Nemo and the inner workings of climate change models

Photo by Robert Page, Jr.

Photo by Robe­spierre Paget.

After winter storm ‘Nemo’ dropped two feet of snow on us a couple weeks ago, it took the City of Boston two days to plow one of the roads leading to my home. The other was still buried beneath a thick, icy blanket for another day. The term “global warming” has been sit­ting on our lips for almost four decades, could it be the city no longer sees a need for plows in the age of cli­mate change? If so, it should take a look at a 2011 study by North­eastern pro­fessor Auroop Gan­guly and grad­uate stu­dent Evan Kodra.

Despite the overall warming trend in our cli­mate, the 2011 paper sug­gests that cold snaps like the ones we’ve seen this winter (but were almost entirely absent here in New Eng­land last winter) will per­sist through the end of the cen­tury. I under­stand that cli­mate is a long term ver­sion of the short term thing we call weather, but what I didn’t under­stand before an e-​​chat with Kodra was how exactly he and his col­leagues are able to make mean­ingful state­ments about it. Below is a recap of what I learned in our “conversation.”

A wide range of experts all over the world have devel­oped a series of cli­mate change models that people like Kodra and Gan­guly use to ana­lyze the impacts of var­ious cli­mate change sce­narios on the planet. The models are big com­pu­ta­tional algo­rithms of extreme com­plexity that inte­grate infor­ma­tion about things like atmos­pheric cir­cu­la­tion pat­terns, the shape of land masses and water­sheds around the globe, moun­tain and veg­e­ta­tion cov­erage, and existing obser­va­tional data, to name a few fac­tors. The models break up the globe into a mas­sive grid in which each parcel is treated as an indi­vidual unit, Kodra told me. “When the models run, they pro­duce dis­crete values over the globe, and vari­ables like tem­per­a­ture and rain­fall are output at each rec­tangle for each time step,” he said. “Often the assump­tion is that these vari­ables rep­re­sent the average weather hap­pening within that cell of the earth for that day.”

Photo courtesy of Evan Kodra.

Evan Kodra is a grad­uate stu­dent in the Col­lege of Engi­neering. Photo cour­tesy of Evan Kodra.

Eval­u­ating cli­mate models is a huge under­taking in its own right, said Kodra, whose work hap­pens to be focused, in part, on that very thing. Two basic approaches look at the “skill” and “con­sensus” of var­ious models, respec­tively. A “skillful” model is able to pre­dict his­tor­ical data with high accu­racy. For instance, a pre­dic­tion about the cli­mate in 2011, based on ear­lier data, would check well against actual 2011 data. The con­sensus approach assumes that when mul­tiple cli­mate models, each of which uses a totally dif­ferent set of data and algo­rithms, come up with the same results, those results are likely to be cor­rect. A third approach “has tried to bal­ance the two notions,” said Kodra. But, eval­u­ating models has turned out to be an incred­ibly dif­fi­cult task.

Addi­tion­ally, the models seem to break down at the very local scale. “This gap is thought to be related to micro-​​scale phys­ical processes that the com­mu­nity cannot rep­re­sent well phys­i­cally,” said Kodra. Also, the cells of the giant grid I men­tioned above seem to be too big right now to get really detailed infor­ma­tion about local envi­ron­ments. “Unfor­tu­nately, local deci­sion makers usu­ally need reli­able pre­dic­tions at these scales and find global pre­dic­tions, where sim­u­la­tions are more real­istic in the aggre­gate, less imme­di­ately rel­e­vant,” Kodra noted.

So where does this leave us? “It’s entirely pos­sible that we could chalk Nemo up to nat­ural vari­ability that we have seeing for so long in New Eng­land, and in fact this is the most plau­sible expla­na­tion,” Kodra told me. But here’s the catch: The nat­ural vari­ability we’re all so used to in this part of the country is not always masked by the overall warming trend. “Hence, extremely cold weather will still pop up in the future,” he said. Some researchers are sug­gesting that warming taking place in one part of the world trig­gers cooling events in other regions. “Our per­sonal view is that these remain hypotheses that need to be proven more rig­or­ously for wider sci­en­tific acceptability.”

And this is gen­er­ally where Gan­guly and Kodra are headed. Since 2011 they have been exploring more deeply the hot and cold extremes that mul­tiple cli­mate models pre­dict along with how those extremes will impact var­ious sec­tors of society, from public health to the energy grid, Kodra said. He is also using his exper­tise as a sta­tis­ti­cian to com­pare the large datasets that come from cli­mate models with large obser­va­tional datasets. “The end goal is to build a suc­cinct and reli­able descrip­tion of uncer­tainty in extremes that could even­tu­ally be helpful in deci­sion making.”

Bottom line? Don’t get rid of the plows! We’re going to need them for a long time to come. But that doesn’t, by any means, negate the overall warming trend.