1 Introduction

Systematic observations of the full solar disk started at the Meudon Observatory in 1908 with Deslandres spectroheliograph (Malherbe and Dalmasse, 2019). Spectroheliograms are obtained by spectroscopic scans of the Sun, producing line profiles and narrow bandpass images at constant wavelength. This is a low cadence instrument dedicated to the study of long term solar activity.

In addition several heliographs performed high cadence (60 seconds) H\(\alpha\) observations from 1956 to 2004 at Meudon and Haute Provence (OHP) observatories, which produced more than 7 million images. This survey of fast solar activity (flares, evolution of active regions and filaments) was based on Lyot filters and started in the frame of the International Geophysical Year (IGY 1957). Unfortunately, only 10% of the collection is digitized, the oldest part being on films. Nowadays, such observations are mainly done with narrow band imagers such as Fabry–Pérot filters (e.g. Global Oscillation Network Group (GONG) H\(\alpha \) network: Harvey et al., 2011), but some Lyot filters remain in activity (e.g. Global H\(\alpha \) Network (GHN): Gallagher et al., 2002).

This work is intended to promote high-cadence observations of the Sun in the context of space weather monitoring. Section 2 recalls the scientific interest of such observations. In Section 3, we present the historical background summarizing the evolutions and main discoveries of the Meudon and OHP heliographs (1956–2004). Section 4 describes the new instrument which is going to replace them at Calern observatory. Section 5 discusses new perspectives in terms of flare onset and Moreton waves detection.

2 Scientific Goal of High Cadence Observations

Solar flares and coronal mass ejections (CMEs) are energetic events associated with the sudden release of magnetic energy in response to the development of resistive (Furth, Killeen, and Rosenbluth, 1963; Antiochos, DeVore, and Klimchuk, 1999) or ideal (Amari and Luciani, 1999; Kliem and Török, 2006) MHD instabilities in the coronal magnetic field. During these events, typically \(10^{28}\) – \(10^{33}\) erg of magnetic energy are released in about \(10^{3}\) – \(10^{4}\text{ s}\) (Shibata and Magara, 2011). The energy is released by magnetic reconnection and converted into thermal, kinetic and radiated energy (Carmichael, 1964; Sturrock, 1966; Hirayama, 1974; Kopp and Pneuman, 1976). Through particle acceleration and ejections, flares and CMEs are the two major drivers of space weather that can cause various environmental hazards at the Earth (Schrijver et al., 2012), and for which high-cadence observations of the Sun are required.

H\(\alpha\) is the best line for fast imaging of chromospheric structures such as filaments, which are cool and dense material that appear as elongated and dark structures. They are formed of chromospheric plasma confined in highly stressed coronal magnetic fields that overlay polarity inversion lines (Aulanier and Démoulin, 1998; Schmieder et al., 2006). Filaments are typical precursors of flares, including confined flares associated with failed filament eruption (Török and Kliem, 2005) and flares for which the filament erupts and leads to the formation of a CME (Moore et al., 2001). In the latter case, the filament is usually characterized by a slow rise (taking hours) until it reaches a critical height beyond which the system becomes unstable and the filament erupts (Kahler et al., 1988; Sterling and Moore, 2005). In this regard, high-cadence observations of filaments are needed for near real-time prediction of their ejection.

H\(\alpha\) imaging also provides observations of flare ribbons, a useful signature for real-time detection of flares for which there is either no filament eruption (Dalmasse et al., 2015) or no filament at all (Masson et al., 2009). Flare ribbons are surface brightenings that can be observed in ultraviolet (UV) and H\(\alpha\) (Schmieder et al., 1987). They are caused by the interaction of energetic particles and thermal energy (produced at the reconnection region) with the lower and denser atmospheric layers (review by Fletcher et al., 2011). The spatio-temporal evolution of H\(\alpha\) ribbons provides information on the flare dynamics. If one further combines H\(\alpha\) data with photospheric magnetograms and some sort of modeling, it is then possible to derive the reconnection rate (Forbes and Priest, 1984; Qiu et al., 2002), the energy release rate (Asai et al., 2004; Isobe, Takasaki, and Shibata, 2005), and the flare energy (Toriumi et al., 2017).

