PNNL

The Science

Floods and droughts in East Asia, influenced largely by monsoon precipitation, affect billions of people. Atmosphere-only simulations generally poorly represent East Asian monsoon precipitation, while coupled climate models that include interactions between the atmosphere and ocean generally perform better. To improve atmosphere-only simulations, it is important to understand how much of their performance is determined by the lack of air–sea coupling versus limitations of the atmosphere models themselves. Here, researchers at the U.S. Department of Energy’s Pacific Northwest National Laboratory (PNNL) analyzed simulations from 18 pairs of atmosphere-only and corresponding coupled simulations from the Coupled Model Intercomparison Project phase 5 (CMIP5). They showed that improved monsoon precipitation in coupled simulations relative to atmosphere-only simulations is largely due to compensation of precipitation biases induced by sea surface temperature biases that originate from the biased atmospheric forcing evident in the atmosphere-only simulations. This finding points to the need to improve atmosphere models to refine simulations of East Asian monsoon precipitation.

The Impact

Modeling East Asian monsoon precipitation has been a longstanding challenge, as evident from the significant systematic errors, or biases, in generations of climate models. This study shows that models with larger atmosphere model errors benefit more from coupling, and vice versa. Although the error compensation between precipitation and sea surface temperature biases improves the representation of monsoon precipitation in coupled simulations, it reduces confidence in the ability of the coupled models to simulate the response of the Earth system to human perturbations. This study shows that improving the atmosphere models is key to simultaneously improving the simulations of East Asian monsoon precipitation and sea surface temperature, thus improving model credibility for projecting future changes in monsoon precipitation.

Summary

Errors in atmosphere-only simulations may be related to both random atmospheric internal variability and deterministic model biases. Coupling atmosphere models with ocean models weakens the atmospheric internal variability through negative atmosphere–ocean feedbacks, which may partly explain the improved skill in coupled models. If random internal variability is the main reason for the lower skill of atmosphere-only models, then sea surface temperature biases in coupled simulations should have a weak relationship with biases in the atmospheric variables in atmosphere-only simulations. In a study led by scientists from PNNL and Nanjing University, researchers analyzed simulations from 18 pairs of atmosphere-only and corresponding coupled simulations from CMIP5 to understand the difference in skill between these simulations for modeling East Asian monsoon precipitation. They revealed significant relationships between the sea surface temperature biases in the coupled simulations and the atmospheric biases in atmosphere-only simulations. These relationships demonstrated how monsoon precipitation biases in the coupled simulations are reduced through sea surface temperature biases that compensate for the deterministic precipitation biases evident in the atmosphere-only models. While the chain of processes involved in the relationships is local, some improvements representing in monsoon precipitation in the coupled models can also be achieved through remote atmospheric circulation effects.

Here, quantitative analyses demonstrate that models with larger atmospheric model errors benefit more from coupling, and vice versa. Therefore, reducing the biases in atmosphere models is important not only for improving monsoon precipitation in atmosphere-only models, but also for simultaneously improving monsoon precipitation and sea surface temperature predictions in coupled models; this is crucial for improving operational climate prediction and understanding future changes in monsoon precipitation.

PNNL Contacts

L. Ruby Leung, Pacific Northwest National Laboratory, Ruby.Leung@pnnl.gov

Funding

This study is jointly supported by the National Key R&D Program of China (Grant No. 2016YFA0602100) and National Natural Science Foundation of China (41675101). Y.Q., F.S., and L.R.L. were supported by the U.S. Department of Energy’s Office of Science as part of the Regional and Global Modeling and Analysis program area. The Pacific Northwest National Laboratory is operated for DOE by Battelle Memorial Institute under contract DE-AC05-76RL01830.