimming at the base of the corona is primarily an effect of plasma rarefaction due to the expansion of the overlying coronal structures such as a CME, which reduces the pressure of the overlying corona. The expansion or eruption of a coronal structure may occur before or after the onset of magnetic reconnection; therefore, dimming may be observed either before or after brightening in the lower atmosphere.

A solar eruptive event involves a rather complex magnetic configuration, and different parts of the system undergo different dynamics and also interact with each other, such as through reconnection. The spatially resolved, full-disk observations, like those provided by the Solar Dynamics Observatory (SDO; Pesnell et al., 2012), allow us to infer the evolution of different parts of the coronal structures during or before the eruption by examining the behavior at their base in the lower atmosphere. In this paper, we conduct such an experiment on two eruptive flares, using the tempo-spatial sequence of brightening and dimming to reconstruct the evolution of coronal structures in the early phase of their eruption. In the following section, we describe the strategy to identify dimming signatures at the base of the corona (S2). We apply the analysis to observations of an X-class eruptive event SOL20120712, which shows post-eruption dimming (S3), and of a C-class eruptive event SOL20110621, which also exhibits pre-eruption dimming (S4), and use these signatures to infer the properties of the overlying coronal structure. We summarize what is learned from this experiment and discuss the potential to reconstruct the evolution of overlying coronal structures from the brightening and dimming signatures at their base (S5).

In this paper, we analyze the brightening and dimming signatures in the lower atmosphere of two eruptive events. They exhibit a variety of dimming–brightening sequences that can be used to diagnose the dynamic evolution of overlying coronal structures in the early phase of the eruption.

An X-class eruptive flare occurred on 12 July 2012 in NOAA-11520. The flare was accompanied by a fast CME with its early-phase motion best captured by STEREO-A (Cheng et al., 2014; Dudík et al., 2014; Zhu et al., 2020). Figure 2 shows several snapshots of the event observed by the AIA. The eruption occurred in the core active region, forming two bright ribbons in magnetic fields of opposite polarities, outlining the feet of post-reconnection flare loops or closed field lines produced by reconnection. We refer to these two ribbons as flare ribbons. Shortly afterward, another set of two long ribbons extended from the flare ribbons in the core region, first briefly brightening and then quickly dimming. We call these two ribbons that became dark dimming ribbons, and they were also located in magnetic fields of opposite polarities.

Figure 3A shows, in red, the time sequence of the brightening in the flare ribbons in the core region and also the brightening adjacent to the dimming ribbons away from the core region. This reconnection mask is derived using the UV 1,600 Å observations, with the method described by Qiu et al. (2007, 2010). In the same figure, blue indicates the time sequence of the dimming as the brightness is reduced to 80% of the base level (see S2 for identification of the dimming pixels). The dimming (and the preceding brightening) appears to spread rapidly along the dimming ribbons at an apparent speed of more than 100 km s ⁻¹, which was noted as a signature of “slipping reconnection” by Dudík et al. (2014).

The magnetic flux integrated in the brightening pixels and in the dimming pixels is measured and shown in Figure 3B, showing that, in this event, the onset of flare reconnection starts at approximately 16 UT, prior to the rapid increase in the dimming. The flare reconnection rate peaks at 16:20 UT and stops at approximately 17 UT, when a total amount of 7 × 1 0 21 Mx has been reconnected. The dimming flux is also measured, which increases quickly after 16:20 UT, and the dimming rate peaks at approximately 17 UT, with the total amount of dimming flux reaching 5 × 1 0 20 Mx encompassed in the dimming ribbons. The associated CME is observed by STEREO; it shows a rapid acceleration starting from 16:10 UT and reaching a maximum at 16:20 UT (Zhu et al., 2020), when the rate of flare reconnection peaks.

The dimming ribbons are most prominently observed in the EUV 304, 171, and 193 Å passbands. Figure 4 shows the time–distance diagrams of the normalized brightness produced along five slits across each dimming ribbon; the locations of the slits are indicated by horizontal or vertical bars in Figure 3A. In particular, the diagrams of the 304 Å passband clearly demonstrate dimming after impulsive brightening. Figure 5 shows the epoch plot of the pixel light curves in the two dimming ribbons, or the evolution of the brightness (normalized to the base brightness) with respect to the time of the peak brightness. The figure illustrates the timescales of impulsive brightening, followed by rapid dimming, both within a couple of minutes. In most of these places, dimming does not recover for more than a few hours.

The tempo–spatial sequence of the brightening and dimming in this event suggests the scenario of magnetic reconnection between an erupting structure from the core region and the overlying arcades or the strapping field. The two dimming ribbons outline the photospheric intersection of separatrices in the complex multi-polar magnetic field where reconnection tends to occur (Dudík et al., 2014). In particular, a potential field extrapolation suggests that the dimming ribbon in the negative magnetic field to the east outlines the feet of an arcade connecting the dimming ribbon to the outer edge of the flare ribbon in the positive magnetic field, and the dimming ribbon in the positive magnetic field to the west maps the feet of another set of arcades with their conjugate feet in the negative magnetic field north of the flare ribbon (Cooper Downs, personal communication). These overlying arcades have to open up for the underlying structure to escape the solar corona.

