An agent-based model reveals lost person behavior based on data from wilderness search and rescue

Area of Science:

• Wilderness Search and Rescue
• Computational Modeling
• Human Behavior Analysis

Background:

• Thousands lost annually in US wilderness necessitate efficient Search and Rescue (SAR).
• Increasing search areas and decreasing survival rates pose challenges for SAR teams.
• Optimizing SAR requires understanding lost person behavior in diverse landscapes.

Purpose of the Study:

• To introduce a novel agent-based model for simulating lost person behavior.
• To integrate landscape dynamics and reorientation strategies into a predictive model.
• To validate the model using real-world Search and Rescue incident data.

Main Methods:

• Developed an agent-based model with agents moving on landscapes.
• Modeled agent behavior using a random variable selecting from six reorientation strategies.
• Simulated various behavior distributions and identified a best-fit profile for hikers.
• Validated the model using a leave-one-out analysis with the International Search and Rescue Incident Database.

Main Results:

• Identified a best-fit behavioral profile for lost hikers based on real SAR data.
• The model successfully simulates time-discrete lost person dynamics.
• Validation confirmed the model's accuracy in representing real-world scenarios.

Conclusions:

• This time-discrete model is the first of its kind validated with real SAR incident data.
• The model has significant potential to enhance current wilderness Search and Rescue strategies.
• Improved understanding of lost person behavior can lead to more effective and timely rescues.