Abstract for Eric Cawi’s award:
Wilderness searches consume thousands of man-hours and millions of dollars per year. Timeliness is critical. After 24 hours, survivability drops by about 20%. Good probability maps could greatly speed intensive searches.
MapScore (mapscore.sarbayes.org) provides a web-based portal for scoring, comparing, and improving models of lost person behavior – or any 2D probability map where ground truth is known.
Researchers receive case data and upload maps which are then scored according to the actual find location.
Cases come from the International Search & Rescue Incident Database (ISRID). MapScore now uses 300 of the thousands of available ISRID cases.
Wilderness Search and Rescue (WiSAR) is decades behind maritime search, and there are no theoretically sound tactical decision aids, in part because it lacks good probability maps. MapScore aims to change that.
Right now three universities are testing and refining their models. We describe the system and present preliminary results. See below.