In summary, in this eruptive event, the dimming analysis does not reveal any significant pre-eruption dynamics of the coronal magnetic structures. Dimming occurs after the onset of flare reconnection in the core region. The pair of dimming ribbons extending outward from the flare ribbons most likely map the feet of several overlying arcades that open up by reconnecting with the erupting structure. In this course, an overlying flux of 5 × 1 0 20 Mx, which is 10% of the flare reconnection flux, is removed from the path of the eruption.

A C-class eruptive flare occurred on 21 June 2011. It has been studied by Zhu et al. (2020) and Vievering et al. (2023). The associated CME is best observed by STEREO-A from the limb during its early phase; following the trajectory of the CME, its onset is determined to be at 2:15 UT (Zhu et al., 2020). Figure 6 shows the evolution of the event observed in the EUV 304 Å (left) and 211 Å (right) passbands. A filament is visible prior to the eruption and is erupted around the time of the CME onset. During the eruption, two flare ribbons are brightened as depicted in a standard model; the eruption also causes coronal dimming in a large area to the south of the source region. In this study, we focus on the core region where the eruption originates, attempting to understand the early-phase evolution.

This event exhibits a variety of brightening and dimming signatures different from those of the SOL20120712 event. As shown in Figure 6, prominent dimming signatures are observed in both passbands and, in particular, in the EUV 304 Å base ratio images (middle panels). The dimming primarily occurs in two regions demarcated by the orange boxes that are located at the two ends of the filament. The dimming morphology resembles the twin-dimming or core-dimming geometry, similar to the events reported by Webb et al. (2000), Cheng and Qiu (2016), and Wang et al. (2019), suggesting that the twin-dimming regions may map the feet of a flux rope. It is also noted that along the edge of the dimming cores at the far ends of the two flare ribbons, brightening occurs as early as 1:30 UT, well before the onset of the eruption and flare reconnection that produced the two ribbons.

In this event, the brightening is relatively weak compared with the X-class flare studied in S3; therefore, we use EUV 304 Å passband observations, which are more sensitive to weak brightening than the 1,600 Å passband, to identify brightening signatures at the base of the corona. Two kinds of brightening signatures are observed: flare ribbons underlying closed flare loops exhibit impulsive brightening, followed by a prolonged brightening that gradually decays for over 10 min; on the other hand, several pixels of weak and brief brightening are also identified. Figure 7B shows the timing of brightening ¹ superimposed on a photospheric magnetogram of B los . The magnetic flux integrated into the brightening pixels is shown in Figure 7A. It should be noted that due to some mixture of the brightening of coronal features, especially during the filament eruption between 2:15 and 2:40 UT, the measured flux is an overestimate of the reconnection flux ψ rec (see the measurement obtained by Zhu et al. (2020) using 1,600 Å passband images, which likely underestimates the total ψ rec ). It is noted that, in this event, the initial brightening occurs adjacent to the twin dimmings at the far ends of the later formed by two flare ribbons, suggesting that pre-eruption reconnection might take place between the hypothetic flux rope and ambient fields. The amount of flux estimated from the early brightening signatures prior to the eruption (2:15 UT) is about one-fifth of the total reconnection flux.

We identify dimming using EUV 304 Å imaging observations following the method outlined in S2. We form the base image as the average of the images between 1:00 and 1:20 UT and define dimming as persistent attenuation at ≤ 80 % of the base brightness I 0 at the same location for more than 10 min. The onset of dimming, which is shown in Figure 7C, is identified as the time when the brightness starts to decrease below 80% of I 0 . The analysis reveals two groups of dimming signatures: one group exhibits gradual dimming over tens of minutes prior to the eruption, and in the other group, dimming occurs after the onset of the eruption (at 2:15 UT), often preceded by impulsive brightening.

Figure 8 shows the time–distance diagrams of the normalized brightness along a few slits shown in Figure 6 and the light curves of a few selected pixels along the slits exhibiting various dimming–brightening sequences similar to those shown in Figure 1. The epoch plots of the two groups of dimming light curves are given in Figure 9, showing the evolution of the pixel brightness with respect to their peak time. The first group exhibits gradual dimming (Figure 9A), with the observed onset of the dimming ranging between 10 and 100 min (average at 54 ± 19 min) before they brightened. The other group of dimming pixels is characterized by rapid dimming following impulsive brightening (Figure 9B), and the peak dimming depth is approximately 42% ± 12%. In most places, dimming did not recover after more than 3 hours.

The locations of the two groups of dimmings are shown in Figure 7C. The pre-eruption gradual dimmings are within the orange circles; they are mostly clustered in the right region in negative magnetic fields and are bounded by early brightening that outlines a triangle-shaped hook. In the left region in positive magnetic fields, only a small area right next to the flare ribbon exhibits pre-eruption gradual dimming. The post-eruption dimming occupies a larger area in both the left and right feet regions. It appears that, at the onset of the eruption, the pre-eruption dimming spread out into the post-eruption dimming, with a rapid brightening during the transition.

