My job is to improve the scientific understanding of what causes climate change. I give a lot of public talks about this research. It’s an important part of my job. As a recipient of government grants, I have a societal responsibility to explain what my colleagues and I have learned about human effects on climate, and why others should pay attention to our findings.
I speak to a variety of different audiences, not just to my peers. I’ve talked to Rotary Clubs, religious organizations, schools and the U.S. Congress. I’ve presented to audiences skeptical of every aspect of my research: to members of Bohemian Grove in the beautiful California redwoods; to the Jonathan Club in Los Angeles and the Pacific Club in Orange County. I always seek respectful dialogue, irrespective of the audience.
The Q&A sessions after public lectures yield some common themes. One widely held misperception is that computer models of the climate system are never compared with observations of real-world climate change. Nothing could be further from the truth. Comparing models with observations is at the core of climate science.
Forty years ago, the comparison of simulated and observed changes in climate was relatively rudimentary. Things changed in the late 1970s and early 1980s. It became apparent that human-caused greenhouse gas increases were likely to lead to planetary-scale warming. It was also clear that computer models would provide critical tools for understanding the size and rate of the expected warming. These tools required rigorous and objective vetting.
The need for comprehensive evaluation of climate models led to the establishment of model intercomparison projects (MIPs). Larry Gates, generally credited with being the intellectual father of climate model intercomparisons, was a key architect of the first MIP—the Atmospheric Model Intercomparison Project, or AMIP, which began in the late 1980s. In this project, different numerical models of the atmosphere were fed the same observed changes in ocean surface temperature and sea ice. AMIP enabled diagnosticians to compare how climate variables like atmospheric temperature, rainfall and winds responded to a common “boundary condition” (the same underlying ocean). Comparisons were made across a dozen different models and between models and observations. It was the beginning of the benchmarking of climate model simulations and of the scientific crowdsourcing of model diagnosis.
Since AMIP, climate models have been tested in a variety of novel ways. There are individual tests of the ocean component, the sea-ice module, the carbon cycle and atmospheric chemistry. Analysts can study the response of virtually all the world’s climate models to specified increases in greenhouse gases, to massive volcanic eruptions or to the imposition of the boundary conditions that existed at the time of the last ice age.
These sensitivity tests yield useful information about model performance in a variety of different settings—the present day, the recent past, deep time and the near- and long-term future. The tests teach us about climate responses to changes in boundary conditions, about intrinsic natural variability and about key physical processes. We learn when the behavior predicted by a computer model matches reality and basic theory. We learn when simulations and observations differ, and we understand why they differ.
Since November 2016, we have also learned about President Trump and his administration. Both have undergone a variety of sensitivity tests. We’ve witnessed the president’s response to catastrophic natural disasters—to Maria, Florence and Michael. We’ve watched his response to death in Charlottesville, Virginia; at Parkland, Florida’s Stoneman Douglas High School; and in Tree of Life synagogue in Pittsburgh. We’ve seen his performance on the world stage, shoulder-to-shoulder with Vladimir Putin and Kim Jong Un. We’ve observed Trump’s response to the murder of a Saudi journalist, to Christine Blasey Ford’s testimony and to criticism from “enemies of the state.” We’ve seen Mr. Trump respond to the existential problem of human-caused climate change by initiating U.S. withdrawal from the Paris climate accord.
In climate model world and in the real world, sensitivity testing provides valuable information. We can reject or down-weight a climate model based on its test results. And we can exercise our constitutional right to vote in favor of decency, compassion for the less fortunate and the long-term viability of our planetary life-support system. Sensitivity testing for the current administration is over. Now it’s time to vote.