We present an empirical study of contamination in deep, rapid, and wide-field optical follow-up searches of GW sources from aLIGO. We utilize dedicated observations during four nights of imaging with DECam. Our search covered $sim56$ deg$^2$, with two visits per night separated by $approx 3$~hours, in $i$- and $z$-band, followed by an additional set of $griz$ images three weeks later to serve as reference images for subtraction, and for the purpose of identifying galaxy and stellar counterparts for any transient sources. We achieve $5sigma$ point-source limiting magnitudes of $i approx 23.5$ and $z approx 22.4$ mag in the coadded single-epoch images. We conduct a search for transient objects that can mimic the $i-z$ color behavior of both red ($i-z > 0.5$~mag) and blue ($i-z < 0$~mag) kilonova emission, finding 11 and 10 contaminants, respectively. Independent of color, we identify 48 transients of interest. Additionally, we leverage the rapid cadence of our observations to search for sources with characteristic timescales of $approx1$ day and $approx3$ hours, finding no potential contaminants. We assess the efficiency of our pipeline and search methodology with injected point sources, finding that we are 90% (60%) efficient when searching for red (blue) kilonova-like sources to a limiting magnitude of $i lesssim 22.5$ mag. Applying these efficiencies, we derive sky rates for kilonova contaminants in the red and blue regimes of $mathcal{R}_{rm red} approx 0.16$ deg$^{-2}$ and $mathcal{R}_{rm blue} approx 0.80$ deg$^{-2}$. The total contamination rate, independent of color, is $mathcal{R}_{rm all} approx 1.79$ deg$^{-2}$. We compare our derived results to optical follow-up searches of the GW events GW150914 and GW151226 and comment on the outlook for GW follow-up searches as additional GW detectors (e.g., KAGRA, LIGO India) come online in the next decade.
Download