search

dstansby / pfsspy

Potential Field Source Surface model package for Python
https://pfsspy.readthedocs.io/
Other
40 stars 17 forks source link

Add gallery example with HPC magnetogram #322

Closed wtbarnes closed 1 year ago

wtbarnes commented 3 years ago

I often find myself wanting to do field extrapolations of particular ARs at particular times to match up with AIA images. This is difficult to do with synoptic observations (because there aren't corresponding AIA images), but this can be mimicked by reprojecting the closest HMI HPC observation into a CAR (or CEA) and then doing the extrapolation from that reprojected magnetogram. It would be nice to have a gallery example that shows that workflow or perhaps the existing AIA fieldline example could be modified to use this approach?

dstansby commented 3 years ago

Reprojecting a single HMI image into CEA doesn't work, as you need the magnetic field over the whole solar surface (not just half) to do the extrapolation.

Perhaps what you're after is a better way of specifying the seed points for tracing lines near the an active region of interest?

wtbarnes commented 3 years ago

What exactly do you mean by "doesn't work"? I only ask because I was able to do what I described above and comparing the loops I get out of that extrapolation to the corresponding EUV images (from a few different observers) seems to yield structures that look reasonably like those found in 171. Note that I'm not concerned at all about the global structure of the field, only the field in a relatively small patch of Sun (~400 by 400").

image

That being said, having a way to better specify seed points in a region of interest would be useful.

dstansby commented 3 years ago

What exactly do you mean by "doesn't work"?

I meant that you can't take a single take a single HMI full-disc image and do a PFSS extrapolation from that, as you would be missing the far-side data that is needed to do a global PFSS extrapolation. Is this what you've done above?

That being said, having a way to better specify seed points in a region of interest would be useful.

Through the magic of sunpy.coordinates you can specify the seed points in whatever frame you want - see https://github.com/dstansby/pfsspy/pull/323/ for an example where the seed points are defined directly on an AIA image.

wtbarnes commented 3 years ago

I meant that you can't take a single take a single HMI full-disc image and do a PFSS extrapolation from that, as you would be missing the far-side data that is needed to do a global PFSS extrapolation. Is this what you've done above?

Yes, that's what I've done. I understand what you're saying and certainly agree when it comes to the global structure of the field. Do you have a sense for how much this matters for the closed field on approximately an AR scale? Does the lack of information from the far-side influence the local field? I guess a good test of this would be to build up a synoptic map "centered" on this particular full-disk HMI observation, do the extrapolation, and compare the two.

Through the magic of sunpy.coordinates you can specify the seed points in whatever frame you want - see #323 for an example where the seed points are defined directly on an AIA image.

Yep! The issue is primarily the timing, i.e. you're never guaranteed that there is a synoptic map that includes the AR you want for a specified AIA image.

dstansby commented 3 years ago

The PFSS equations are linear, so a full-map solution = the sum of two half map solutions. Assuming you've set the far-side data to 0, this means you can take your half-map solution, and look at the magnetic field solution on the far-side to understand how much one hemisphere of a map affects the magnetic field solution in another hemisphere. I can believe that on small scales it wouldn't have much of an effect.

You might also be interested in https://github.com/dstansby/pfsspy/pull/260

wtbarnes commented 3 years ago

The PFSS equations are linear, so a full-map solution = the sum of two half map solutions.

Ah ok. That is good to know. I can't imagine that the farside solution would have much of an effect on the closed field at this scale, but I should probably double-check this.

Assuming you've set the far-side data to 0

Yes, this is what I've done.

dstansby commented 2 years ago

To close the loop on this, I am 👍 to adding an example, but not too motivated to do it myself, so PRs welcome 😄

wtbarnes commented 1 year ago

I think I'm convinced this is actually not a good workflow to demonstrate so I'm going to close this issue.