Implement Differential Rotation for Maps

[gbear605]

Slightly modified version of #1530 (see comments).

Nice things but not needed before pulling:

• Add the rotation time to the new map, maybe peek() has to be updated to show that time when plotted, maybe a new keyword in Map
• Work out how to update the coordinates mostly in the case of diff-rotating a not full-disk image

Tasks before pulling:

• Add unit test for _warp_sun

[Cadair]

Thanks @gbear605

I would like to see the test for warp_sun before we merge this. I will let @dpshelio and @wafels comment on the other things in the todo list.

@wafels from #1530 I was not suggesting we move this functionality into sunpy.coordinates just that we use those objects to specify things like vstart and vend.

[dpshelio]

Thanks @gbear605!!! This is what I love from github, people can continue the work I've left half-way. I've been trying to figure out why my attempts for not full-disk was not working... but I think it's to do with some columns/row, if I'm right I will be able to update this soon.

[wafels]

Not sure how this code will work, or should work, with maps that have masks.

[wafels]

Also, handling of NaN's might be better handled by doing a first pass on the data itself, and then a second pass on a map that has 1's where all the NaN's are. NaN's in the warped image are located where pixels are non-zero.

[wafels]

The _to_norm function has some odd behavior. If the minimum value of the input array is less than zero, then the output array does have a minimum value of zero. But if the minimum value of the input array is more than zero, the minimum value in the output array is arr.min() / arr.max(). Surely the minimum value should be zero to be consistent. Also, nanmin and nanmax functions should be used - these automatically ignore NaN's in the input.

[wafels]

If you have nans and infs in your map, a masked map is probably the best way to inform further code about what to do. This moves the handling of nans and infs on to numpy, and the user. In this case, there should be documentation in this code to explain to the user that masked maps are the best way forward. However, there should also be warnings in the docstrings to explain what might happen should nans and infs be passed through the code.

[Cadair]

@gbear605 are you going to have time to continue this? If not would you have time to pick it up @wafels?

[gbear605]

I am not intending to continue this, since I don't really have the time.

…

On Fri, Jan 6, 2017 at 06:31 Stuart Mumford ***@***.***> wrote: @gbear605 <https://github.com/gbear605> are you going to have time to continue this? If would you have time to pick it up @wafels <https://github.com/wafels>? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1876 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAqKHD3oVpcyOV8MQgpbPtgC22Sx94xRks5rPiYsgaJpZM4JieTg> .

[Cadair]

no problem @gbear605. Thanks for letting us know, and starting it off.

[wafels]

Sure, assign it to me.

…

On Fri, Jan 6, 2017 at 7:10 AM, Stuart Mumford ***@***.***> wrote: no problem @gbear605 <https://github.com/gbear605>. Thanks for letting us know, and starting it off. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1876 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA8CF0h8X3pvHcAuldIqT5-tDry7xG3kks5rPi8xgaJpZM4JieTg> .

[agneet42]

@Cadair if could any work be done upon this as this issue is related with the sunkit image project? If you could guide me through. Thanks.

[wafels]

Agneet, I have been working on this beyond what gbear605 was doing. If you want to pick it up, please take a look at my branch *https://github.com/wafels/sunpy/tree/ji_diffrot_fixes/sunpy <https://github.com/wafels/sunpy/tree/ji_diffrot_fixes/sunpy>* If you want to work on it, I can help you. Jack

…

On Mon, Mar 6, 2017 at 2:57 PM, Agneet Chatterjee ***@***.***> wrote: @Cadair <https://github.com/cadair> if could any work be done upon this as this issue is related with the sunkit image project? If you could guide me through. Thanks. — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#1876 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA8CF7U361WUfrZkZixX_4fmYErp4thcks5rjGU7gaJpZM4JieTg> .

[agneet42]

Wafels, yes I'd want to.
Could you help me out? Thank you.

