Climate Change and Extreme Weather: 32 Takes on 28 Events from 2014

The science of deciphering how much long-term climate change influences shorter-term weather and climate events continues to blossom. On Thursday, the American Meteorological Society (AMS) released its fourth annual special issue of the Bulletin of the AMS devoted to these attribution studies. Launched in 2012 as an experiment, the project hit a nerve: researchers and the public were both intensely interested in the connection between human-produced greenhouse gases and high-profile, high-impact weather. This year’s batch of studies, which focuses on events from 2014, is the largest yet: a total of 32, including more than 100 researchers from 20 countries looking at 28 extreme weather and climate events from all seven continents. New topics this year include tropical cyclones, forest fires, and anomalies in sea surface temperature and sea level pressure. For about half of the events studied in this year’s AMS report, scientists found that human-induced climate change played a measurable role in making the event stronger and/or more likely.


Figure 1. Icons denoting the locations and nature of the 28 events examined in the 2015 special issue of BAMS devoted to climate change attribution for extreme events from the year 2014. ETS = extratropical storms; SST = sea surface temperature; MSLP = mean sea level pressure. Image credit: NOAA/NCEI.


Each year the project editors invite researchers from around the world to choose a particular event and examine it through the lens of climate change and its potential influence. Because the report is an open digest of sorts, it includes a variety of techniques in which particular events are put in a larger context using models and observations. Attribution science remains a young field, and there’s still room for experimentation.

A couple of U.S. highlights
There is far too much interesting science in this BAMS report to cover in a short blog post. For those most interested in U.S. weather, the report has plenty to chew on, including coverage of 2014’s burst of hurricane activity near Hawaii; the year’s drought-stoked wildfire season in California; the active winter storm track across much of North America in 2013-14; that winter’s Midwestern cold; and the chilly conditions that prevailed over much of the East. Strikingly, each of these events recurred to at least some extent in 2015. This is itself a fascinating phenomenon, one that goes unaddressed in the report--but that’s understandable, given that the report’s mandate is to focus on one year at a time. Below is some more detail on a couple of the U.S. events that I found particularly interesting.


Figure 2. Hurricane Iselle, with 90 mph winds, and Hurricane Julio, with 75 mph winds, steam west-northwest towards the Hawaiian Islands in this GOES-West image taken at 8 pm EDT Wednesday, August 6, 2014. Image credit: NASA/Goddard Space Flight Center.

Hawaiian hurricanes
Three tropical cyclones passed within 500 kilometers (300 miles) of the Hawaiian Islands in 2014: Iselle, Julio, and Ana. A research team led by Hiroyuki Murakami (NOAA Geophysical Fluid Dynamics Laboratory) used GFDL’s Forecast-oriented Low Ocean Resolution model (FLOR) to simulate hundreds of hurricane seasons with and without the influence of human-produced greenhouse gases. Even the amount of greenhouse gas already present by 1990 was enough to boost the odds threefold of getting at least one tropical storm or hurricane within 500 kilometers (300 miles) of the Hawaii coastline, compared to the simulations with preindustrial (year-1860) levels of greenhouse gas. One would not necessarily have expected a big year for tropical cyclones near Hawaii in 2014, given the unfavorable values of Pacific and Atlantic oscillations (IPO, AMO, and PDO) combined with a moderately favorable El Niño state. However, the years in which this blend of factors showed up in the greenhouse-boosted simulations produced a 340% greater chance of activity near Hawaii. “It is possible that global warming increased the odds of the extremely large number of Hawaiian TCs in 2014, in combination with the moderately favorable condition of El Niño,” write the authors. They add: “The ensemble experiments with FLOR indicate a continued increasing probability of active seasons around Hawaii over the next few decades...although there will be substantial modulation on interannual and decadal time scales from internal variability.”


Figure 3. The Las Pulgas Fire lights the night on May 16, 2014 at Camp Pendleton, California. The fire was one of three large wildfires in San Diego County that scorched more than 26,000 acres. Image credit: David McNew/Getty Images.

California fire risk
A group led by Jin-Ho Yoon (Pacific Northwest National Laboratory) studied 2014’s highly destructive Western US. fire season. Their results pointed in the same direction as several recent studies emphasizing the role of long-term warming in boosting drought impact and fire risk.The group drew on satellite analyses of burned area together with the Keetch-Byram Drought Index (KBDI), which incorporates both temperature and precipitation. It’s well known that decades of forest preservation coupled with population growth have helped bump up fire risk throughout the West. This study looks directly at how a warming climate affects fire risk by using the Community Earth System Model (CESM1) to directly calculate the KBDI and the probabilities of wildfire during preindustrial, recent, and future climates. Natural climate variation will make some years more fire-friendly than others, but this study found that climate change will push the overall threat beyond the range of natural variability over the next decade or so, as measured in three ways: the annual average KBDI, the amount of land under extreme fire risk, and the number of extreme fire-danger days per year. Interestingly, this is all despite the fact that the CESM1 produces a somewhat wetter California climate later this century (in line with several other recent climate projections). There’s no telling whether any given high-risk year will actually produce catastrophic fire: that depends on day-to-day weather, as well as factors such as the aggressiveness of firefighting, the presence or absence of arsonists, and sheer luck. The authors weren’t able to rule out natural variability as a factor in 2014’s highly destructive fire season. However, they did conclude that “man-made global warming is likely one of the causes that will exacerbate the areal extent and frequency of extreme fire risk.”

