Consensus Report

Assessment of Intraseasonal to Interannual Climate Prediction and Predictability (2010)

Each report is produced by a committee of experts selected by the Academy to address a particular statement of task and is subject to a rigorous, independent peer review; while the reports represent views of the committee, they also are endorsed by the Academy. Learn more on our expert consensus reports.

Accurate forecasts of climate conditions over time periods of weeks to a few years—called intraseasonal to interannual timescales—can help people plan agricultural activities, mitigate drought, and manage energy resources. However, current forecast systems have limited ability on these timescales because models for such climate forecasts must take into account complex interactions among the ocean, atmosphere, and land surface, as well as processes that can be difficult to represent realistically. To improve the quality of intraseasonal to interannual forecasts, this report recommends the continued development of tools used in forecasting, and sets specific research goals for improving understanding of sources of predictability. In addition, the report also suggests best practices to improve methods of making and disseminating forecasts to make the information more accessible to decision-makers and researchers.

Key Messages

  • Broadening the metrics used to assess forecast quality would provide more information. No perfect metric exists to convey all the information about a forecast; therefore multiple metrics should be used.
  • Errors in the representation of physical processes in dynamical models should be investigated and corrected.
  • Establishing public archives of forecasts, comparisons of past forecasts to actual climate conditions, the measurements used to produce forecasts, and the details of forecasting models could enhance transparency. Archives of the inputs, outputs, and tools used in forecasts could help quantify and identify sources of forecast error, assist users as they interpret forecasts to their own needs, and document how forecasts are made using a variety of climate prediction tools.
  • Exploration of multi-model ensembles, tools that combine predictions from different climate models when making a forecast, should be continued. The development of standards and metrics could help to determine the optimal selection and weighting of ensemble members.
  • Increased collaboration between operational centers and the research community would encourage the exchange of ideas between the two cultures. For example, operational centers could hold workshops focused on forecast development and grant short-term appointments for visiting researchers.
  • Knowledge of the "sources of predictability," variables or processes in the atmosphere, ocean, or land surface that can influence climate in predictable ways on intraseasonal to interannual timescales, is critical to the quality of forecasts. Research efforts should be expanded to provide a better understanding of sources of predictability such as the Madden-Julian Oscillation, feedbacks between the ocean and atmosphere and between the land and atmosphere, and interactions between the stratosphere and lower layers of the atmosphere.
  • State-of-the-art data assimilation systems should be used by forecast centers, and the types of data used to make forecasts should be expanded.
  • Statistical techniques should be used to complement predictions from models. Statistical techniques, especially non-linear methods, could identify errors in dynamical models, sets of equations that describe the motions of the atmosphere and ocean. Statistical techniques could also be used to translate forecasts to more local scales, which can benefit decision-makers.