Automatic Solar Filament Detection Using Image Processing Techniques

We present an automatic solar filament detection algorithm based on image enhancement, segmentation, pattern recognition, and mathematical morphology methods. This algorithm cannot only detect filaments, but can also identify spines, footpoints, and filament disappearances. It consists of five steps: (1) The stabilized inverse diffusion equation (SIDE) is used to enhance and sharpen filament contours. (2) A new method for automatic threshold selection is proposed to extract filaments from local background. (3) The support vector machine (SVM) is used to differentiate between sunspots and filaments. (4) Once a filament is identified, morphological thinning, pruning, and adaptive edge linking methods are used to determine the filament properties. (5) Finally, we propose a filament matching method to detect filament disappearances. We have successfully applied the algorithm to Hα full-disk images obtained at Big Bear Solar Observatory (BBSO). It has the potential to become the foundation of an automatic solar filament detection system, which will enhance our capabilities of forecasting and predicting geo-effective events and space weather.     

Crossing Filaments

Solar filaments show the position of large-scale polarity-inversion lines and are used for the reconstruction of large-scale solar magnetic field structure on the basis of Hα synoptic charts for the periods that magnetographic measurements are not available. Sometimes crossing filaments are seen in Hα filtergrams. We analyze daily Hα filtergrams from the archive of Big Bear Solar Observatory for the period of 1999 – 2003 to find crossing and interacting filaments. A number of examples are presented and filament patterns are compared with photospheric magnetic field distributions. We have found that all crossing filaments reveal quadrupolar magnetic configurations of the photospheric field and presume the presence of null points in the corona.     

Automated Detection of Filaments and Their Disappearance Using Full-Disc Hα Images

A new algorithm is presented that automatically detects filaments on the Sun in full-disc Hα images. Pre-processing of Hα images includes corrections for limb darkening and foreshortening. Further, by applying suitable intensity and size thresholds, filaments are extracted, while other solar features, e.g. sunspots and plages, are removed. Filament attributes such as their position on the solar disc, total area, length, and number of fragments are determined. In addition, every filament is also labelled with a unique number for identification. The algorithm is capable of following a particular filament through successive images, which allows us to detect their changes and disappearance. We have analysed ten cases of filament eruption from different observatories, and the results obtained are presented. The algorithm will eventually be integrated with an upcoming telescope at the Udaipur Solar Observatory for real-time monitoring of activated/eruptive filaments. This aspect should prove to be of particular importance in studies pertaining to space weather.     

Filament Recognition In Solar Images With The Neural Network Technique

We describe a new technique developed for an automated recognition of solar filaments visible in Hα hydrogen line full-disk spectroheliograms. These filaments are difficult to recognize because of variability in the background caused by atmospheric conditions. The presented technique is based on an artificial neural network (ANN) consisting of two hidden neurons and one output neuron which learn to exclude the contribution of a changeable background to a filament. The ANN is trained on a single image fragment labeled manually to recognize the filament elements depicted on a local background. The background contribution is approximated with linear and parabolic functions. This technique applied to the filament recognition in 54 cropped images reveals better detection results for a parabolic approximation than for a linear one approaching an accuracy of about 82% of the total filament pixels.     

Solar Feature Catalogues In Egso

The Solar Feature Catalogues (SFCs) are created from digitized solar images using automated pattern recognition techniques developed in the European Grid of Solar Observation (EGSO) project. The techniques were applied for detection of sunspots, active regions and filaments in the automatically standardized full-disk solar images in Caii K1, Caii K3 and Hα taken at the Meudon Observatory and white-light images and magnetograms from SOHO/MDI. The results of automated recognition are verified with the manual synoptic maps and available statistical data from other observatories that revealed high detection accuracy. A structured database of the Solar Feature Catalogues is built on the MySQL server for every feature from their recognized parameters and cross-referenced to the original observations. The SFCs are published on the Bradford University web site http://www.cyber.brad.ac.uk/egso/SFC/ with the pre-designed web pages for a search by time, size and location. The SFCs with 9 year coverage (1996–2004) provide any possible information that can be extracted from full disk digital solar images. Thus information can be used for deeper investigation of the feature origin and association with other features for their automated classification and solar activity forecast.     

