Modeling the impact of Weather on Distance Traveled by Lost Persons

Abstract

Missing Persons cases are a race against time, where every minute is critical to save a life. The more information a Search and Rescue (SAR) team has to work with, the more likely the success of the search. dbS Productions created a Search and Rescue database with over 20,000 search and rescue cases across the world to assist rescuers in their SAR efforts. The database includes search-specific information such as location, eco-division, and limited weather information. It also includes personal data, including sex, age, clothing, and equipment, as well as various characterizations of the missing person, such as whether they are a hunter, a hiker, or have various medical conditions, such as dementia. All of these factors can be used to determine where a missing person may have headed while they were lost and try to locate them more efficiently. The primary goal of this research is to create a predictive model by augmenting existing spatial models implemented by dbS Productions with additional weather features, determining how weather conditions impact the distance traveled by lost persons, thus improving the efficiency of search and rescue operations. This process was established through regression modeling and other machine learning methods. Several models included in order to determine the effect of weather on the distance traveled, including regression models, models using support vector machines (SVM), and the most successful model using XGBoost. The results showed that there was a relationship between the distance traveled and the maximum temperature and the minimum temperature. Overall showing that the weather extremes have a significant impact on the distance traveled by lost persons.