Moreton waves are another flare-related signature for which high-cadence imaging is required. They are large-scale disturbances propagating in the solar atmosphere in H\(\alpha\) observations (Moreton, 1960). They materialize as arc-shaped, bright fronts in the center and blue wing of the H\(\alpha\) line, which propagate out to distances of up to 600 Mm in the extreme cases (review by Warmuth, 2015). Their typical propagation speed is in the range 400 – 2500 km s−1 (Narukage et al., 2002; Muhr et al., 2010; Liu et al., 2013). The speed and signature argue in favour of a chromospheric counterpart of a coronal perturbation that compresses and pushes the chromosphere downwards (Uchida, 1968). Coronal waves were detected by the Extreme ultraviolet Imager Telescope (EIT) onboard SOHO (Klassen et al., 2000) and later by the Atmospheric Imaging Assembly (AIA) onboard SDO (Nitta et al., 2013), with better temporal and spatial resolution. Their physical nature (CME-driven or freely propagating MHD shock-wave) is not yet fully understood.

Moreton waves can disturb the ambient coronal magnetic field and remote filaments, and seem to be associated with strongest flare or CME events (Warmuth, 2015), which are more likely to be very geo-effective. Moreton waves detection at high cadence, by running- or base-difference H\(\alpha\) imaging of the chromosphere (Section 5), should allow one to identify in real-time the potentially most energetic and dangerous eruptive events in space weather applications.

In this context, the new instrument presented here has two major goals:

  1. i)

    solar activity monitoring in chromospheric lines as H\(\alpha\) center (flares and filaments) and Caii K (magnetic proxy);

  2. ii)

    flare, CME onset and Moreton wave detection using two high cadence (15 seconds) H\(\alpha\) filters (center and blue wing).

3 Historical Background of the French Heliographs

More than 7 million images were produced both at Meudon and OHP between 1956 and 2004 in the context of high-cadence (60 seconds) observations to investigate solar activity events (Table 1 and Figure 4).

Table 1 1956-2004 H\(\alpha\) observations at Meudon and OHP (FS = full Sun; PS = part Sun). Three wavelengths mean that line center, blue and red wings (\(\pm0.75\) Å) are observed in sequence, while for five wavelengths, we have four wing positions (\(\pm1.0\) Å, \(\pm0.5\) Å).

The story began with the invention of the monochromatic birefringent filter by Lyot (1944). The first H\(\alpha\) heliograph was built by Grenat and Laborde (1954). Systematic observations started in 1956 with 35 mm films (Figure 1). The 45 meters long films (Kodak Technical Pan TP2415) recorded each 2200 full Sun images, 15 mm diameter and 0.75 Å FWHM. The film resolution (better than 125 lines/mm or less than \(1''\) at the Sun) was limited by the seeing (\(2''\) typical). A similar instrument started at OHP in 1958 and worked until 1995, producing 2.6 million images. The pair of sites allowed for an increase in temporal coverage, as clear sky at OHP is more frequent. The 1957 IGY motivated the scientific program of both instruments.

Figure 1
figure 1

Example of 11 mm H\(\alpha\) monochromatic full Sun images got with Meudon heliograph between 1965 and 1985, 0.75 Å FWHM (4 October 1965 from 16:08 to 16:13 UT).

Between 1960 and 1965, Meudon observations were interrupted but continued at OHP. A new three-wavelength instrument was built for observations at H\(\alpha\) center, blue and red wings (±0.75 Å) using a three-stage tunable Lyot filter with motorized rotating plates (Figure 2, Michard, 1965, Demarcq et al., 1985). Observations started in 1965, covering about 30% of the solar area (\(16 \times 20\) mm FOV selected in the 35 mm image of the Sun). For that reason, a second routine, full disk and only at H\(\alpha\) center, started at the same time (11 mm solar diameter). Both instruments had 0.75 Å FWHM and were in production until 1984.

