[Loops] [1802.05093] X-ray flare oscillations track plasma sloshing along star-disk magnetic tubes in Orion star-forming region

[Fabio Reale]

Hi Frederic,
    thank you for your interest and for your comments. I discussed with Javier Lopez-Santiago and in the following please find some clarifications. We were aware of the problems in determining robust confidence levels in de-trended time series, particularly when assuming an AR(1) model as the background behavior. Some years ago, we carried out a probabilistic study to understand whether spurious detections (i.e. high confidence levels for non-physical signatures) were possible in our light curves when using the Torrence & Compo code. The results showed that this code may overestimate the significance of the features in the scalogram (wavelet power spectrum) of the de-trended light curve. The work was presented in a Cool Stars Workshop held in Barcelona. Since then, we started using several methods to prevent false positives, including investigating the PSD and de-trending the light curve using polynomial fitting or frequency filtering. The boxcar or moving average shown in some of our papers is mostly for visualization purposes.

Fortunately, our case is simpler. The oscillatory pattern is also detected in the original time series, before applying any de-trending and also when different time bins are applied, but we also found other light curves where the results were not so clear and we did not use them for our research. 

Your method for determining confidence levels seems robust to us. We will use it with our data to compare with the current method and save your reference for future citations. 

Thank you again. 

Best regards,  
Fabio & Javier


> On 23 Feb 2018, at 12:15, Frédéric Auchère <frederic.auchere at ias.u-psud.fr> wrote:
> 
> Dear Fabio,
> 
> I've read your paper with great interest. These are very interesting results.
> 
> I would however draw your attention to the fact that used as-is, the Torrence & Compo wavelet code can potentially output erroneous confidence levels, especially if used on detrended time-series. There are three reasons for this:
> 
> 1) UV or EUV light curves tend to exhibit an overall power-law behavior. I don't know if this is indeed the case for your light curves but if it is , then the built-in AR(1) red noise model (the one you're using if I'm not mistaken) is generally not a good fit to the spectrum (except in the special case  in which the power law exponent is -2).
> 
> 2) Detrending of the time-series that initially has a power-law power spectrum artificially creates a resulting signal that has dominant power in a narrow band of frequencies. In that case the AR(1) red noise model is likely not a good fit.
> 
> 3) The wavelet confidence levels output by the code are "local" confidence levels, i.e. they don't take into account the total number of degrees of freedom in the spectrum.
> 
> The details of these effects are described in this paper
> 
> http://adsabs.harvard.edu/abs/2016ApJ...825..110A
> 
> I'm not implying that the conclusions that you draw are incorrect, only that the Torrence & Compo code generally cannot be used as-is to robustly detect periodic signals. It was actually not intended to be an all-purpose solution by its authors. The annex of the above paper (corrected in the erratum) gives practical details on how to call the Torrence & Compo routine to use  any custom background noise model, and "global" confidence levels (taking into account the total number of degrees of freedom in the spectra). A demonstration code is available at:
> 
> https://idoc.ias.u-psud.fr/MEDOC/wavelets_tc98
> 
> I hope that  you find these remarks useful for future work.
> 
> All the best,
> 
> Frédéric
> 
> 
> 
> Le 19/02/2018 à 16:34, Fabio Reale a écrit :
>> Dear friends,
>>      here is a paper just accepted for publication in the ApJ. Although it is not strictly on solar loops, it is an extrapolation of a solar loop modeling study (Reale 2016) to very intense and long stellar X-ray flares detected on pre-main sequence stars (Orion cluster). In this case, the detection of long-period oscillations is a strong evidence of very long magnetic tubes, as long as to connect the star to its still-existent accretion disk.
>> Best regards
>> Fabio
>>> https://arxiv.org/abs/1802.05093 <https://arxiv.org/abs/1802.05093>
>>> 
>>> X-ray flare oscillations track plasma sloshing along star-disk magnetic tubes in Orion star-forming region
>>> (Submitted on 14 Feb 2018)
>>> 
>>> Astrophysics > Solar and Stellar Astrophysics
>>> 
>>> 
>>> Pulsing X-ray emission tracks the plasma echo traveling in an extremely long magnetic tube that flares in an Orion Pre-Main Sequence (PMS) star. On the Sun, flares last from minutes to a few hours and the longest-lasting typically involve arcades of closed magnetic tubes. Long-lasting X-ray flares are observed in PMS stars. Large-amplitude (~20%) long-period (~3 hours) pulsations are detected in the light curve of day-long flares observed by the Advanced CCD Imaging Spectrometer (ACIS) on-board Chandra from PMS stars in the Orion cluster. Detailed hydrodynamic modeling of two flares observed on V772 Ori and OW Ori shows that these pulsations may track the sloshing of plasma along a single long magnetic tube, triggered by a sufficiently short (~1 hour) heat pulse. This magnetic tubes are as long (>= 20 solar radii) as to connect the star with the surrounding disk.
>>> Submission history
>>> 
>>> From: Fabio Reale [view email <https://arxiv.org/show-email/b5d3a8b3/1802.05093>]
>>> [v1] Wed, 14 Feb 2018 13:46:37 GMT (480kb)
>> _______________________________________________
>> Loops mailing list
>> Loops at solar.physics.montana.edu
>> https://mithra.physics.montana.edu/mailman/listinfo/loops
> 
> _______________________________________________
> Loops mailing list
> Loops at solar.physics.montana.edu
> https://mithra.physics.montana.edu/mailman/listinfo/loops