Extreme Weather and Climate Change
In late October, HUCE director Daniel Schrag met with three other Harvard climate scientists to discuss the relationship between weather anomalies and climate change, and considered how the general public’s concerns about extreme weather events might affect the scientific research agenda. The group included Zhiming Kuang, McKay professor of atmospheric and environmental science; Peter Huybers, professor of earth and planetary sciences; and Brian Farrell, Burden professor of meteorology.
Schrag: The United States population suddenly is thinking about climate change again, ironically because of a hurricane that may or may not have had anything to do with climate change.
Kuang: That’s right, there was a similar event like this in the past, when a hurricane was tracking with a low and caused a lot of damage.
Schrag: Late October hurricanes themselves are not unheard of, but there are two aspects of this hurricane that have me intrigued. One is that if you look at the data, sea surface temperatures off the coast of New Jersey down to the Carolinas averaged about four degrees warmer than usual. This hurricane strengthened as it traveled from North Carolina up to New Jersey, with 75 mile-an-hour winds increasing to about ninety mile-an-hour winds. That’s pretty unusual when a hurricane moves north along the Atlantic coast. Even more interesting is a point that Peter has made about steering.
Huybers: If you look at conditions 1,000 kilometers north of New Jersey during the summer and the fall, we’ve been losing sea ice on the Labrador Sea. We’ve been losing snow cover in Northeastern Canada. And there’s warming associated with that. And associated with the warming is the establishment of a high pressure system that’s been much more prevalent there during the last five years than during the prior thirty years, the period for which we have decent records.
This system, centered right over Labrador, is associated with a clockwise circulation, which decreases the average wind speed over New Jersey, resulting in an increased prevalence of winds going from east to west than had been observed previously. I think it’s clear that this is associated with Arctic warming.
Schrag: When you say prevalence, you don’t mean a small change in prevalence. Since 2007 you’ve calculated something on the order of a fivefold change.
Huybers: It really depends how you count storm systems, but essentially, there is a fivefold increase in storm systems going east to west, as opposed to west to east. Those numbers are still early.
Schrag: Now you say storm systems. Do you mean tropical storm systems?
Huybers: No—all storm systems. The number of tropical storms that are coming through is small.
Farrell: If you look just at low-pressure systems, comparing storm systems that are moving west to east to those moving east to west, Peter saw that, before 2007, something like six percent were moving east to west.
Huybers: Yes. And then by our count it went up to thirty percent during October and November because, apparently, there’s a routine block pattern [a blocking high, or large-scale pattern in the atmospheric pressure-field that can remain in place for days or weeks, effectively redirecting migratory hurricanes].
Schrag: In our community, beginning around 2005 when MIT professor of meteorology Kerry Emanuel linked rising ocean temperatures to increased tropical cyclone strength, the debate was defined around total energy dissipation. But energy dissipation measured during the season is actually not that interesting since most storms actually never hit land. If you had a decrease in storm intensity, but an increase in landfall, that would be much more troubling than the reverse, right?
Kuang: Right. The more recent work Kerry is doing addresses that point.
Huybers: There was another paper in 2010 that found that the loss of sea ice—and the associated pattern of warming—tended to generate a high-pressure system over the Labrador and Greenland region. That’s potentially another piece of evidence for this steering effect.
Kuang: Yes, but a similar type of storm did happen before.
Farrell: “The Perfect Storm” of 1991 was actually one of those.
Schrag: The question is, could we see meteorological conditions set up by the loss of sea ice that make this a much more common occurrence? It’s all about probabilities.
Kuang: Agreed. Do you think you can ascertain that from the data we already have, or are you talking about a modeling exercise?
Huybers: I think it’s quite solid. There are mean circulation pattern changes, so we can think about the likely consequences for tropical storms.
There are other factors that one has to take into account because the circulation anomalies are complicated. It’s not simply this one high-pressure system, but I think it’s worth looking into in more detail.
Schrag: Fifteen years ago, independent of storms and tracking, Kevin Trenberth [head of the Climate Analysis Section at the USA National Center for Atmospheric Research] was writing about changes in rain rate. And what’s interesting is we have now started to see a lot of unusual precipitation events. Did you hear what happened in the summer of 2012 at the Duluth Zoo? The polar bear and seals escaped because of incredible downpours. Seven to ten inches of rain fell in twenty-four hours, enabling the polar bear to swim out of its cage. Seals were crawling across the roads.
