Creating a Solar Stormwatch Catalogue from YOUR clicks.
Hello, I’m Luke and I work with Chris Scott (formerly Davis!) at the University of Reading as a postdoctoral research assistant. Recently I’ve been doing some work with the large amount of Trace-it data that has been generated over the last few years. We thought it was a good time to update everyone on the work we have been doing.
The short story is that we have turned the roughly 40000 time-elongation (t-e) profiles generated by Trace-it into a catalogue of CMEs seen by Heliospheric Imager (HI) instruments aboard STEREO-A (STA) and STEREO-B (STB). The Solar Stormwatch catalogue provides profiles of the CME fronts in the remarkable field-of-view (FOV) of the HI instruments. The HI FOV covers regions of the inner heliosphere not accessible to the coronagraph instruments that are more commonly used to build CME catalogues. Therefore the Solar Stormwatch catalogue should allow us to study the structure and dynamics of CME fronts in a way not previously possible using other presently available CME catalogues. This has been written up into a paper which is currently under review for publication in the journal Space Weather. So, first things first, thank you to everyone that has contributed to Solar Stormwatch. I think that we have produced a useful catalogue of CMEs, which will hopefully be of use to the wider space weather community – this wouldn’t have been possible without all of the contributors to Solar Stormwatch.
Let’s begin with a quick review of the raw data produced by Trace-it. Trace-it analysed J-maps made from HI1 and HI2 images, for both STA and STB, over 18 distinct position angles separated by 5 degrees, except for one position angle, which was centered on the ecliptic plane.
The J-maps covered a time span of January-2007 to February-2010. As of a few months ago, the Trace-it results consisted of database of 38171 t-e profiles, 22007 from STA and 16164 from STB, generated by 4599 Solar Stormwatch users.
If elongation angles and position angles are unfamiliar to anyone reading this, Figure 1 shows an image from HI1A, over which contours of constant position angle (in blue) and constant elongation angle (in red) have been overlaid, to make these coordinates clear.
Figure 1. An example of a differenced image from the HI1A camera, overlaid with contours of
constant PA (in blue) and constant elongation angle (in red). The elongation and PA contours
are in 5◦ increments. A CME is visible to the right of the image, between 5◦ and 10◦ elongation
and with maximum extent in PA between 65◦ and 135
To separate the t-e profiles into groups which represent individual CMEs we looked for periods when many t-e profiles were clustered in a short space of time. To do this, we counted how many t-e profiles began in a 7-hour window, for every hour covering the data set, and whenever the count of profiles was higher than a threshold of 22 counts we defined that as an event. This happens whenever lots of us have seen features over multiple position angles but at similar times. Figure 2 shows an example of this. Panel A) shows a STA J-map, at PA=110 degrees, overlaid with the t-e profiles generated by the Solar Stormwatchers as red dots, whilst the blue dots mark the earliest occurring point in each profile. In this instance, this position angle was tracked 11 times by 8 different Solar Stormwatchers. Panel B) shows the count of these profiles as a function of time, using the 7 hour sliding window. Note that this count is done over every position angle, whereas the J-map shows the t-e profiles at one position angle only. There is a well defined maximum in the count, which we use to define the onset of this event and identify the t-e profiles that describe it. The thresholds we picked are arbitrary but sensible, we could have used different ones and had similar results – for anyone interested in how we picked these numbers, we go into a bit more detail in the paper.
This gives us groups of t-e profiles for each CME – 113 from STA and 80 from STB. However, we had to do a bit of quality control, as it is not good enough just to have to t-e profiles that start at similar times – they could come from coincident but unrelated solar transients that are widely separated in position angle. So we used another set of rules to exclude any t-e profiles which look like they may belong to a different solar transient. This process is detailed more in the paper, but the result of it is that we have to discard 6 events that we are too unsure about, 3 each from STA and STB. This leaves us with 110 events from STA and 77 from STB.
At this point, we have defined sets of t-e profiles which we think robustly identify CMEs seen by the Solar Stormwatchers. The next step is to average these profiles along each position angle the event was observed. An example of this averaging is shown in Figure 3. The black dots show the t-e profiles generated by the Solar Stormwatchers for one event and along one PA, which includes 13 t-e profiles, generated by 9 different Solar Stormwatchers. The red-dots and red-lines show the average profile and the uncertainty in the average profile.
Figure 3. An example of an average t-e profile, for CME number 59 from STA, tracked along a PA of 110 degrees. The black dots show the individual t-e profiles and the red dots mark the consensus profile while the two red lines indicate the uncertainty in the mean time coordinates.
Now we can turn this around and overlay the average t-e profiles for each position angle back onto the original differenced images that made the J-maps they were tracked in. Figure 4 shows a movie of the evolution of an event through the HI1A field-of-view. The yellow lines mark the maximum extent of PAs that the J-maps used by Stormwatch cover, whilst the regions bounded in red mark the locations where the consensus profiles (like figure 3) suggest the CME front should be. The width of the bounded region arises from the uncertainty in the consensus profile at that position angle, so that wider regions mean we are less sure where the CME front is.
Figure 4. This movie shows a sequence of HI1A differenced images in which a CME can be seen to enter and propagate across the HI1A field-of-view. The yellow lines mark the outer limits of the position angles of the J-maps analysed by Trace-it. The red lines mark the location of the CME front, and are calculated from the averaged t-e profiles (see Figure 3) along each position angle the event was tracked.
We are in the process of making this CME catalogue available in an easily usable form so that the rest of the space weather community can get involved and hopefully start using it for some research. In November we will be taking this work to the European Space Weather Week conference in Belgium, to present this work to other researchers. In the meantime, we have some plans for some things we would like to do with the Solar Stormwatch catalogue, which we will update you with when there is more to say.