| |
 |
| Paper: |
Automated Anomaly Detection at Scale with the Cloud-Based Roman Data Monitoring Tool |
| Volume: |
541, ADASS XXXIII |
| Page: |
107 |
| Authors: |
O. Justin Otor; Jesse Averbukh; Tyler Desjardins; John Wu; Jonathan Hargis |
| DOI: |
10.26624/LFHJ6916 |
| Abstract: |
The Nancy Grace Roman Space Telescope (Roman) is an upcoming
NASA flagship mission that is planned for launch no later than 2027. Roman is primarily a survey mission with the Wide Field Instrument (WFI) as its main instrument. The
WFI consists of eighteen 16-megapixel detectors and is anticipated to produce 20 PB
of science data during its five-year primary mission, an order of magnitude more than
the current and planned yields of all active NASA flagship missions in astrophysics. If
the effort to inspect data by eye scales with the number of pixels, WFI-sized data may
dictate a full-time job’s worth of attention. Thus, we present the Roman Data Monitoring Tool, a scalable, cloud-based platform built to automate the anomaly detection
process for WFI science observations. Hosted on Amazon Web Services (AWS), our
proof of concept is an event-driven pipeline using AWS Lambda that judges the astrometric alignment of newly uploaded simulations against Gaia’s Data Release 3 and
logs results in a database. Future plans include building parallel pipelines to track other
phenomena of interest and a dashboard of all monitoring results to aid in triage when
follow-up support is necessary. |
|
|
|
|
|