We have seen more and more extreme precipitation events, including thunderstorms in Maryland last July that shut down a huge swath of suburban neighborhoods, and people were out of power for weeks. Is something unusual happening?
Kuang: It’s hard to say—we don’t have a theory to predict that this was going to happen. This is an attempt to link events that occur to a cause.
Farrell: Here’s the problem: we all know that as climate scientists we don’t want to say anything about specific weather events. And yet the public is freaking out because of specific weather events.
As scientists, we are afraid of the topic because we can’t nail it down. And it’s a hard problem, but a year and half ago, we talked about tornadoes…
Schrag: Yes, we did, when all those tornadoes were happening. There’s almost no work on how climate change affects tornadoes—we talked about how crude that work was. Do you think the climate community is missing something? We know it’s a hard problem, but part of the problem is that there’s a communication disconnect, right?
Huybers: I think of George Lakoff [professor of cognitive science and linguistics at the University of California, Berkeley] making a distinction between systemic causation and specific causation. He points out that we don’t have a problem saying smoking causes cancer, but for any given person who gets cancer, you can’t really say, “Well, it’s the smoking that did it.” It just changes the odds. Yet we still use causative language when we talk about health.
For climate science you won’t find people talking in the equivalent language even if there is a statistical or probabilistic connection between events. I don’t know if that’s a cultural issue or if that’s because our problem is different because we have a much smaller number of samples to analyze.
Farrell: I wonder if the public is actually going to force us to work on this problem in a more serious way. What’s remarkable is how little work has been done.
Kuang: Fifteen to twenty years ago there was little work on climate change and hurricanes. That’s changed. So maybe going forward, things are going to change with respect to tornadoes and extreme weather events.
Farrell: I think that in the case of cancer and smoking there were plausible connections that had to do with mutations and DNA damage. So you could point to plausible connections that scientists could agree on.
In the case of increasing hurricane intensity, the theory has gone nowhere. Sometimes, when hurricanes pass over warm water, they intensify. Sometimes they don’t. So the theory is not good. In the case of tornadoes, there is no really good theory.
Schrag: We don’t know yet how to count them—is any extreme convection event a tornado? At least with cyclones we have data. We don’t even have data on tornadoes.
Farrell: Right. With tornadoes we do know that you need a lot of CAPE [convection available potential energy]. And it helps a lot if you have shear. Maybe we will end up saying something like this: if, when you raise the mean atmospheric temperature, this generates more shear, that might be conducive to more tornadoes. But the problem is, the science isn’t strong.
Schrag: The science isn’t strong, but we’re also a conservative community, so that when the science isn’t strong we like to assume that there is no connection, which is in some ways a miscommunication of the state of understanding, right?
Huybers: Maybe this conservative bent is a bad way of capturing the risk that might exist.
Schrag: Most researchers have stayed away even from thunderstorms: if you look at the literature on how thunderstorms will be affected by climate change, it’s miniscule. And yet it’s actually one of the ways that people are most affected.
Farrell: If you look at drought—the dust bowl—I think it is fairly well associated with climate change. Observed temperature anomalies of as little as 0.2 degrees can give you a dust bowl. It’s not well understood why, but the models tend to show it. So clearly you can get droughts of great magnitude with very small forcing. And those connections are straightforward. You don’t need large climate change.
Schrag: We tend to view extreme weather events like the heat wave in March or the heat wave in Russia in 2010 in a statistical context. We’re trained to think about things statistically, but in a system that’s changing there’s a question of how valid statistical approaches are.
Huybers: You just need to use non-stationary statistics. It’s just a matter of whether or not your statistical model is up to the task.
Farrell: You have statistics showing a correlation. And then you have the science trying to show why that is.
Schrag: I think the summer drought in the United States and the heat wave in Russia are reasons why we see a lot of scientists now starting to focus on soil moisture as a mechanism. That’s a step forward, but we need to talk more about specific mechanisms.
