The National Research Council weighed in on US climate modelling programmes Friday, calling for a more systematic methodology for developing, interpreting and comparing the full suite of climate and weather models.

The 250-odd-page report provides a primer on the challenges ahead, ranging from a shortage of young modellers entering the business to the need for a new category of “climate model interpreters” who can take the data and help convert it into something more useful for the world at large. But the core challenge going forward, the council reports, is how to “disrupt the inertia of the U.S. climate science enterprise” and promote a broader integration of efforts, on both scientific and technical fronts. As climate models become ever more detailed and precise, they could also be integrated with weather models.

The idea has been around for some time, but the council sees opportunity in the coming years as the modelling community makes a likely difficult transition toward larger supercomputing platforms with ever more processors working simultaneously to get the calculations done. This has implications for how the models are designed and run, and the question is whether modellers can standardize solutions, which might in turn help scientists address thorny problems that have plagued their efforts since the very beginning: Given that different models produce different answers for different parts of the climate system, how do we determine which parts of which models are right?

This is a particularly daunting question because climate models are so complex that the entire idea of reproducibility – a cornerstone of experimental research – is mostly a theoretical construct in this arena (for more on these challenges, see our recent feature on advances in modelling the atmosphere here and a prior feature talking about broader complexity in the models here).

Currently, modelling teams try to address these challenges by coordinating experiments through initiatives such as the World Climate Research Programme’s Coupled Model Intercomparison Project, which is at the center of the climate modelling effort for the Intergovernmental Panel on Climate Change’s ongoing fifth assessment. But even here, comparing results is difficult because models are a bit like children: though they may behave differently, it’s seldom clear why. But what if all of the models could speak to each other, and scientists could pull the atmosphere or the ocean out of one and simply plug it into another?

“Having all of the Nation’s models buy into a common framework would allow this research to be systematized,” the council says, and “could hasten scientific progress significantly.”

As usual there are costs and institutional barriers, and the path forward remains unclear. The report nonetheless underscores a variety of “bottom-up” initiatives in which scientists coordinate work in order to maximize there own time. And here again, the council says, one tool for encouraging such activity is the development of a common infrastructure.