Multi-model biases for precipitation (mm day-1) for week 1 (left), week 4 (middle), and week 4 minus week 1(right) for re-forecasts initialized in Dec-Jan-Feb (top row), Mar-Apr-May (second row), Jun-JulAug (third row), and Sep-Oct-Nov (bottom row). Biases are calculated as model minus verification.
A well-known gap in prediction skill extending from two weeks to several months exists in our climate modeling systems. With the inability of climate models to include all local and mesoscale land and atmospheric processes, reliable subseasonal forecasts and predictions have been difficult to produce. This paper describes the Subseasonal eXperiment (SubX), an effort in which seven operational and research models coordinated to produce a suite of historical re-forecasts and prove weekly real-time forecasts, to assess S2S skill, and prospects for improvements in forecasts.
In a new American Meteorological Society journal, authors Emerson LaJoie, Ray Bell, Dughong Min, Yuejian Zhu, Wei Li, Erik Sinsky, Hong Guan, Emily Becker, Joseph Metzger, Neil P Baron, Deepthi Achuthavarier, Jelena Marshak, Randal D Koster, Normand Gagnon, Michael Bell, Stanley G Benjamin, Benjamin W Green, Rainer Bleck, Shan Sun, Kathy Pegion, Ben P Kirtman, Dan C Collins, Robert Burgman, Timothy DelSole, Jon Gottschalck, Hai Lin, Michael K Tippett, Andrew Robertson, many of whom were funded by the MAPP program working with interagency partners (DoD/Navy and NASA), analyze and evaluate model biases and skill of seven global models that have produced seventeen years of retrospective forecasts and more than a year of weekly real-time forecasts in real-time, on an operational schedule.
Evaluation of SubX model biases reveal that model bias patterns are established at week 1 and grow to week 4. Useful temperature and precipitation skill over the U.S. exists for week 3-4 predictions for specific regions and seasons. The SubX multimodel ensemble was found more skillful than any individual model overall, and the dataset allows comparison of tradeoffs between weather and climate/seasonal modeling prediction approaches. In addition, skill in simulating the Madden-Julian Oscillation and the North Atlantic Oscillation was also evaluated and found to be comparable to other subseasonal modeling systems.
The addition of research and operational models and availability of both real-time and retrospective forecasts in SubX provides a unique contribution to community efforts in subseasonal predictability and prediction. Both the re-forecasts and forecasts are archived at the Data Library of the International Research Institute for Climate and Society, located at Columbia University.
The Modeling, Analysis, Predictions, and Projections (MAPP) Program is a competitive research program in NOAA Research’s Climate Program Office. MAPP’s mission is to enhance the Nation’s and NOAA’s capability to understand, predict, and project variability and long-term changes in Earth’s system and mitigate human and economic impacts. To achieve its mission, MAPP supports foundational research, transition of research to applications, and engagement across other parts of NOAA, among partner agencies, and with the external research community. MAPP plays a crucial role in enabling national preparedness for extreme events like drought and longer-term climate changes. For more information, please visit www.cpo.noaa.gov/MAPP.