Using Ensemble Prediction Systems for Decision Support Activities

Cost effective decisions to mitigate loss due to meteorological phenomena require knowledge of the cost to take action and the loss if no action is taken. Once a cost-loss ratio is established for a particular event type, forecasts can be used to determine when action must be taken to mitigate loss. Ensemble prediction systems contain probabilistic and uncertainty information and are well suited to provide information to support such decision making activities.

Single deterministic forecasts can be used in decision making to mitigate losses. However, these forecasts do not provide probabilities of an event occurring nor do they provide a measure of uncertainty. Thus, critical information is missing in the decision making process when the decision is based on a single model.

Ensemble prediction systems provide important information which can be of value to decision makers. These data can reveal both the high probability of a significant event, such as heavy rain or snow, or a low probability event with potentially high impacts and large potential losses. The low probability forecast of a major hurricane making landfall near a major population center might be a good example of the latter. Failure to take action could be extremely costly. Decision makers need to be aware of the probability of an event and cost-loss ratios when making a decision to act or not to act.

This paper examines the use of ensemble prediction systems to aid in decision making. The focus is on how forecasters can leverage probabilistic and uncertainty information provided by ensemble prediction systems to assist in making effective decisions. Simple cost-loss models are presented. Clearly, there is a need to establish cost-loss models for both a wide range of weather events and a wide range decision making activities.