Cold in the Midwest and East
A pair of studies looked at the frigid winter of 2013-14 and found no evidence that human-produced climate change was responsible. Both studies concluded that, if anything, there seems to be less rather than more variability in winter temperatures as time goes by. The Midwest team, led by Klaus Wolter (CIRES/University of Colorado Boulder), focused on a broad area they call the “greater Upper Midwest,” or GUM, that extends from the eastern Dakotas to New York and Pennsylvania. Using a variety of simulations from 30 climate models, the group focused on the odds of getting a winter as cold as 2013-14 in today’s climate vs. the colder climate that was in place in the late 1800s, when human-produced greenhouse gases were far less prevalent. A winter like 2013-14 would have been expected about once per decade in the late 19th century, but in today’s climate, such a winter is now “extraordinarily unlikely,” as the authors put it--perhaps only once in every few hundred years. I asked Wolter about the recurrence of intense cold and snow over the eastern GUM region in 2014-15, which seems to fly in the face of the long odds found in his study. Wolter noted that snow cover is a big factor, since early snowfall helps set the stage for a cold winter. “The models are telling us that snow cover should be declining so rapidly that this kind of cold winter is on its way out. But the observations are giving us either no trend or even an upward trend in early winter snow cover.”


Figure 4. Snowfall plasters the streets of New York on February 13, 2014, part of a winter storm that knocked out power to more than a million homes and business across the eastern U.S. Image credit: Spencer Platt/Getty Images.


The study of Eastern cold, led by Laurie Trenary (George Mason University/COLA), focused on whether increased variability from 1950 to 2014 could explain the winter chill of 2013-14 in three regions: north-Atlantic (Pennsylvania northward), mid-Atlantic (from the Mason-Dixon line to North Carolina), and south Atlantic (Alabama, Georgia, South Carolina, and Florida). The team used several diverse measures in each region, including the coldest single January-to-March reading in each year, the variability of daily temperature within each winter, and the number of days per winter that fall into the coldest 10% in the long-term climatology. The winter of 2013-14 stands out dramatically in that final measure, ranking above any other year on record in all three regions. Otherwise, though, the findings are consistent: “the variability of winter daily temperature, and therefore of the range of realized temperature, has been decreasing for the past six decades.” Another wrinkle is that no significant change in the variability showed up when the period 1950-2014 was simulated by 12 leading global climate models as part of the Climate Model Intercomparison Project (CMIP). However, the observed drop in variability is still within the confidence interval of all models for the north and mid-Atlantic and about half the models for the south Atlantic.

The authors present their findings as an alternative to the “Francis hypothesis”--the concept put forth by Jennifer Francis (Rutgers University) and colleagues that a weaker, more meandering jet stream attributed to polar warming and reduced Arctic sea ice is leading to greater extremes and more “stuck” weather patterns. If anything, these authors suggest, high-latitude effects might be tamping down extremes rather than goosing them. “The decrease in variance is a plausible consequence of polar amplification of global warming, since a decrease in the pole-to-equator temperature gradient reduces the strength of fluid dynamical instabilities,” they write. As with the Midwestern study, this one does not address the recurrence of severe cold during the winter of 2014-15, but Trenary told me that NOAA/NCEP reanalysis data into 2015 maintains the decreasing long-term trend in winter temperature variability over the study area. “There are a number of open research questions,” she added. “For example, it would be interesting to look at the spatial changes in wintertime temperature variability across the US and see if there is any relation between the jet-stream variability and spatial characteristics of wintertime temperature variability. Fundamentally, it’s important to understand why these observed changes in wintertime temperature variability are occurring.”

Commentary: A work in progress
Climate change is a sprawling, long-term phenomenon that doesn’t always fit neatly into 365-day boxes. Its influence is obvious when you’re looking at slow-burn phenomena, such as the year-over-year rise in global sea levels or the gradual shift poleward in planting zones. The closer you get to a single weather/climate extreme, the more challenging it can be to connect the dots--and sometimes the dots lead you to preliminary, less-than-dramatic answers. Most of the temperature-related extremes in this year’s BAMS report were found to have a climate-change link of some type, while the connections with rain and snow extremes were less consistent and harder to ferret out. Drought, in particular, is a tough nut to crack, largely because it includes so many natural and human cofactors and because it can be defined in multiple ways.

In a press packet released on Thursday, lead editor Stephanie Herring (NOAA National Centers for Environmental Information) emphasized one crucial caveat: the absence of a clear climate-change influence in any particular study doesn’t prove that no influence exists. “Any of the following could explain the absence of a signal: there was no human influence on the event; the particular factors investigated were not influenced by human-caused climate change; [or] the human influence could not be identified with the scientific tools available today.”

Lurking in the background of any attribution study is the natural variation inherent in our weather and climate system. Researchers go to great lengths to separate any influence of climate change from the kinds of ups and downs one would expect in a climate that didn’t have ever-increasing greenhouse gases. This year, the states of New York and Vermont had their coldest January-to-March period since records began in 1895, with all of the Northeast in their top-ten coldest. Surely, the odds of getting such widespread, persistent, intense cold over this large an area in a warming climate must be phenomenally low--but sometimes, even very unlikely things happen. This paradoxical, high-impact event gets my vote for inclusion in next year’s BAMS attribution report. Stay tuned!

We’ll be keeping an eye on Cyclone Megh and on the potential for tropical development in the Gulf of Mexico and the Atlantic. See Jeff Masters’ post from earlier today for more on these systems.

Wunderblogger Steve Gregory has a new Friday afternoon post, Stormy Pattern As Arctic Cold Develops - Hints of Pattern Change.

Bob Henson