Hindcasts – A Crucial Component of Water Resources Forecast Services

Hindcasts are records of the forecasts that would have been made in the past using current forecast tools. They are absolutely essential if water resource managers are to make good use of real time forecasts.

The real-world performance of forecast based operating rules depends on the accuracy, precision, and bias of the forecast. But it depends just as much on the way in which the forecasts are actually used to make decisions – both real-time operating decisions and capital investment decisions. This means that the decision making process, especially operating rules for water management must be tuned to the forecasts that will be available.

Operating decision rules for water resource systems are generally tested by simulating the operation of the system using historical meteorological and hydrologic time series. If and only if hindcasts are available can decision rules that use forecasts be similarly tested. Just as important, it is not necessarily true that a statistical improvement (in meteorological terms) in forecast performance will lead to an improvement in water system performance when used with an existing forecasts based rule. In fact, a degredation in performance could occur in such a case. This makes it critical that when forecast procedures are changed, new hindcasts be produced and operating rules be re-tuned to the new forecasts.

This presentation will demonstrate the concepts presented above in light of real world examples, including the NYC water supply system.