The first of the four elements of the optimal search problem is having a probability density distribution (predicting the likelihood that an object is in any particular search area or region.)
To achieve this during a Maritime Search and Rescue Incident one takes into account the accuracy of the initial location report, the current, wind and so on. Computer models can then accurately map the likelihood of the boat (or whatever search object is being sought) being in any particular area.
However, the variables for a vulnerable missing person search are not yet known with any particular accuracy. They may choose any direction; may stay on paths or tracks, or depart from them; camp or find shelter; try to cross rivers; go uphill or down; and so on and so on.
Computer models of missing person behaviour then are not as useful or accurate as Maritime models. This does not, however, mean that we cannot come up with useful probability density distributions. Take a quick glance at Robert Koester’s, Lost Person Behavior book to see that the world SAR community has over 50,000 incidents’ data to draw upon.
From this we can predict the likelihood a given misper will travel a certain distance from their initial location and misper “type” or “category”. This is sufficient to draw a couple of circles on a map and calculate probability density’s for concentric regions on a map. It is a small step from this to calculate actual Probability of Areas (POAs) for specific search sectors.
Other potential methods for calculating probability density distributions include the consensus method – allowing for subjective calculation based upon search planners’ experience, the individual misper intelligence and the actual search terrain limitations.
