The Value of Seasonal Climate Forecasts and Why Water Managers Don't Use Them

A number of government agencies and academic institutions now disseminate seasonal climate forecasts based on the output of general circulation models or statistical models developed from historical data. While verification of these forecasts often shows that they have significant skill, adoption by water management agencies appears to be slow. Some possible reasons for this are a lack of understanding of probabilistic forecasts, no straightforward means of revising policies based on forecasts, or simply a lack of will to depart from current policies and practices. We propose another reason, which is that in many locations seasonal forecasts are not reliable enough to warrant changes in operating policies. Through a simple economic-optimization model, we show that forecast skill does not translate directly into forecast value, and in fact a surprisingly high level of skill may be needed to provide sufficient improvement in climatology-based operating rules. One implication is that the skill measures commonly used to assess forecast accuracy and reliability may be inappropriate for water management applications.