In 2006, Held and Soden published Robust Responses of the Hydrological Cycle to Global Warming, an analysis of climate simulations of the 21st century included in the 2007 IPCC report. Models predict that global water vapor will increase at a rate of 7.5%/K of global surface warming, consistent with the Clausius-Clapeyron relationship, which predicts how saturated water vapor increases with temperature. The approximation using surface temperature in the C-C equation instead of “local” temperature, that cools as you move up in the troposphere, was justified on the grounds that atmospheric moisture is concentrated near the surface.
Meanwhile, in a collaboration of scientists led by Zhengyu Liu of the University of Wisconsin-Madison and Bette Otto-Bliesner of the National Center for Atmospheric Research, a ground-breaking paleoclimate simulation was planned. Then grad student Feng He carried out the simulation of the last 22,000 years. It ran from the Last Glacial Maximum to the present, the first fully coupled, transient (read complex) paleoclimate simulation of this length. The research team considered the forcings of paleoclimate. The orbital cycles were known, the ice sheets documented, and greenhouse gas levels were obtained from Antarctic ice cores. But as the ice sheets melted, fresh water flowed into the ocean, becoming a climate driver as well. The light water served as a lid in the North Atlantic, weakening the meridional overturning circulation. This cooled the Northern Hemisphere, allowing the Southern Hemisphere to warm via a mechanism called the bipolar see-saw. Scientists didn’t really know how much ice melted from which geographic regions at which times. The meltwater from the Laurentide ice sheet flowed to the Arctic Ocean, the North Atlantic and the Gulf of Mexico. Feng conducted a series of branch experiments testing different pathways and quantities during the simulation. After each experiment, the scientists selected the branch that best represented the geological evidence and then Feng continued the simulation. The whole simulation took 1 ½ years, resulting in a Science paper (Liu et al., 2009) and Feng’s dissertation.
Working with professors Larissa Back and Zhengyu Liu, I attempted to test the Held and Soden results using this paleoclimate model data in place of the 21st-century model data. The professors thought that I would get a result close to Held and Soden’s water vapor increase of 7.5%/K. But for most of the paleoclimate run, water vapor only increased 4.2%/K. It only sped up to 6.7%/K over last century or so. Had the grad student made a mistake? Further investigation upheld the surprise finding.
During the rapid global warming of today, the Southern Ocean and its overlying atmosphere are in a state of disequilibrium. The Southern Ocean draws heat down from the atmosphere, so that the air is not permitted to warm much. During the slow warming of the deglaciation, a state of quasi-equilibrium was reached, so that the Southern Hemisphere atmosphere was allowed to heat up. You’ve heard of the Arctic amplification of the current global warming? Well, in the paleoclimate simulation, warming was bipolar. Relatively less warming occurred over the tropics. The warm, moist tropics have the most potential for adding water vapor to the atmosphere. If the tropics don’t warm much, then the overall global water vapor does not increase as much… only 4.2 vs. 7.5%/K. You can read our Journal of Climate paper (Back et al., 2013) and check out my plot below showing different spatial warming patterns for rapid vs. slow warming.
Figure 1: “Local” normalized temperature change patterns. Rapid-change (top) shows relatively greater change in tropics and less change over the Southern Ocean as compared to the slow-change (bottom) scenario. Temperature anomalies are normalized by globally-averaged temperature change.