Details
-
Type:
Story
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: atmospec
-
Labels:None
-
Story Points:40
-
Epic Link:
-
Team:Data Release Production
Description
Write a hacky, zeroth order version of the pipeline so that we have some skeleton tasks and structures in place to iterate on.
This should probably be done on a user branch, but I wanted to have a ticket number for this work to save git hassles later on as this will touch more than just atmospec.
This will initially be pretty hacky in places and probably not adhere to all standards, but will be tidied up before review.
Checklist/to-do list/steps in Augustin's pipeline:
Prepare Image:
|
Load image from disk
|
Subtract overscan and trim & assemble
|
|
|
Initial ISR Step
|
Load the dark by filename
|
Scale the dark by exposure time and subtract
|
(you don not want to see the details of this!)
|
|
|
Source extraction (using SExtractor):
|
Find the objects and centroids
|
|
|
Load SExtractor catalog
|
|
|
Process catalog to find the primary object
|
return the x,y of the brightest, roundest object's centroid
|
|
|
Calculate the spectrum's bounding box, depending on dispersion direction and the order used
|
Dispersion direction is either just x or y, not arbitrary angles yet
|
Will need extending for arbitrary angles
|
|
|
Secondary ISR:
|
Load masks from disk if masking
|
removal of periodic pattern on chip
|
Create initial dispersion relation
|
Apply initial flatfielding
|
Find and interpolate over cosmics
|
|
|
Extraction (choice between three methods:)
|
Spectral extraction method 1:
|
Calculate row-wise mean and sigma of background and subtract
|
Row-wise, calculate:
|
aperture flux
|
PSF 1D-Gaussian flux
|
Moffat flux
|
Gauss-Moffat Flux
|
Voigt flux
|
|
|
Spectral extraction method 2:
|
Calculate row-wise sum for aperture flux without subtracting background)
|
Fit 1D Gaus for PSF flux (again, without bgSub, and without pedestal in fit function)
|
Same for Voigt fit (no bgSub here either)
|
|
|
Spectral extraction method 3 (two pass method):
|
1st pass: fixed aperture, no centroid, meaning that:
|
Per row:
|
Fit 1D Gaus for PSF flux (again, without bgSub, and without pedestal in fit function)
|
|
Between passes, fit the row-wise Gaussian's parameters with linear and quadratics functions describing the position and width of them as a function of row number i.e. position in the spectrum, respectively
|
2nd pass: applying a variable aperture, using the 1D centroid and 1D sigma provided, meaning that:
|
Per row:
|
Regular aperture flux summing (still without bgSub I think)
|
aperture is centered at the row's centroid, and 6 sigma around it
|
fit Gaussian over a very narrow range (120-180 pixels) up spectrum
|
|
|
Post-extraction:
|
|
|
Reflatfield
|
|
|
Find lines in spectrum
|
|
|
Match to known lines
|
|
|
Refine the dispersion relation calculation
|
Attachments
Issue Links
Activity
| Field | Original Value | New Value |
|---|---|---|
| Status | To Do [ 10001 ] | In Progress [ 3 ] |
| Status | In Progress [ 3 ] | In Progress [ 3 ] |
| Epic Link |
|
| Team | Data Release Production [ 10301 ] |
| Remote Link | This issue links to "Page (Confluence)" [ 19621 ] |
| Description |
Write a hacky, zeroth order version of the pipeline so that we have some skeleton tasks and structures in place to iterate on.
This should probably be done on a user branch, but I wanted to have a ticket number for this work to save git hassles later on as this will touch more than just atmospec. This will initially be pretty hacky in places and probably not adhere to all standards, but will be tidied up before review. |
Write a hacky, zeroth order version of the pipeline so that we have some skeleton tasks and structures in place to iterate on.
This should probably be done on a user branch, but I wanted to have a ticket number for this work to save git hassles later on as this will touch more than just atmospec. This will initially be pretty hacky in places and probably not adhere to all standards, but will be tidied up before review. Checklist/to-do list/steps in Augustin's pipeline: {noformat} Prepare Image: Load image from disk Subtract overscan and trim & assemble Initial ISR Step Load the dark by filename Scale the dark by exposure time and subtract (you don not want to see the details of this!) Source extraction (using SExtractor): Find the objects and centroids Load SExtractor catalog Process catalog to find the primary object return the x,y of the brightest, roundest object's centroid Calculate the spectrum's bounding box, depending on dispersion direction and the order used Dispersion direction is either just x or y, not arbitrary angles yet Will need extending for arbitrary angles Secondary ISR: Load masks from disk if masking removal of periodic pattern on chip Create initial dispersion relation Apply initial flatfielding Find and interpolate over cosmics Extraction (choice between three methods:) Spectral extraction method 1: Calculate row-wise mean and sigma of background and subtract Row-wise, calculate: aperture flux PSF 1D-Gaussian flux Moffat flux Gauss-Moffat Flux Voigt flux Spectral extraction method 2: Calculate row-wise sum for aperture flux without subtracting background) Fit 1D Gaus for PSF flux (again, without bgSub, and without pedestal in fit function) Same for Voigt fit (no bgSub here either) Spectral extraction method 3 (two pass method): 1st pass: fixed aperture, no centroid, meaning that: Per row: Fit 1D Gaus for PSF flux (again, without bgSub, and without pedestal in fit function) Between passes, fit the row-wise Gaussian's parameters with linear and quadratics functions describing the position and width of them as a function of row number i.e. position in the spectrum, respectively 2nd pass: applying a variable aperture, using the 1D centroid and 1D sigma provided, meaning that: Per row: Regular aperture flux summing (still without bgSub I think) aperture is centered at the row's centroid, and 6 sigma around it fit Gaussian over a very narrow range (120-180 pixels) up spectrum Post-extraction: Reflatfield Find lines in spectrum Match to known lines Refine the dispersion relation calculation{noformat} |
| Status | Admin Review [ 3 ] | In Progress [ 11605 ] |
| Remote Link | This issue links to "Page (Confluence)" [ 19752 ] |
| Status | Review [ 11605 ] | In Progress [ 3 ] |
| Resolution | Done [ 10000 ] | |
| Status | In Progress [ 3 ] | Done [ 10002 ] |
| Story Points | 20 | 30 |
| Story Points | 30 | 40 |
We discussed this at our meeting of 2018-12-10. We agreed that it was appropriate to have a “big bang” ticket for bootstrapping initial atmospec development, but that it'd be helpful to expand this ticket with a checklist of work that it includes.