NWS/NOAA climate prediction services: evolution and new directions

A revolution in NOAA operational climate prediction took place in 1997 in the process of preparation of long-lead U. S. seasonal forecasts for the 1997-98 El Nino episode. A largely linear-statistical approach, capable of producing just low-confidence predictions of seasonal averages with marginal skill only at small leads, was replaced by an entirely new approach. This mixed physical/statistical approach using experimental model prototypes and emphasizing specific modes of climate variability not only produced high-confidence predictions of seasonal means at long leads with record skill but also opened the door for long-lead forecasts of intraseasonal variability, i.e. predictions of risks of storm types and major weather events. New developments leading to even higher skills and promising even the capability to make useful forecasts of the timing of storm activity up to a month or two in advance have since followed at an unprecedented pace. Both institutional changes and scientific advancements made the revolution possible. The latter represented milestones in several research directions reinvigorated in the wake of the highly-publicized El Nino of 1982-83. These directions will be described along with the new developments currently altering operational climate prediction and the technological advances necessary to fully realize new product potential. A brief summary will also be provided of the institutional initiatives to facilitate new product development and to provide users with the products they need in a form they can use.