Figure 2
figure 2

Example of three-wavelength H\(\alpha\) part Sun images got with Meudon heliograph between 1965 and 1984: line center, blue and red wings (±0.75 Å), 0.75 Å FWHM, here images of 4 October 1965 at 12:38 UT).

In 1985, a new tunable Lyot filter, with better performance (H\(\alpha\) center, blue and red wings (±0.50 Å, ±1.0 Å, 0.50 Å FWHM) was developed (Figure 3, Demarcq et al., 1985). It was mounted between two lenses of 360 mm focal length; this afocal system was fed by a 150/2250 mm telescope. Observers chose three or five wavelengths according to solar activity level, plus a long exposure image for prominences at line center. This five-stage Lyot filter (11 Å distance between maxima) was optimized to cut secondary lobes. The H\(\alpha\) peak was isolated by a three cavity 3.5 Å FWHM blocking filter. This device provided full disk images (21 mm diameter), such that the old 11 mm H\(\alpha\) center routine was stopped. A \(1536 \times 1024\) cooled CCD camera from Princeton Instruments, supporting a 140 mm objective, replaced the film in 1999 (9 \(\mu\) pixel, 12 bits, cooled KAF1600 sensor), producing 0.7 million FITS images until 2004 (Table 1 shows that, contrarily to the film, CCD images were undersampled). FITS data are off line (CD/DVD), but freely available upon request. Light curves and quick look MPEG movies are on line at the BASS2000 solar database (http://bass2000.obspm.fr/home.php?lang=en).

Figure 3
figure 3

Example of five-wavelength H\(\alpha\) full Sun images got with Meudon heliograph between 1985 and 1997, line center, red and blue wings (±0.1 Å, ±0.50 Å) and overexposed line core for prominences (0.50 Å FWHM, flare of 2 June 1991 at 14:11 UT).

Figure 4 summarizes observations made between 1956 and 2004. About 6.5 million images have been recorded on 35 mm films (a total of 130 kilometers), and 0.7 million of CCD images are archived. The detailed list is available at: http://www.lesia.obspm.fr/perso/jean-marie-malherbe/heliograph/index.html

Figure 4
figure 4

Number of H\(\alpha\) images got with Meudon and OHP heliographs between 1956 and 2004 (green: Meudon, 11 mm images; pink: Haute Provence, 15 mm images; blue: Meudon, three-wavelength (part Sun until 1985, full Sun after); red: Meudon, three-wavelength CCD full Sun).

The cost makes impossible a systematic scan of the 3000 films of this exceptional collection, but films of interest can be digitized individually upon request (8 bits, 157 pixels/mm scans, sampling better than \(1''\)/pixel, TIF format, one file for 100 mm of film).

Observations have been mainly used by the Meudon group and exploited qualitatively, as long as films were not digitized, to investigate chromospheric flares and filament instabilities. The multi-wavelength filters allowed one to study mass motions in active regions.

Concerning flares, Martres and Pick (1962) found a relationship with radio bursts observed in Nançay. Mouradian, Martres, and Soru-Escaut (1983) suggested that a flare is composed of several elementary eruptive phenomena in relation with magnetic emerging flux, involving the presence of both cold (the surging arch at \(10^{4}\) K) and hot (the flaring arch at \(10^{7}\) K) magnetic loops. Martres (1989) found evidence for homologous flaring. Mouradian et al. (1989) compared H\(\alpha\) features and X-ray brightenings at flare onset. Later with the digital version of the three-wavelength instrument, Pick et al. (2005) and Maia et al. (2003) investigated CMEs and flares using H\(\alpha\) and radio data.

As for filaments, Mouradian and Soru-Escaut (1989) discovered the existence of rigid rotation points (the pivot points) in some filaments which could play a role in instabilities. Soru-Escaut and Mouradian (1990) suggested that heating and cooling mechanisms could explain several cases of sudden disappearances and reappearances. Hence, Mouradian, Soru-Escaut, and Pojoga (1995) proposed two physical classes of disappearances, involving either thermal or dynamic processes.

