Large-Scale Image Processing with the ROTSE Pipeline for Follow-Up of Gravitational Wave Events

Electromagnetic (EM) observations of gravitational-wave (GW) sources would bring unique insights that are not available from either channel alone. However, EM follow-up of GW events presents new challenges. GW events will have large-sky error regions on the order of 10-100 deg(2). Therefore, there is potential for contamination by EM transients unrelated to the GW event. Furthermore, the characteristics of possible EM counterparts are uncertain, making it desirable to assess the statistical significance of a candidate EM counterpart. Current image-processing pipelines are not usually optimized for large-scale processing. We have automated the ROTSE image analysis and supplemented it with a post-processing unit for candidate validation and classification. We also propose a simple ad hoc statistic for ranking candidates as more likely to be associated with the GW trigger. We demonstrate the performance of the automated pipeline and ranking statistic using archival ROTSE data. EM candidates from a randomly selected set of images are compared to a background estimated from the analysis of 102 additional sets of archival images. The pipeline's detection efficiency is computed empirically by re-analysis of the images after adding simulated optical transients that follow typical light curves for gamma-ray burst afterglows and kilonovae. The automated pipeline rejects most background events and has a similar or equal to 50% detection efficiency for transients up to the real limiting magnitude of the images. However, similar to 10% of the image sets show a residual background tail that impedes assigning a high significance to any putative candidate. This motivates the use of information beyond simple light curves for background rejection.