Forecasting for the In-Between
We can do a good job forecasting the weather for a week or two, and we can settle on what the climate is likely to do season to season, a month to a year into the future. But what about in-between?
Last week hundreds of scientists from around the world gathered online and at Lamont-Doherty Earth Observatory to talk about that gap, and what progress is being made in forecasting on what’s called the “sub-seasonal to seasonal” time scale.
The Workshop on Sub-seasonal to Seasonal Predictability of Extreme Weather and Climate was organized by Columbia’s International Research Institute for Climate and Society and Initiative on Extreme Weather and Climate, in conjunction with the WWRP/WCRP Sub-seasonal to Seasonal Prediction Project and NOAA’s Modeling, Analysis, and Prediction Program, the workshop took place over the course of two days full of diverse presentations and lively discussions.
The sub-seasonal to seasonal time frame, dubbed “S2S,” presents a significant challenge to scientists and forecasting centers because, as organizer and scientist Andrew Robertson put it, the S2S time frame has historically been a “predictability desert.” As anyone planning a vacation months in advance can appreciate, weather forecasts are unreliable more than a week or two into the future. Seasonal climate forecasts, on the other hand, are largely based on ocean conditions, and predict only averages over monthly or longer periods, rather than actual weather states.
The S2S time scale sits between these limits, and represents a gap in prediction skill which has, until recently, proven difficult to fill. In the last decade, though, useful predictions on this time scale have become possible, as phenomena such as the Madden-Julian oscillation, stratosphere-troposphere interactions, and others which influence weather on the time scale of two to four weeks have become better understood and better simulated in models.
Attendees at last week’s workshop were especially interested in “extreme” events, such as floods, droughts, heat waves, tornadoes or hurricanes. While hugely impactful on communities and companies, these extreme events remain inherently unpredictable and difficult to forecast. A diverse range of presenters demonstrated the progress being made as well as the pressing need for continued improvement on S2S prediction.
Erin Coughlan demonstrated how the Red Cross/Red Crescent makes use of S2S forecasts to anticipate and mitigate disasters, and explained how better S2S prediction would assist with budgeting and asset allocation. Michael Ventrice, from IBM’s The Weather Company, discussed the advantages S2S prediction offers to traders or clients in the energy industry looking to anticipate heat waves or cold snaps that drive demand and affect energy prices. Stefano Materia, from the Euro-Mediterranean Center on Climate Change, shared how a pair of companies in the agribusiness and water management sectors could benefit from better information predicting droughts, and discussed how to design forecasting products with policy-makers in mind.
In addition to private companies, other attendees affiliated with operational forecast and government centers, such as the European Centre for Medium-Range Weather Forecasts, the World Meteorological Organization, and the Australian Bureau of Meteorology, all brought unique perspectives. Also in attendance were research scientists from labs like NOAA’s Geophysical Fluid Dynamics Laboratory or NASA’s Jet Propulsion Laboratory, in addition to professors and academics from universities around the world.
The presentations addressed pragmatic considerations such as how to best construct and evaluate forecast systems; specific problems in atmospheric dynamics, such as how the stratosphere influences “atmospheric rivers,” giant plumes of water vapor which can cause floods on the U.S. West Coast; or broader climate science questions, such as where is the boundary between predictable “signal” and unpredictable “noise.”
The workshop organizers encouraged early-career scientists and graduate students such as myself to attend, and underscored the possibilities of and the necessity for the next generation of scientists to continue work on S2S issues. S2S research is an emerging area where there are ample opportunities to contribute to advancing scientific understanding, as well as the possibility to have meaningful impacts on the business, government and nonprofit sectors. The S2S time scale is a frontier in our efforts to gain understanding and improve prediction of weather and climate, and progress will naturally require collaboration among individuals from diverse communities and viewpoints.
The S2S project, now officially in its third year, has already advanced research significantly through the S2S database—an ongoing collection of forecasts and re-forecasts on S2S time scales up to around 60 days which are provided by modeling centers around the globe. The S2S database offers a relatively new and unprecedented resource for scientists, modeling groups and other interested parties to better study, understand and ultimately improve S2S prediction – in the same way that the Coupled Model Intercomparison Project archives have advanced climate change research—by making output from many models available publicly in a single place.
Several presenters, including one of the database’s main organizers, senior scientist Frederic Vitart from the European Centre for Medium-Range Weather Forecasts, demonstrated new or preliminary results from the S2S database, but it is only beginning to be exploited. Many participants are supported by the NOAA Modeling, Analysis, and Prediction Program for research projects using the database (including two projects at Columbia, one led by Suzana Camargo on tropical cyclones and one led by Shuguang Wang on the Madden-Julian oscillation) are still in the early phases of their work.
Zane Martin is a PhD student at the Fu Foundation School of Engineering and Applied Science, in the Department of Applied Physics and Applied Mathematics.