4 The New Heliograph

A small rolling house has been built at Calern observatory (1270 m elevation) in order to have better weather and seeing conditions than at Meudon or OHP. The equatorial mount was constructed by Valmeca. It supports the instruments which are enclosed in a \(1.7 \times 0.5 \times 0.5~\mbox{m}^{3}\) box, thermally regulated above ambient temperature at \(27^{\circ}\) C by active heating and passive cooling.

The new instrument is composed of three telescopes (Tables 2 and 3) corresponding to the design of Figure 5. The two H\(\alpha\) telescopes are identical, except for the Fabry–Pérot etalons manufactured by DayStar corporation (professional series). The first filter is line core centered, in order to observe active regions, flares and filaments. The second one is adjusted to the blue wing (choice discussed in Section 5). The third telescope is centered on Caii K with an interference filter from Barr company. The equivalent focal length is 983 mm providing a 9.13 mm solar image and \(35'\) FOV. Each detector (Table 4) is a 12 bits cooled camera from Quantum Scientific Imaging using shutterless interline CCD sensors from Sony (ICX694 and 814, respectively, for H\(\alpha\) and Caii K). The readout noise is 7 electrons RMS at 8 MHz. Exposure times are shorter than 10 ms.

Figure 5
figure 5

Optical design of the two H\(\alpha\) (top) and the Caii K (bottom) telescopes.

Table 2 Optical design (the two H\(\alpha\) telescopes are identical, apart the filter).
Table 3 Filter characteristics.
Table 4 Detector characteristics.

H\(\alpha\) telescopes are afocal systems. O1 is the entrance objective (Takahashi TSA102, 102 mm diameter) protected in full aperture by a Baader energy rejection filter (ERF). The Fabry–Pérot etalon is located in the pupil image at F/30 between O2 and O3 (maximum beam aperture constrained by the manufacturer). O4 (a set of two separate lenses) is an amplifier and field corrector. A 80 mm diaphragm limits the optical resolution to \(2.0''\).

The calibration of the two H\(\alpha\) filters was done at F/60 with the high resolution spectrograph (R = 300000) of the Meudon solar tower. The surface filter (31.75 mm diameter) was scanned by the slit. Both filters exhibit 0.32 Å FWHM (homogeneous over the filter), but the scans show that the CWL varies from place to place. The pupil plane location of the afocal design allows one to correct this effect and delivers a uniform CWL image, but, as a counterpart, it increases the global FWHM (Figure 6). Hence, filter 1 (2019) is almost perfect (0.34 Å global FWHM) and is dedicated to H\(\alpha\) center images. Filter 2 (2009) has some optical defects (0.46 Å global FWHM) and is used for H\(\alpha\) wing (Moreton waves detection); it exhibits a 10% parasitic internal reflection which vanishes in the running-difference process (Section 5). Each Fabry–Pérot etalon is solid (mica spaced) and \(\lambda\)/10 surfaced. The H\(\alpha\) peak is selected by an internal 2-cavity 8-10 Å FWHM blocking filter. Adjacent peaks are about 20 Å apart with only 0.34% transmission of the central peak. However, the blocking filter of etalon 2 is shifted by 2.8 Å. A colored longpass glass filter (RG630) suppresses short wavelengths (UV, blue, green), while IR radiation is rejected by a shortpass filter.

Figure 6
figure 6

Calibration at F/60 of the two H\(\alpha\) filters (0.34 Å FWHM top, 0.46 Å FWHM bottom). Left: H\(\alpha\) observed line at disk center with or without the filter. Right: transmission functions; black line: observed line profile; dashed line: local typical transmission (0.32 Å FWHM); red line: global transmission averaged over the filter surface for pupil plane application.

