Statistical technique seeks to reduce climate uncertainty

Three measures of the climate system—climate sensitivity, vertical ocean diffusivity, and sulfate aerosol forcing—underpin current understanding of the power of anthropogenic climate change. Climate sensitivity reflects the equilibrium temperature change that would occur given a doubling of atmospheric carbon dioxide, vertical ocean diffusivity affects the rate at which the ocean is able to redistribute heat, and sulfate aerosol forcing describes how anthropogenic sulfate aerosols affect the radiation budget. Other projections rest on these measures, such as changes in weather patterns or precipitation rates, with the derivative predictions sensitive to changes in the more fundamental properties. Given their importance, a key research effort revolves around minimizing the uncertainty in the portrayal of climate sensitivity, ocean diffusivity, and aerosol forcing in climate models. A conventional approach to doing so is to systematically vary climate model parameters in an attempt to minimize the discrepancy between model results and the observational record. The time and financial requirements of running large‐scale climate models, however, reduce this technique's viability. To overcome these barriers, Olson et al. refined an approach to statistically emulate a complex climate model, allowing them to simulate a massive number of model runs at a fraction of the cost.