A core-collapse supernova releases the vast majority of the gravitational binding energy of its compact remnant in the form of neutrinos over an interval of a few tens of seconds. In the event of a core-collapse supernova within our galaxy, multiple current and future neutrino detectors would see a large burst in activity. Neutrinos escape a supernova hours before light does, so any prompt information about the supernova's direction that can be inferred via the neutrino signal will help to enable early electromagnetic observations of the supernova. While there are methods to determine the direction via intrinsic directionality of some neutrino-matter interaction channels, a complementary method which will reach maturity with the next generation of large neutrino detectors is the use of relative neutrino arrival times at different detectors around the globe. To evaluate this triangulation method for realistic detector configurations of the next few decades, we generate random supernova neutrino signals with realistic detector assumptions, and quantify the error in expected time delay between detections. We investigate a practical and robust method of estimating the time differences between burst detections, also correcting for detection efficiency bias. With this method, we determine the pointing precision of supernova neutrino triangulation as a function of supernova distance and location, detectors used, detector background level and neutrino mass ordering assumption. Under favorable conditions, the 1$σ$ supernova search area from triangulation could be reduced to a few percent of the sky. It should be possible to implement this method with low latency under realistic conditions.

12 pages, 17 figures