Both etalons are thermally regulated; the CWL is temperature dependent (\(9^{\circ}\)C/Å typical). As the wavelength shift is very slow (about 0.1 Å/minute), contrarily to Lyot tunable filters based on rotating plates, it is not possible to scan line profiles, so that two filters are required for two wavelength positions (Figure 7).

Figure 7
figure 7

Images obtained with the 0.46 Å FWHM filter (29 August 2017) in H\(\alpha\)-0.5 Å (blue wing, left) and with the 0.34 Å FWHM filter (21 March 2019) in H\(\alpha\) line core (right).

The Caii K telescope provides a magnetic field proxy for active regions and faculae (Pevtsov et al., 2016). As the filter is broad (1.5 Å FWHM), the design does not need to be afocal. O1 is the entrance objective (Takahashi FS102, 102 mm diameter). O2 and O3 constitute a focal amplifier. The 3934 Å filter is located close to the image plane and protected by an ERF. A 80 mm diaphragm limits the resolution to \(1.25''\).

Meteospace is an autonomous station: opening and closing the dome, weather control, catching the Sun, observations, data processing, real-time dissemination, database archiving: all operations are automated. Sensors permanently control the instrument and can decide to stop and close in case of cloud, rain, alarm or failure; alerts are sent to the local staff by a Short Message Service (SMS). The entry point to the data is the BASS2000 service, even if the full archive is physically located at Nice.

All data will be delivered freely to the solar community. Raw (level 0) data as well as standard processed data (level 1) will be available in real-time JPEG for quick look purpose and FITS for scientific use. Level 1 includes corrections as dark current, distortion and solar image rotation to present solar north up. The observing cadence is, respectively for H\(\alpha\) and Caii K, 15 and 60 seconds. The system will run systematically (weather permitted) under seeing conditions better than \(2''\) (sampling reported in Table 4). Following Thompson (2006), FITS headers include World Coordinates System (WCS) information concerning image pointing and orientation, allowing for the use of positioning tools.

5 New Perspectives for Flare Onset and Moreton Wave Detection

The purpose of the two high-cadence H\(\alpha\) telescopes is to improve the detection of flare onset and Moreton waves. Such events are suspected to be the counterpart of fast coronal EUV waves in the chromosphere, which could be compressed by the coronal shock above, producing 5 – 10 km s−1 downflows (Warmuth, 2015).

We simulated the detection of Moreton waves by the 0.46 Å FWHM filter using the difference signal between a shifted profile and the quiet sun profile (Figure 8). We used the atlas profile from Delbouille, Roland, and Neven (1973) altered by 15% scattered light and shifted from −10 to +10 km s−1. The filter CWL was moved from the blue to the red wing (−1.0 Å to +1.0 Å by 0.25 Å step). The best sensitivity is obtained, for chromospheric downflows, when the filter selects the blue wing (−0.5 Å, 9% and 17% enhancement respectively for 5 and 10 km s−1). In the case of simultaneous up or downflows, both wings are convenient. These results are illustrated by movie 1 showing running differences of the 28 October 2003 event observed by the tunable Meudon Lyot filter (H\(\alpha\) center and wings).

Figure 8
figure 8

Running-difference signal as a function of Doppler velocity for different shifts of the filter bandpass (−1.0 to +1.0 Å by 0.25 Å step). Solid lines: blueshifted bandpass; dashed lines: redshifted bandpass. Bandpass shift: no shift (black), \(\pm0.25\) Å (yellow), \(\pm0.50\) Å (red), \(\pm0.75\) Å (green), \(\pm1.0\) Å (blue).

As Moreton waves will probably not occur before the next maximum, a simulation of typical data which will be provided by the new instrument, based on 28 October 2003 data, is displayed in Figure 9. Real-time images and movies will be produced, as central intensities \(I_{c}(t)\) or contrasts \(C_{c}(t)\) and running differences of the blue wing intensity \(I_{b}(t)-I_{b}(t-1)\) or contrast \(C_{b}(t)-C_{b}(t-1)\). Contrasts give smoother results and are defined by \(C = \frac{I}{LD} - 1\), where LD is the limb darkening function. It is built from the image by taking median values in concentric rings.

