LSSTDESC / SprintDayChicago2022

Repository to host Sprints at the Chicago July 2022 Collaboration Meeting
Creative Commons Zero v1.0 Universal
3 stars 0 forks source link

[SPRINT] A notebook for processing raw images using the Rubin Science Platform #18

Open mpwiesner opened 2 years ago

mpwiesner commented 2 years ago

A Notebook for processing raw images

I want to build a Jupyter notebook that can take a raw FITS image for a single chip from the focal plane and do all basic data reductions (flat fielding, bias correction, etc.) and complete object identification, deblending and measurement.

Contacts: Time: All day Main communication channel: GitHub repo:
In-person/Virtual/Hybrid: Hybrid Zoom room (if applicable): https://benu.zoom.us/j/3240135641?pwd=bnhjRFdXYUZFVUlhaVFCQlVZRWh6Zz09

Goals and deliverable

I would like a functional Jupyter notebook that can take a raw image and end with a calibrated exposure and a catalog of objects in the image.

Resources and skills needed

Understanding of the LSST Data Management systems, especially lsst.ip.isr.isrFunctions. Understanding how to use the DM Stack on the Rubin Science Platform. Python.

Detailed description

This is a similar notebook, but doesn't quite start from a raw single chip image: https://github.com/LSSTScienceCollaborations/StackClub/blob/master/ImageProcessing/BrighterFatterCorrection.ipynb

Results

We worked with Merlin and he advised us that processing raw PhoSim images from scratch will be prohibitively difficult, especially due to the cost of generating flats. We will try in the future to process our current images, but are going to continue this project through synthetic source injection into DC2. We expect to take our existing magnitudes based on Kasen predictions and the locations determined in DC2 galaxies and use those to inject the kilonovae directly into DC2 images. We still need to learn how to do that step by step, but hope to make progress during source injection and DIA sessions at PCW.

Thanks to @mfisherlevine, @DouglasLeeTucker and @arunkannawadi for advice!

For today, we tried increasing the magnitude of the kilonovae using PhoSim. The KN in this galaxy went from 25.24 to 19.0. These are raw PhoSim FITS images produced by @amnatasneem: ezgif com-gif-maker

Here is a comparison of the image we generated in PhoSim (galaxy of interest in blue circle) on left, compared to the DC2 image served from RSP DP0.2 (right; galaxy of interest in green): compare

mfisherlevine commented 2 years ago

It might well turn out that the sythetic source injection method can take postage stamps as input (I know literally nothing about it, other than that is exists and this is what it's for), so you might well find that this work can be transferred directly. I'd suggest joining the #dm-source-injection channel and asking in there.