Scientists have developed a new analytical technique for minimizing uncertainty in climate prediction models. Photo by NASA/UPI | License Photo
March 18 (UPI) -- By looking at which climate mechanisms and variables drive uncertainty among climate models, scientists have found a way to narrow the range of predictions.
Scientists and their models agree, the world is getting warmer and the climate is changing as a result of growing greenhouse gas emissions. But even the best models can't say exactly what the planet's climate systems will look like in 2100.
Even when they're fed the same datasets, climate models produce different ranges of possibilities. The uncertainty surrounding climate feedbacks, like the snow-albedo feedback, explains the discrepancies among model predictions.
The snow-albedo feedback suggests the planet's declining snow coverage will amplify warming. With less snow, the planet's surface will be darker and absorb more of the sun's energy. There are many similar feedback mechanisms that are poorly understood and cause uncertainty in climate models.
Researchers at the University of Exeter in England and UCLA realized they could use prediction differences, or emergent constraints, to their advantage. By comparing differences in climate model predictions, scientists were able to identify which current climate variables are responsible for the greatest amount of uncertainty in the future.
If scientists can tweak models to more closely replicate the current climate's most important variables, then they can shrink the range of future climate scenarios.
"Emergent constraints will help developers make models that better predict the future because they identify which observations they should get their model to replicate," Steve Klein, researcher at the Lawrence Livermore National Laboratory in California, said in a news release. "This is particularly valuable on the subject of clouds, for which it is not easy to know which of the many diverse aspects of the clouds we observe are relevant to their future evolution."
Scientists used their new analysis technique to improve snow-albedo models.
"We found that the seasonal variation in the amount of snow-cover was closely related to the strength of the snow albedo feedback in the future, across a wide-range of climate models," said UCLA professor Alex Hall. "As we have satellite measurements of snow-cover variations in the recent past, we can use these observations to select the most likely values of snow-albedo feedback across the models."
In their new paper on the novel analytical approach -- published Monday in the journal Nature Climate Change -- authors warned against misusing emergent constraints. The new technique isn't a cure-all.
Some level of uncertainty is inevitable, according to researchers. But by reducing that uncertainty, scientists can help policy makers craft more specific climate change mitigation plans.
"An enormous amount of effort has gone into developing climate models by research groups around the world. Unfortunately, there remain significant differences between their projections," said Chris Huntingford, ecology professor at Exeter. "This uncertainty has to be reduced to help policymakers plan. At present, the only game in town to aid uncertainty removal is that of emergent constraints."