Automated Technique For Comparison Of Magnetic Field Inversion Lines With Filament Skeletons From The Solar Feature Catalogue

We present an automated technique for comparison of magnetic field inversion-line maps from SOHO/MDI magnetograms with solar filament data from the Solar Feature Catalogue created as part of the European Grid of Solar Observations project. The Euclidean distance transform and connected component labelling are used to identify nearest inversion lines to filament skeletons. Several filament inversion-line characteristics are defined and used to automate the decision whether a particular filament/inversion-line pair is suitable for quantitative comparison of orientation and separation. The technique is tested on 551 filaments from 14 Hα images at various dates, and the distributions of angles and distances between filament skeletons and line-of-sight (LOS) magnetic inversion lines are presented for six levels of magnetic field smoothing. The results showed the robustness of the developed technique which can be applied for a statistical analysis of magnetic field in the vicinity of filaments. The method accuracy is limited by the static filament detection which does not distinguish between filaments, fibrils, pre-condensations and filament barbs and this may increase the asymmetries in magnetic distributions and broadening in angular distributions that requires the incorporation of a feature tracking technique.     

Statistical study of the north–south asymmetry of the solar differential rotation based on various solar structures during 1966–1985

The time variation and latitude dependence of the solar rotation are found using observational data on Hα filaments and compact magnetic features with different polarities during solar activity cycles 20 and 21 (1966–1985). Statistical analysis of the observational data shows that there is a north–south asymmetry in the rotation, both for the Hα filaments and for compact magnetic features (structures) with negative and positive polarities. The N-S asymmetry in the differential rotation of the Hα filaments and the compact magnetic features with both polarities shows up quite distinctly in solar activity cycles 20 and 21, but the asymmetry for the compact magnetic features with positive polarity is comparatively lower in cycle 21. The confidence level is lower the compact magnetic features with positive polarity than for the compact magnetic features with negative polarity.     

Observation of Interactions and Eruptions of Two Filaments

We present new observations of the interactions of two close, but distinct, Hα filaments and their successive eruptions on 5 November 1998. The magnetic fields of the filaments are both of the sinistral type. The interactions between the two filaments were initiated mainly by an active filament of one of them. Before the filament eruptions, two dark plasma ejections and chromospheric brightenings were observed. They indicate that possible magnetic reconnection had occurred between the two filaments. During the first filament eruption, salient dark mass motions transferring from the left erupting filament into the right one were observed. The right filament erupted 40 minutes later. This second filament eruption may have been the result of a loss of stability owing to the sudden mass injection from the left filament. Based on the Hα observations, we have created a sketch for understanding the interactions between two filaments and accompanying activities. The traditional theory of filament merger requires that the filaments share the same filament channel and that the reconnection occurs between the two heads, as simulated by DeVore, Antiochos, and Aulanier (Astrophys. J. 629, 1122, 2005; 646, 1349, 2006). Our interpretation is that the external bodily magnetic reconnection between flux ropes of the same chirality is another possible way for two filament bodies to coalesce. Electronic Supplementary Material The online version of this article () contains supplementary material, which is available to authorized users.     

Magnetic reconnection as the cause of a photospheric canceling feature and mass flows in a filament

Magnetic reconnection in the temperature minimum region of the solar photosphere can account for the canceling magnetic features on the Sun. Litvinenko (1999a) showed that a reconnection model explains the quiet-Sun features with the magnetic flux cancelation rate of order 10¹⁷ Mx hr⁻¹. In this paper the model is applied to cancelation in solar active regions, which is characterized by a much larger rate of cancelation ∖ ge10¹⁹ Mx hr⁻¹. In particular, the evolution of a photospheric canceling feature observed in an active region on July 2, 1994 is studied. The theoretical predictions are demonstrated to be in reasonable agreement with the measured speed of approaching magnetic fragments, the magnetic field in the fragments, and the flux cancelation rate, deduced from the combined Big Bear Hα time-lapse images and videomagnetograms calibrated against the daily NSO/Kitt Peak magnetogram. Of particular interest is the prediction that photospheric reconnection should lead to a significant upward mass flux and the formation of a solar filament. Hα observations indeed showed a filament that had one of its ends spatially superposed with the canceling feature. Supplementary material to this paper is available in electronic form at http://dx.doi.org/10.1023/A:1005284116353     

Periodic Motion Along Solar Filaments

We present observations of four filaments that exhibit large-amplitude periodic mass motion. Observations are obtained using the high resolution (2″) and high cadence (1 min) Hα telescope system at the Big Bear Solar Observatory (BBSO). The motions found in these events are along the axis of the filaments, and are associated with the activity of a nearby flare or filament. The most characteristic properties of these motions are long period (≥ q80 min), large distance (≥ q 4 × 10⁴ km) of mass transport at much higher velocity (≥ q 30 km s⁻¹) than ever detected from filament motions. The velocity, period, dimension and damping timescale measured for these motions are presented, and discussed to identify the most plausible restoring force and damping mechanism.     