The post-eruption dimming light curves in the second group are similar to those in the SOL20120712 event, suggesting that the erupting structure reconnects with overlying fields and escapes from the corona. However, the geometry of the dimming is different; in this event, the post-eruption dimming appears to map the feet of the erupting structure that has shifted outward after the onset of the flare reconnection, rather than the feet of overlying arcades. This should be confirmed with future data-driven MHD modeling.

The pre-eruption gradual dimming at the feet of the filament is likely a signature of the quasi-equilibrium expansion of a coronal structure. We can estimate the mean speed of the expansion from the slope of the dimming depth (Qiu and Cheng, 2017). At the base of the corona, the optically thin transition-region emission can be modeled as the “pressure gauge” (Qiu et al., 2013, and references therein) so that the brightness is proportional to the mean pressure of the overlying corona. As the overlying corona gradually expands, the dimming in terms of the ratio of the brightness to the base brightness, R ≡ I / I 0 , roughly varies as R ̇ ≈ α ( L 0 / L ) α ( ⟨ v ⟩ / L ) , where α ≥ 1 is a factor close to unity dependent on the gas expansion model, L is the equivalent height along the line of sight, and L 0 is the height before the expansion. For slow (subsonic) expansion, approximating L ≈ L 0 and α ≈ 1 , the expansion velocity is roughly ⟨ v ⟩ ≈ R ̇ L 0 . Fitting the dimming light curves by R ( t ) ≡ I ( t ) / I 0 = R 0 − R ̇ t , we derive the dimming slope R ̇ ≈ 0.001 − 0.01 min − 1 , the average being 0.003 min − 1 , or about 0.3% decrease in brightness per minute. For L 0 of a few tens of Mm, ⟨ v ⟩ is approximately a few kilometers per second. The estimated subsonic expansion speed is consistent with the expansion speed directly observed from limb observations of a different event, which also exhibits persistent gradual dimming, when viewed from the disk, prior to the eruption (Wang et al., 2019).

In summary, the eruptive event SOL20110621 displays different brightening–dimming geometries and sequences from those of the SOL20120712 event. Both pre-eruption and post-eruption dimming occur at the feet of a filament adjacent to flare ribbons, and its geometry resembles the core twin dimming. The twin dimming likely maps the conjugate feet of a coronal structure that gradually expands and then erupts, and this structure carries the total (axial) flux of up to a few times 10 19 Mx, estimated by integrating the flux in the dimming area. The early brightening surrounding the gradually dimming cores may indicate the coronal structure expanding through (and interacting with) the ambient field.

In this paper, we present an experiment that analyzes the brightening and dimming signatures in the lower atmosphere and uses the tempo-spatial sequence of these signatures to identify overlying magnetic structures that are undergoing dynamic evolution in the early phase of their eruption, such as reconnection or expansion. The experiment is applied to two eruptive events, in which pixel light curves exhibit a variety of brightening–dimming sequences.

For the SOL20120712 event, we did not find signatures indicative of dynamic evolution prior to flare reconnection, which occurs nearly simultaneously with the eruption. The erupting structure then reconnects with the overlying arcades, producing impulsive brightening and rapid dimming at the feet of the arcades as the overlying field lines are opened up and removed from their path. We did not find any signatures likely to map the feet of the erupting structure in this event (e.g., Gou et al., 2023). This is possibly due to the stringent criteria for selecting dimming pixels in this study, which require persistent dimming (for more than 10 min) observed with low-temperature lines characterizing signatures at the base of the corona, such as in the transition region. These requirements are reinforced to help minimize projection effects and noise fluctuations; on the other hand, dimming at these lines, particularly in the EUV 304 Å passband, is relatively weak. Therefore, dynamic dimming variations on short timescales may not be detected with the method used in this paper, and the estimated dimming flux is likely to be the lower limit.

The SOL20110621 event exhibits gradual pre-eruption dimming at the feet of a filament that erupts later, and the pre-eruption dimming is accompanied by early brightening that outlines the boundary of the gradual dimming at the far ends of flare ribbons formed later. These are likely signatures of the quasi-equilibrium expansion of a coronal structure through the ambient field, which persists for tens of minutes before the explosive loss of equilibrium, leading to eruption. If this structure embodies a pre-existing flux rope, then its gradual expansion and accompanying reconnection with the overlying fields (e.g., Longcope and Forbes, 2014) would bring it to a larger height in favor of the onset of the ideal instabilities, such as the torus instability (Kliem et al., 2014).

Comprehensive modeling (e.g., Rempel et al., 2023) is needed to reconstruct the three-dimensional magnetic configuration and the evolution toward eruption in both events. The identified tempo-spatial sequence of dimming and brightening provides additional observational constraints, or the observed boundary conditions, for the successful modeling of real eruptive events.