Kuang: Yes. I think that should be an area of a lot more official research. Mechanistic explanations provide a motivation for the theorists to focus on particular areas: for example, how the weakening of the jet stream can change the climate. That hypothesis may not be correct, but the theory provides some motivation for further study.
Huybers: It’s not surprising that mechanistic understanding is lagging behind the changes themselves because it’s an observation-driven science. On the whole, I would say we don’t predict phenomena so much as observe them and then search for explanations.There are a few instances where we actually predicted the phenomenon, and then went out and measured it. Usually it works the other way.
To the extent that we identify things that don’t fit into our current understanding of the climate system, that’s going to provide motivation for seeking the mechanisms.
Schrag: Do you think that is right?
Farrell: Well, Ed Lorenz [the MIT meteorologist famous for describing “the butterfly effect”] wrote in the first part of his book on the general circulation of the atmosphere that a person who attempts to explain the general circulation of the atmosphere without first observing it places himself at a considerable disadvantage. Essentially, if you were a desert island physicist you would not have conjured up most of what we see in the atmosphere—that is, ab initio [relying on basic and established laws of nature].
Schrag: The earth is such a complex system, that a simple theory is never going to predict the existence of most natural phenomena.
Huybers: I’m saying we should pay really close attention to things that seem out of the ordinary as a way of determining what our mechanistic explorations ought to focus on.
Schrag: Yes. Why did it get so hot in Russia? Why did it get so hot in the States for so long? And the answer is there must be something that was broken. And what was broken, probably, was that soils were so dry that they didn’t provide the evaporation that cooled everything off.
Farrell: That’s just been advanced as an explanation for the Midwestern floods too.
Schrag: Right. You see this also in the heat wave in Texas in 2011. If you look at a soil moisture map from that period, that area is dry as a bone. What about other types of mechanisms?
Huybers: Are you trying to connect this to climate change? Are you going to argue that it was dry because of climate change and it could cause a little change like that?
Schrag: I think we don’t know yet. But at least it suggests where we should begin to look. You can at least ask, How would climate change affect soil moisture? Because if you don’t ask that question you’re not going to guess how climate change might cause heat waves. You might suggest that with more greenhouse gases in the atmosphere you’re going to get more heat waves because it gets warmer on average. And you’d be wrong. There’s actually an amplification of heat because of the soil moisture feedback, but if you didn’t ask the right question you might not recognize the link.
Kuang: So how would one go about this? This probably provides motivation for particular studies on dynamics focusing on how changes in the new state are affecting the extreme. But I think from the data alone it’s hard to draw a causal conclusion.
Schrag: But you might look at things that you wouldn’t have looked at otherwise. For example, in Russia, you might look at the snowpack and notice that the snowpack is reduced. Intuitively, that wouldn’t be the first thing you’d look at. You’d say, Why is there a heat wave in Russia? Because there’s too little snow in the winter? That doesn’t make a lot of sense until you think about a mechanism, right? The snow cover in the Western U.S. was minisculein early 2012. And the following summer there was an incredible drought and heat wave.
Huybers: Another way to phrase this is that this offers an opportunity to test our models. In some of the models we’re not able to get the heat wave that was observed in Russia. That tells you that probably something is missing. Even when simulating a blocking pattern there, they seemed not to be able to model the actual excessive heat. That may be because the model didn’t correctly account for soil moisture. So that’s a nice falsification of the model that suggests, okay, we need to include more factors that adequately represent what’s likely to happen in the future.
Schrag: I am intrigued by the idea that public concerns are going to start steering our scientific community a little bit.
Farrell: That’s already happened! Our entire focus is on climate because of public concern. And the 1982 El Niño and the conditions in South America and the Southern U.S. set an entire research agenda.
Huybers: Your point is that fluctuations in the environment that have consequences for society have garnered a lot of attention. And in the case of El Niño it’s a natural event. Now we’re seeing some extreme events that may have a connection with climate change.
Farrell: Hurricanes are the tip of the iceberg. The heat wave in France in 2003 instigated a number of efforts. I think that the research agenda does respond to extreme weather events quite strongly. But if this winter there’s a lot of snow in the west and next year we have a cool summer, do you think everybody will forget about this?
Kuang: Well, until the next big event happens, I guess.
Huybers: We won’t forget about it.