Atmosphere Ocean Science Colloquium

Understanding climate change past, present, and future: new methods for signal detection and attribution

Speaker: Kate Marvel, Columbia/GISS

Location: Online

Date: Wednesday, February 9, 2022, 3:30 p.m.


What does climate change mean, what’s causing it, and how bad is it going to get?  Human activities, notably emissions of greenhouse gases and aerosols, have had a detectable influence on the global mean temperature.  But the detection of external influences on other variables is complicated by several factors: the brevity of high-quality observational records, large internal variability, and errors in the computer models used to estimate natural climate variability and future climate trajectories.  Moreover, climate models project a wide range of future temperature rises in response to the same changes in atmospheric carbon dioxide, and attempts to constrain this "climate sensitivity" using recent observations have proved challenging.   In this talk, I’ll explain how new methods can help to identify clear signals amidst the noise.  I'll discuss "smart fingerprinting" techniques that exploit known physics of the climate system to differentiate between the responses to different anthropogenic forcings and internal variability, and I'll highlight methods that use paleoclimate data to contextualize current changes.  Finally, I will demonstrate how Bayesian approaches can help quantify uncertainty, evaluate simple and complex climate models, and constrain projections of future climate.  Along the way, I will highlight the real-world conclusions enabled by these techniques: that humans are already affecting global and regional rainfall patterns, that the signal of human influence on global drought conditions emerged earlier than previously thought, and that while the worst-case scenarios for high climate sensitivity now appear unrealistic, the best-case scenario of extremely low climate sensitivity is now likely ruled out.