I have been following this blog, off and on, for the past year or two. Many of the topics are fascinating, often with very useful references. I do have a question concerning the seasonal forecasts. Are the forecasted probabilities based on statistical modeling or dynamic numerical modeling methodologies? It seems that even a somewhat simple dynamic models could produce good generalized results (something a bit more complex than the first models using the so called “primitive equation” – it seems SSTs, the forecasted behavior of the MJO based on the current ENSO trends, etc. could all be incorporated into the dynamic model). After all, only generalized seasonal trends are needed, not specifics. If my understanding is correct, generalized temperatures trends would be easier to forecast than generalized precipitation amounts.