Another important factor that prevents a large use of long-range forecasts is that their economical value has not been demonstrated. It has been shown that the CMC seasonal forecast system has some skill, although modest, for the prediction of surface air temperature anomaly in Canada. However, the relation between skill and economical value is complex and reliance on skill alone may give a misleading impression of forecast value. In this paper, a simple cost-loss decision model will be used to study the relative economical value of CMC seasonal forecasts.
Using twenty-six years of historical data, the relative economical value of CMC deterministic and probabilistic long range forecast systems for forecasting three equi-probable categories for surface air temperature and precipitation is evaluated. Results show that there is clearly some value in the surface air temperature anomaly forecasts in each season for the two extreme categories. The deterministic system adds value to the probabilistic one in Winter only, although it benefits from an error correction process (called Best Linear Unbiased Estimator or BLUE). For the precipitation forecasts, it is found that the current system has little or no relative economical value in any season.