Modeling Human Movement in the Wilderness: Navigating Lost Person Behavior and Searcher Response

Abstract

Search and rescue (SAR) outcomes hinge on two behavioral uncertainties: how lost people move through complex landscapes and how ground teams search those landscapes. This dissertation develops a unified, data-driven view of both. First, we present a dynamic, agent-based model of lost person behavior that adapts empirically observed reorientation strategies (e.g., trail-following, staying put, backtracking), integrates map layers of terrain-specific features (e.g., roads, streams, powerline easements), and is validated against real incident data. Because navigation is so heavily-dependent on the landscape, we further extend the model to include more landscape-dependent behaviors (like following elevation contours), explicit spatial and temporal decision points, and direct comparisons to observed tracks. Through simulation of all possible behavior distributions, we find an average behavioral profile for a hiker in the wilderness that can be adapted to accommodate other categories of lost people in different landscapes and locations. Recognizing that the lost person is only part of the SAR operational equation, we then turn to the searchers themselves by quantifying human searcher movement and coordination using GPS logs from real missions. In analyzing elevation effects on speed, differences in search tactics, and coupled coordination within search teams, we gain a better understanding of searcher dynamics with actionable metrics that can be used to test and refine current SAR assumptions. The overarching contribution is an empirically grounded foundation that links lost person movement and searcher performance, bridging operational practice with formal search theory.