Figure 9
figure 9

Simulation (based on the 28 October 2003 event at 11:07:12 UT) of data which will be produced by the new heliograph. Top: real-time images; central intensity (left) and running difference of the blue wing intensity (right). Bottom: post-processed images; contrast at line center suppressing the limb darkening (left) and Doppler proxy (right) issued from the intensity difference between the blue wing and line center, after correction by an offset observed before the event.

Post-processed images will also be delivered. The two H\(\alpha\) filters can be combined to give \(I'(t)= I_{b}(t)-I_{c}(t)\), \(I_{b}\) and \(I_{c}\) being, respectively, the blue wing and central intensity. Then an offset \(I'(t_{0})\), measured before the event, is subtracted to the current signal \(I'(t)\), as suggested by Muhr et al. (2010). This provides a proxy of the Doppler velocity. Results issued from both wings would be better, but this cannot be done with our instrument. The method can be applied to contrasts instead of intensities, as shown by movie 2.

Hence, efforts to implement real-time detection of Moreton events using running differences of blue wing intensities are under way and will be refined using initial datasets from the instrument as it is commissioned. This tool, together with the survey of flare brightening in line core, could be used to monitor flare onset and anticipate their possible impact at the Earth.

6 Discussion and Conclusion

More than 7 million images at 60 seconds cadence (flares, CME onset, filament instabilities, Moreton waves) have been obtained at Meudon and OHP from 1956 to 2004 using H\(\alpha\) Lyot filters, either line centered or three-wavelength (center + wings). Systematic observations will restart in 2020 at Calern observatory using more compact and cheaper technology. The new heliograph is based on two commercial Fabry–Pérot etalons. However, such filters have far photospheric wings (Lorentzian shape), contrarily to Lyot filters which mainly pass the chromosphere (Figure 10). Let us define the energy bandpass (EB) as the wavelength interval containing half of the energy transmitted by the filter. It is computed from the spectral energy (the product of the filter transmission curve by the quiet H\(\alpha\) disk center line profile). The EB of the Meudon three-wavelength Lyot filters (1965–1984 and 1985–1997), respectively 0.62 Å and 0.34 Å, is always smaller than their FWHM (the second filter was optically more optimized). In comparison, the distance between inflexion points of the H\(\alpha\) line is 1.0 Å. On the contrary, our industry Fabry–Pérot devices provide 0.53 Å and 0.77 Å EB, greater than their respective FWHM. Hence, more photospheric light will pass in comparison to previous Lyot prototypes, reducing for instance filament contrasts. However, adding 2.0 or 3.0 Å FWHM pre-filters would allow one, in the future, to minimize photospheric wing contributions and achieve former performance.

Figure 10
figure 10

Spectral energy transmitted by old H\(\alpha\) Meudon Lyot prototypes (blue, 1985–2004 solid, 1965–1984 dashed) and the two DayStar Fabry–Pérot etalons (red, solid and dashed).

Three telescopes are available: high cadence (15 seconds) H\(\alpha\) center and blue wing, and medium cadence (60 s) Caii K (active regions and magnetic field proxy). Real-time and post-processed images for scientific purposes will be produced: H\(\alpha\) center for filament eruption and flare brightening detection, as well as running difference of blue wing for Moreton waves (often involved in large flares). All images (quick look JPEG and scientific FITS) will be freely available to the solar community through BASS2000, without any delay.

After 16 years of interruption (due to the lack of observers), this new automatic instrument will resume the exceptional survey started at the IGY. It will complete the GHN and GONG networks. GONG has only one european station in Tenerife. Weather conditions in Côte d’Azur are the best available in France for continuous observations. This opportunity allows one to offer more time coverage to existing networks, improved cadence, core and wing observations. The new routine is dedicated to space weather scientific studies related to flares, CME and filament instabilities, as well as operational monitoring and forecast by the french Air Force. Detailed opto-mechanical drawings are available for reproduction at other places.