How Many Kilonovae Can Be Found in Past, Present, and Future Survey Datasets?

TL;DR

This paper estimates the number of kilonovae detectable in past, present, and future surveys using simulations based on known models, highlighting the potential for future discoveries and the importance of optimized search strategies.

Contribution

It provides the first comprehensive estimates of kilonova detections across multiple surveys without gravitational-wave triggers, using realistic spectral models and BNS rate assumptions.

Findings

Past surveys likely found fewer than 0.3 KNe.

Future surveys like LSST and WFIRST could detect dozens of KNe.

Contamination from supernovae is manageable with current methods.

Abstract

The discovery of a kilonova (KN) associated with the Advanced LIGO (aLIGO)/Virgo event GW170817 opens up new avenues of multi-messenger astrophysics. Here, using realistic simulations, we provide estimates of the number of KNe that could be found in data from past, present and future surveys without a gravitational-wave trigger. For the simulation, we construct a spectral time-series model based on the DES-GW multi-band light-curve from the single known KN event, and we use an average of BNS rates from past studies of $10^3 \rm{Gpc}^{-3}/\rm{year}$, consistent with the $1$ event found so far. Examining past and current datasets from transient surveys, the number of KNe we expect to find for ASAS-SN, SDSS, PS1, SNLS, DES, and SMT is between 0 and $0.3$. We predict the number of detections per future survey to be: 8.3 from ATLAS, 10.6 from ZTF, 5.5/69 from LSST (the Deep Drilling / Wide Fast Deep), and 16.0 from WFIRST. The maximum redshift of KNe discovered for each survey is z = 0.8 for WFIRST, z = 0.25 for LSST and z = 0.04 for ZTF and ATLAS. For the LSST survey, we also provide contamination estimates from Type Ia and Core-collapse supernovae: after light-curve and template-matching requirements, we estimate a background of just 2 events. More broadly, we stress that future transient surveys should consider how to optimize their search strategies to improve their detection efficiency, and to consider similar analyses for GW follow-up programs.