Extremely energetic 4B/X17.2 flare and associated phenomena

We observed 4B/X17.2 flare in Hα from super-active region NOAA 10486 at ARIES, Nainital. This is one of the largest flares of current solar cycle 23, which occurred near the Sun’s center and produced extremely energetic emission almost at all wavelengths from γ-ray to radio-waves. The flare is associated with a bright/fast full-halo earth directed CME, strong type II, type III and type IV radio bursts, an intense proton event and GLE. This flare is well observed by SOHO, RHESSI and TRACE. Our Hα observations show the stretching/de-twisting and eruption of helically twisted S shaped (sigmoid) filament in the south-west direction of the active region with bright shock front followed by rapid increase in intensity and area of the gigantic flare. The flare shows almost similar evolution in Hα, EUV and UV. We measure the speed of Hα ribbon separation and the mean value is ∼ 70 km s^-1. This is used together with photospheric magnetic field to infer a magnetic reconnection rate at three HXR sources at the flare maximum. In this paper, we also discuss the energetics of active region filament, flare and associated CME.     

The Eruptive Flare of 15 November 1991: Preflare Phenomena

We present and interpret observations of the preflare phase of the eruptive flare of 15 November, 1991 in NOAA AR 6919. New flux emerged in this region, indicated by arch filaments in Hα and increasing vertical flux in vector magnetograms. With increasing frequency before the eruption, transient dark Hα fibrils were observed that crossed Hα bright plage and the magnetic inversion line to extend from the region of flux emergence to the filament, whose eruption was associated with the flare. These crossing fibrils were dynamic, and were often associated with sites of propagating torsional motion. These sites propagated from the region of flux emergence into the filament flux system. We interpret these morphological and dynamic features in terms of relaxation after magnetic reconnection episodes which create longer field lines within the filament flux system, as envisioned in the tether cutting model, and transfer twist to it, as well. Supplementary material to this paper is available in electronic form at http://dx.doi.org/10.1023/A:1005086108043     

Developing an Advanced Automated Method for Solar Filament Recognition and Its Scientific Application to a Solar Cycle of MLSO Hα Data

We developed a method to automatically detect and trace solar filaments in Hα full-disk images. The program is able not only to recognize filaments and determine their properties, such as the position, the area, the spine, and other relevant parameters, but also to trace the daily evolution of the filaments. The program consists of three steps: First, preprocessing is applied to correct the original images; second, the Canny edge-detection method is used to detect filaments; third, filament properties are recognized through morphological operators. To test the algorithm, we successfully applied it to observations from the Mauna Loa Solar Observatory (MLSO). We analyzed Hα images obtained by the MLSO from 1998 to 2009 and obtained a butterfly diagram of filaments. This shows that the latitudinal migration of solar filaments has three trends in Solar Cycle 23: The drift velocity was fast from 1998 to the solar maximum, after which it became relatively slow. After 2006, the migration became divergent, signifying the solar minimum. About 60 % of the filaments with latitudes higher than 50^° migrate toward the polar regions with relatively high velocities, and the latitudinal migrating speeds in the northern and the southern hemispheres do not differ significantly in Solar Cycle 23.     

A Non-Linear Force-Free Field Model for the Evolving Magnetic Structure of Solar Filaments

In this paper the effect of a small magnetic element approaching the main body of a solar filament is considered through non-linear force-free field modeling. The filament is represented by a series of magnetic dips. Once the dips are calculated, a simple hydrostatic atmosphere model is applied to determine which structures have sufficient column mass depth to be visible in Hα. Two orientations of the bipole are considered, either parallel or anti-parallel to the overlying arcade. The magnetic polarity that lies closest to the filament is then advected towards the filament. Initially for both the dominant and minority polarity advected elements, right/left bearing barbs are produced for dextral/sinsitral filaments. The production of barbs due to dominant polarity elements is a new feature. In later stages the filament breaks into two dipped sections and takes a highly irregular, non-symmetrical form with multiple pillars. The two sections are connected by field lines with double dips even though the twist of the field is less than one turn. Reconnection is not found to play a key role in the break up of the filament. The non-linear force-free fields produce very different results to extrapolated linear-force free fields. For the cases considered here the linear force-free field does not produce the break up of the filament nor the production of barbs as a result of dominant polarity elements.     