[wafels]

Start with looking at my branch. The first thing to do is to change *https://github.com/wafels/sunpy/blob/ji_diffrot_fixes/sunpy/physics/differential_rotation.py#L100 <https://github.com/wafels/sunpy/blob/ji_diffrot_fixes/sunpy/physics/differential_rotation.py#L100>* to accept and return SunPy co-ordinates.

…

On Mon, Mar 6, 2017 at 3:18 PM, Agneet Chatterjee ***@***.***> wrote: Wafels, yes I'd want to. Could you help me out? Thank you. — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#1876 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA8CFz3-ZgdnfZTB-29KP2xMh5J4pWvsks5rjGoagaJpZM4JieTg> .

[agneet42]

@wafels I'm on it.

[nitinkgp23]

Hey @wafels , I have been pondering over this since days, and I had a few things to discuss with you.
First, why do we want to shift to using sunpy.coordinates? Is it just to maintain consistency, or is there any extra feature that we are trying to implement? As far as I could understand, the function takes inputs in plain astropy quantities and returns the same. Only 3 of astropy.coordinates classes are being used i.e. Angle , Latitude and Longitude, so are we only converting these to use sunpy.coordinates ?

Are we trying to implement support of any other frame, apart from helioprojective ?

What are we trying to achieve by converting it? Any extra implementation?

Is it going to help in some way to implement image warping?

[agneet42]

@wafels
Sir, if we take an input in the form of,

c = SkyCoord(xu.arcsec, yu.arcsec, frame='helioprojective')
then we can access x and y in solarpy.coordinates using c.Tx and c.Ty.
Would that be a plausible approach?

[agneet42]

@wafels could you check if what i suggested above would work to convert input and output data into sunpy coordinates?

[wafels]

You need to ensure that the input to rot_hpc is a sunpy-coordinate, and then convert it to heliographic coordinates.

…

On Tue, Mar 7, 2017 at 3:13 PM, Agneet Chatterjee ***@***.***> wrote: @wafels <https://github.com/wafels> could you check if what i suggested above would work to convert input and output data into sunpy coordinates? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1876 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA8CF5zZSU7vjmSwSVf3R4PxbErXfEo5ks5rjbpfgaJpZM4JieTg> .

[agneet42]

@wafels I am working on this. Could we discuss this and MGN as part of the sunkit-image project, whenever you have the time to?

[stale]

This pull request has been automatically marked as stale because it has not had any activity for the past five months. It will be closed if no further activity occurs. If the ideas in this pull request are still worth implementing, please make sure you open an issue to keep track of that idea!

[prateekiiest]

@wafels, anything that I can start working on from here ?

I had this one query regarding the input for solar_rotate_coordinate in the current code.
It takes SkyCoord as input and since its under astropy.coordinates.

So by using sunpy.coordinates only do we mean some thing like using this as an input

c = Helioprojective(-570*u.arcsec, 120*u.arcsec, obstime=obstime, observer=get_earth(obstime))
solar_rotate_coordinate(c, '2010-09-10 13:34:56')

instead of using the SkyCoord for the same input c = SkyCoord(-570*u.arcsec, 120*u.arcsec, obstime=obstime, observer=get_earth(obstime), frame=frames.Helioprojective) ?
and then solar_rotate_coordinate(c, '2010-09-10 13:34:56')

[nabobalis]

Is the issue for sunpy.coordinates still open?

[wafels]

I don't quite understand what the issue regarding sunpy.coordinates is. What I can say is that the function *solar_rotate_coordinate* https://github.com/sunpy/sunpy/blob/master/sunpy/physics/differential_rotation.py#L98 is intended to accept SkyCoords only now, although it does not explicitly test for this. Is there a reason we would specify a position on the Sun in another way?

…

On Tue, Mar 27, 2018 at 4:58 AM, Nabil Freij ***@***.***> wrote: Is the issue for sunpy.coordinates still open? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#1876 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AA8CFxqzVlJcGgVHth3zdX7eGCMQINK_ks5tif9RgaJpZM4JieTg> .