Can Abrupt Seasonal Transitions be Predicted in Climate Forecasts?

Abstract:

There is on ongoing debate in the seasonal prediction community as to whether high frequency weather statistics in climate forecasts have any inherent predictability, and ultimately prediction skill. The North American Multi-Model Ensemble (NMME) seasonal-to-interannual prediction experiment is the ideal test-bed to evaluate the predictability and prediction of weather within climate. NMME is multi-institutional multi-agency system to improve operational monthly and seasonal forecasts based on the prediction systems developed at the major US climate modeling centers (NOAA/EMC, NOAA/GFDL, NCAR, NASA) and Canada. Although currently in an experimental stage, the NMME prediction system has been providing routine real-time monthly and seasonal forecasts since August 2011 that adhere to the CPC operational schedule. In addition to the monthly data, daily output from some of the retrospective forecasts are now being archived. Based on the NMME daily output this talk evaluates the predictability and prediction of two aspects of weather within climate: (i) monsoon onset in India and in South West North America and (ii) onset of spring severe weather in the mid-west US. The analysis estimates predictability by examining how well the individual models “predict” themselves and how well they “predict” other models. Prediction quality is assessed based on comparisons with observational estimates.