Counterstreaming in a Large Polar Crown Filament

The motion of small-scale structures is well resolved in high-resolution filament images that were observed on 19 June 1998 with the Swedish Vacuum Solar Telescope, La Palma. The filament was between 80 000 and 100 000 km high. The study is based on two hours of narrow-band observations at three wavelength positions in Hα. Velocities along the line of sight and in the transverse direction, respectively, V _los and V _tr, were measured for a large number of individual small-scale filament structures. Small features are all moving along nearly parallel threads, some in one direction along the threads and the remainder in the other direction, a pattern of motion known as counterstreaming. The net flow velocities in the two directions are about 8 km s⁻¹ and both are tilted by an angle δ≃16° relative to the plane of the sky. This angle is less than expected, by factors between 2.0 and 2.5, relative to the local horizontal plane. We believe that V _los is underestimated by these factors due to a line-shift reducing effect by the underlying Hα absorption line of the chromosphere. Supplementary material to this paper is available in electronic form at http://dx.doi.org/10.1023/A:1026150809598     

Development of an Automatic Filament Disappearance Detection System

This paper presents an efficient and automatic method for detecting filament disappearances. This method was applied to the Big Bear Solar Observatory's (BBSO) full-disk H images. The initial step is to detect the filaments in the solar image, then determine if they are growing, stable or disappearing. If a disappearing filament is found, the solar community can be automatically alerted in near real time. This system is proven to be accurate and fast. In addition, three statistical studies of the appearance and disappearance of all filaments in 1999 are presented.     

Filament activity in a quiet region flare

We have taken the case of a circular Hα filament observed on May 9,1979 erupting into a double-ribbon flare associated with a non-spot region. The plage motions are responsible for the filament reorientation and, here as a special case, wherein the filament attains a clear circular shape before the onset of a flare. We conclude that the change in the orientation of the Hα filament marks the instability giving rise to the flare.     

Preflare state

Discussion on the preflare state held at the Ottawa Flares 22 Workshop focused on the interpretation of solar magnetograms and of H filament activity. Magnetograms from several observatories provided evidence of significant build up of electric currents in flaring regions. Images of X-ray emitting structures provided a clear example of magnetic relaxation in the course of a flare. Emerging and cancelling magnetic fields appear to be important for triggering flares and for the formation of filaments, which are associated with eruptive flares. Filaments may become unstable by the build up of electric current helicity. Examples of heliform eruptive filaments were presented at the Workshop. Theoretical models linking filaments and flares are briefly reviewed.Report of Team 1, Flares 22 Workshop, Ottawa, May 25–28, 1993     

Variations of the Solar Differential Rotation Associated with Polarity Reversal

Variations of solar differential rotation have been studied using observations of solar quiescent Hα filaments obtained during 1965–1993 at the Abastumani Astrophysical Observatory. In both hemispheres of the Sun, propagation of a quasi-biennial pulse of residual rotation velocities of filaments was found. There is a pulse drift from high latitudes to the equator in the northern hemisphere in 1968–1970, 1979–1981, 1988–1990 and in the southern one in 1969–1971, 1979–1981, 1989–1991. Propagation of a pulse starts near the time of the polarity reversal of the circumpolar regions of the Sun. High-latitude double peaks of rapid motion were found in the northern hemisphere for cycle 20 and in the southern hemisphere for cycle 22. The relation of the appearance of suggested double pulse peaks of residual velocities with the threefold polarity changing of the circumpolar areas is suggested.     

Activity cycle of solar filaments

Long-term variation in the distribution of the solar filaments observed at the Observatorie de Paris, Section de Meudon from March 1919 to December 1989 is presented to compare with sunspot cycle and to study the periodicity in the filament activity, namely the periods of the coronal activity with the Morlet wavelet used. It is inferred that the activity cycle of solar filaments should have the same cycle length as sunspot cycle, but the cycle behavior of solar filaments is globally similar in profile with, but different in detail from, that of sunspot cycles. The amplitude of solar magnetic activity should not keep in phase with the complexity of solar magnetic activity. The possible periods in the filament activity are about 10.44 and 19.20 years. The wavelet local power spectrum of the period 10.44 years is statistically significant during the whole consideration time. The wavelet local power spectrum of the period 19.20 years is under the 95% confidence spectrum during the whole consideration time, but over the mean red-noise spectrum of α = 0.72 before approximate Carrington rotation number 1500, and after that the filament activity does not statistically show the period. Wavelet reconstruction indicates that the early data of the filament archive (in and before cycle 16) are more noiseful than the later (in and after cycle 17).