British Astronomical Association

Supporting amateur astronomers since 1890

Hunting flares on red dwarfs

2020 June 4

Andrew Smith

Having decided to relocate my telescope to the PixelSkies site at Castilléjar in Spain,1 I was looking for a project that would benefit from the frequent clear skies, build on my interest in automated spectroscopy and not overwhelm me with data to process. I decided that, at least initially, high-resolution spectroscopy might be a step too far, so I went for ultra-low-resolution spectroscopy but at high cadence.

I had always been impressed by a serendipitous flare on DN Tau, captured by Robin Leadbeater when preparing for low-resolution observations of T Tauri stars.2 While T Tauri stars are pre-main sequence stars prone to irregular flaring, it is thought that all main sequence stars with significant convective regions flare. This includes the late-M dwarfs, i.e. red dwarfs, which are nearly – if not fully – convective, and for which low intrinsic luminosity makes the flares very conspicuous. These are the UV Cet variables which I decided to study.3

Figure 1. Ultra-low-resolution transmission spectrograph.
Figure 1. Ultra-low-resolution transmission spectrograph.

The spectroscope I built (Figure 1) is a development of Robin’s ‘junk box’ spectrograph,4 but optimised for remote operation. The rotator allows for setting the position angle of the camera so that the target spectra can be kept free of contamination from field stars and their spectra. The filter wheel allows a choice of Star Analyser (100 or 200 lines/mm) or photometric filters. The use of a ZWO ASI1600 MM CMOS camera allows for fast download, minimising the dead time between exposures.

The spectrometer is coupled to an Orion Optics ODK 16, mounted on a Software Bisque Paramount ME II. Used together with The Sky and CCDAutoPilot software, it is possible to acquire calibration star(s)/target(s) and take spectra all night without intervention. PixelSkies manages the roof, so I can sleep in peace.

To check if a flare was captured during a night’s observing (typically resulting in some 1,000 to 2,000 images), I use AstroImageJ to align and then perform multi-aperture photometry.5 This is done on the remote observatory computer, as one disadvantage of a remote dark site is the need to use satellite Internet. This requires keyboard discipline, patience and limiting data transfer to essentials. It takes about 20 minutes a day to check for a flare and set up for the next night’s run.

Figure 2. Discovery plot of a flare on HIP 25953, 2019 Dec 14.
Figure 2. Discovery plot of a flare on HIP 25953, 2019 Dec 14.

Overwhelmingly, this produces a negative result. However, very occasionally, a flare is found (Figure 2). On finding one I then perform normal image calibration (again using AstroImageJ), crop the relevant images and transfer them to my home computer. The spectra are then extracted and processed using BASS Project.6

Figure 3. Spectra of the rising phase of a flare on HD 25953. The flare reached a peak in approximately 320s. The black spectrum is the average of the 20 spectra captured just before the flare. The three other spectra show the rapid rise. Each spectrum is a 100s exposure, with a 6s delay between them.
Figure 3. Spectra of the rising phase of a flare on HD 25953. The flare reached a peak in approximately 320s. The black spectrum is the average of the 20 spectra captured just before the flare. The three other spectra show the rapid rise. Each spectrum is a 100s exposure, with a 6s delay between them.

After some 18 hours and 4,600 images, I captured a flare on HIP 25953 (dM4e). Figure 3 shows the spectra taken with the Star Analyser 200 during the rise of this flare, which took approximately 320s to reach the observed maximum. Figure 4 shows a comparison of the peak flare spectrum with the average pre-flare spectrum subtracted, together with a Planck curve for 9,500K. M-dwarfs typically have photospheres with effective temperatures of 3,800K at M0, falling to 2,300K at M9 and being about 3,100K for M4 so the flare is significantly hotter. Living in the habitable zone of a flaring M-dwarf would be a challenge!

Figure 4. Spectrum of the peak of the flare, with the average pre-flare spectra subtracted. The spectrum is well approximated by a Planck curve for 9,500K with additional emission from the alpha, beta and gamma Balmer lines of Hydrogen.
Figure 4. Spectrum of the peak of the flare, with the average pre-flare spectra subtracted. The spectrum is well approximated by a Planck curve for 9,500K with additional emission from the alpha, beta and gamma Balmer lines of Hydrogen.

As I accumulate more flare data (currently obtained at about one per month), I will seek to identify the best way to extract information from them and refine my methods of capture and processing. I am not sure what, if anything, of scientific value will come of this but the thrill of discovering a flare is in itself reward enough for the effort involved.

References

1   PixelSkies: pixelskiesastro.com/

2   Leadbeater R., Three Hills Observatory: bit.ly/2ZDprOm

3   AAVSO: bit.ly/2LZVvUE

4   Leadbeater R., Three Hills Observatory: bit.ly/3d0Rxag

5   AstroImageJ: bit.ly/3d2KTAf

6   BASS Project: groups.io/g/BassSpectro

The British Astronomical Association supports amateur astronomers around the UK and the rest of the world. Find out more about the BAA or join us.