| |
 |
| Paper: |
Automation of VLASS Quick Look Image Quality Assurance |
| Volume: |
541, ADASS XXXIII |
| Page: |
87 |
| Authors: |
Trent Seelig; Sergio Garza; Aaron Lawson; Amy Kimball |
| DOI: |
10.26624/FRMB6694 |
| Abstract: |
In September 2017 the Very Large Array (VLA) began the first three
epochs of observations for the Very Large Array Sky Survey (VLASS). Each epoch
of the survey is split into two observing cycles with 6 cycles total to be completed
over 7 years. During each epoch the VLA will survey ∼80% of the sky with declination > -40° in full polarization between 2-4 GHz and generate 35,500 sets of products,
each covering∼1 square degree of sky. To ensure the survey meets its science goals a
Quality Assurance (QA) workflow was developed whereby each product was manually
inspected before being released to the community. However, this manual workflow has
been found to be prone to random human error and the pace depended on the efficiency
of those performing the QA. We have sought to decrease the time between observation and the delivery of the image product to the community and to standardize the
QA of each product by developing an automated QA workflow. In doing so we have
transcribed the manual QA ruleset for Quick Look (QL) image products into a python
code that employs heuristic methods and a neural network to identify image products
that contain unwanted artifacts. After applying our new automated QA workflow to
QL images produced during the first half of the third epoch of VLASS we present the
results of our automation. We show that compared to previous observing cycles we
have significantly increased the efficiency of QL image QA through our automation by
decreasing delivery time to the community as well as other overhead costs of manually
performing QA. |
|
|
|
|
|
