The World in 2050_ Four Forces Shaping Civilization's Northern Future - Laurence C. Smith [58]
MacDonald points out that by the time Schwarzenegger declared a state of emergency in 2009, most of the southwestern United States was actually in its eighth year of drought, not third. “Arguably, we are now into the great Twenty-first Century Drought in western North America,” he mused to me. “Could we be in transition to a new climate state? Absolutely. Should we be worried? Absolutely.” His concerns are echoed by Richard Seager at Columbia University’s Lamont-Doherty Earth Observatory. In a widely read Science article,248 Seager and his colleagues showed consensus among sixteen climate models that projected greenhouse warming will drive the American Southwest toward a serious and sustained baking. Their result, of course, is dependent on the group of models analyzed, and the simulation is imperfect because today’s coarse-scale climate models don’t represent mountainous areas very well (e.g., the Rockies, which produce most of the region’s snowpack water). But if these model projections prove correct, then the drought conditions associated with the brief American Dust Bowl could conceivably become the region’s new climate within years to decades.
Risky Business
“Stationarity Is Dead,” announced another Science article in 2008, sending a cold shiver through the hearts of actuaries around the world.249 A hydrology dream team of Chris Milly, Bob Hirsch, Dennis Lettenmaier, Julio Betancourt, and others had just told them that the most fundamental assumption of their job description—reliable statistics—was starting to come apart.
Stationarity—the notion that natural phenomena fluctuate within a fixed envelope of uncertainty—is a bedrock principle of risk assessment. Stationarity makes the insurance industry work. It informs the engineering of our bridges, skyscrapers, and other critical infrastructure. It guides the planning and building codes in places prone to fires, flooding, hurricanes, and earthquakes.
Take river floods, for example. By continuously measuring water levels in a river for, say, twenty years, we can then use the stationarity assumption to calculate the statistical probability of rarer events, e.g., the “fifty-year flood,” “hundred-year flood,” “five-hundred-year flood,” and so on. This practice, while creating enormous misunderstanding with the public,250 has also made us safer. Hard statistics, rather than the whims of developers or mayors, are used to design bridges and for zoning. But flood prediction, and most other forms of natural-hazard risk assessment, rest on the core assumption that the statistics of past behavior will also apply in the future. That’s stationarity. Without it, all those risk calculations go straight out the window.
A growing body of research is showing that our old statistics are starting to break down. Climate change is not the sole culprit. Urbanization, changing agricultural practices, and quasi-regular climate oscillations like El Niño all influence the statistical probabilities of flooding. However, the dream team’s paper and others like it251 tell us that climate change is fundamentally altering the statistics of extreme floods and droughts, two things of enormous importance to humans. “In view of the magnitude and ubiquity of the hydroclimatic change apparently now under way,” they wrote, “we assert that stationarity is dead and should no longer serve as a central, default assumption in water-resource risk assessment and planning. Finding a suitable successor is crucial for human adaptation to changing climate.”252
Unfortunately, we have no good replacement for stationary statistics yet, certainly nothing that works as well as they once did. Moreover,