Climate models are powerful tools that scientists use to study how the climate system works today and how it will evolve under different scenarios of global warming. When models are updated with new scientific information, they should be evaluated to see how well they represent different climate characteristics, including weather patterns found in particular geographic regions.
A new study led by Graham Taylor, a doctoral student in Portland State’s Earth, Environment and Society program, and Paul Loikith, an associate professor of geography at PSU, tested how well climate models represent weather patterns at large scale in the Pacific Northwest. Researchers from Oregon State University and the Jet Propulsion Laboratory also contributed to the study, which was published in the journal Climatic dynamics.
“These complex computer models that simulate the Earth system can be thought of as virtual laboratories for experimenting with climate science,” says Loikith. “If models can’t reproduce important features of the observed climate, they won’t be very useful for studying future climate.”
Since all computer models have different strengths and weaknesses based on differences in physics, scientists often use the results of many different climate models to assess projections of future climate change. For this study, researchers used data from the sixth peak phase of the Coupled Model Intercomparison Project (CMIP6) to test how well 26 different climate models could simulate the range of large-scale atmospheric circulation patterns (such as wind and pressure) found over the Pacific Northwest. These patterns range from those associated with hot and dry weather, to cold and stormy and everything in between.
To test the models, the team used a machine learning technique called self-organizing maps to group the daily weather patterns simulated by the climate models into a set of 12 categories. They did the same for observed historical weather data. They then compared the two sets of data to see how well they lined up.
The researchers found that the climate models generally simulated observed wind and pressure patterns very well, and that the temperature and precipitation patterns created by the models closely matched the correct patterns found in historical data.
These results are important because they suggest that current climate models represent large-scale weather patterns in the Pacific Northwest reasonably well and can be used to better understand future climate change under continued global warming.
“These results strengthen our confidence in the ability of these models to help us better understand how the climate will change in the region and why these changes are happening,” says Loikith.