Prediction of Near-term Climate is Nearly There

What will the climate be like in the Mediterranean six years from now? In southeast Asia four years from now? In an era of rising sea level, superstorms, extreme temperature events, and recurrent drought these are not questions of mere academic interest. For four decades a consortium of scientists called the World Climate Research Program (WCRP) has been working toward providing answers to questions like these.

The product they have been developing recently is called near-term climate prediction (NTCP) and the method of development is the coupled-model inter-comparison project (CMIP). Over the 40 years since its inception, the WCRP has been able to advance its work through continual tuning of its models—they are now beginning CMIP6—but they have also benefited from growing computer power and more and better observational data. Soon—via an initiative called The Grand Challenge—they will be able to tell us what the next 10 years will be like.

“The NTCPs are still not operational,” said Yochanan Kushnir, a Lamont Research Professor at Lamont-Doherty Earth Observatory and the co-chair of the WCRP Grand Challenge on NTCP. “The seasonal predictions are also still in process.” The goal, he said, is to provide a product useful to a number of different users outside the academy, namely those in charge of re-insurance, agriculture, transportation, energy, and water resources.

Making Model Results Available to the Private and Public Sectors

In a January 2019 Nature Climate Change Perspective paper, “Toward operational predictions of the near-term climate,” Kushnir and 20 co-authors provide a progress report and describe the intentions of the WCRP.

While Kushnir works primarily on NTCP development, Peter deMenocal, director of the Center for Climate and Life and dean of science at Columbia University, is committed to bringing these emerging models to the business and government sectors. DeMenocal readily lists important end-users of NTCPs.

“They include private sector companies with large assets that are vulnerable to future climate risks, specifically, single-point failure assets—refineries, factories, infrastructure—in regions exposed to near-term storm surge and/or sea level risk,” he said. “Also, developing countries need this to prepare for near-term risks to their economy and population. Already, seasonal predictions—like those for El Niño/Southern Oscillations (ENSO) events—are improving climate resilience in ENSO-affected regions, especially in underdeveloped regions.”

While precedents exist for communicating the products of academic research into the private sector—e.g., pharmaceutical development—deMenocal and others are dedicated to making model results available in the private and public sectors. “Some NGOs specialize in translating scientific guidance from NTCP to agencies within impacted countries to improve adaptation and resilience,” deMenocal said. “At Columbia, Lisa Goddard, director of the International Research Institute for Climate and Society, leads a group that has built these bridges, mainly for ENSO climate impacts, but it can serve as a model for NTCP as well.”

“An emerging model is one adopted by Jupiter Intel and other companies,” he continued, “who determine which elements of NTCP are of interest, and hence of value, to large private-sector companies. Identifying this climate and its link to private-sector business-model risk is an emerging theme.”

A question remains about who might carry such information over the proverbial bridge. DeMenocal cites an advisory committee formed during the Obama administration in 2016 as a success story. The committee was disbanded by the Trump administration but reconstituted as the Science for Climate Action Network (SCAN) with funding from Columbia’s Earth Institute and other sources. SCAN uses the findings of the National Climate Assessment released every four years to help communities and business translate climate science into local action.

What’s new about NTCPs?

Traditional climate models provide multi-decadal to century-scale projections of climate at a particular time in the future. As Kushnir explains in the Nature Climate Change paper, the internal variations of these models while they are running are not expected to synchronize with the internal variations in the real-world climate. In other words, if you stopped the model in the middle of a run and looked at the proverbial picture of the world it was drawing, it would not correspond to a realistic climate prediction. The goal of the traditional model is to map the general future trajectory of the climate under the effect of the projected increase in greenhouse gas concentration.

In contrast, the NTCP is designed to be a reliable forecast of the actual evolution of both the externally-forced and internally-generated components of the climate system. Which is to say, when you look back over a CMIP near-term prediction run, different parts of its journey to the future correspond to predictions of actual climate states at several future dates along the way.

Kushnir describes the process of improving the NTCPs as one of honing the skill of the models. In the model world, time can be run both forward and backward so NTCPs can be tested by starting a model at a particular date and running it into the recent past. The predicted climate can then be compared to observations, what actually transpired in the oceans, atmosphere, and on land through several years.

The model results are correlated point by point to establish the relationship between predictions for temperature and precipitation in model space and time versus those environmental parameters as measured in the real world. If the correlation is high, then the model is skillful.

The Grand Challenge

The final challenge is to develop informal formats that allow non-scientists to use the NTCPs. WCRP has instituted the Grand Challenge to do this. The immediate initiative of the Grand Challenge is to begin issuing a multi-decadal forecast that will be of use to planners of economic strategies. They will update it annually, incorporating that year’s improvements in skill and intend this to be a transition toward providing reliable NTCPs.

“My personal view,” deMenocal said, “is that this science-to-marketplace lever is essential to real action on reducing emissions and mitigating climate risk. Activating this lever requires useful, actionable data from NTCP and mechanisms to link this to business risk.”