After another extreme winter in the eastern United States – symbolized by Boston’s historic snow blitz of 90 inches in just over three weeks – scientists, the media and the public are asking once again: Is climate change causing more extreme snowfall events?

The argument that climate change is leading to greater snowfalls is based on a very simple law of physics – a warmer atmosphere can hold more moisture. But on the other hand, snowfall is dependent on sub-freezing temperatures, so what does it mean for places like the eastern United States, where climate change might make it too warm for snowfall?

The best tools we have to answer such a complex question, which incorporate all the physics and competing factors, are the climate models. Put simply, climate models are similar to weather models but run on a much larger scale in both space and time. They take into account interactions between land, oceans and ice, and importantly, they attempt to predict changes in our climate as greenhouse gases increase in the atmosphere.

All of the climate models predict that mid-latitude storm tracks (like the winter storms we see in the United States) will shift toward the poles with climate change — north in the Northern Hemisphere, and south in the Southern Hemisphere. They also predict more precipitation overall.

However, with a more northward storm track and warmer temperatures, the precipitation increase actually leads to more rainfall and less snowfall for the Northern Hemisphere mid-latitudes in winter, including the eastern United States. The models predict an increase in snowfall across parts of the Arctic, and in particular Siberia, where temperatures for much of the year are well below freezing and where an atmosphere with more moisture can more than compensate for warmer temperatures.

But contrary to the predictions of the climate models, recent winters, especially since 2009-2010, have produced some historic snowstorms and seasonal totals in the Northeast. So, despite being much warmer now, the large cities in the Northeast are seeing an increase in snowfall, similar to the model projections for Siberia. Once-moderate snowfalls are now blockbuster snowstorms instead, pushing both storm and seasonal snow totals higher.

So are the models wrong? Does climate change actually result in an increase in snowfall?

To try to begin to answer this interesting and societally important question, we have plotted the top 10 snowfalls for eight major Northeastern cities from the District to Boston. The first thing that jumps out from the plot is the clustering of historic snowfalls in the most recent years. Clearly the frequency of extreme or historic snowfalls is increasing in the Northeast. Whether this is simply natural variability or a consequence of climate change can be answered only by more rigorous analysis and modeling studies; nonetheless, the increase in historic snowfalls is striking.

Another interesting finding from this plot is that the upper limit of the most intense snowfall events appears to be rather constant at about 30 inches. This upper limit is probably due to the geography and the topography of the Northeast and seems to be universal for all the cities plotted despite a wide range in seasonal snowfall or the number of moderate to large snowstorms per year (Albany, N.Y., being a notable exception during the Blizzard of 1888).

But if the atmosphere can hold more moisture as the climate warms, shouldn’t the upper limit be increasing as well? That the number of large snowfalls are increasing while the upper limit of the snowfall totals is not suggests that the answer lies in something more than just moisture — that there is a dynamical reason for the increase in extreme snowfall events and that it is not as simple as the thermodynamic argument that a warmer atmosphere can hold more moisture.

The above plot indicates that the trend in the most extreme snowfalls has not been increasing. But there is still the possibility, for example, that a 10-inch snowfall 50 years ago is now a 20-inch snowfall today because of more moisture availability in a warmer climate. To examine this hypothesis, we’ve plotted the vertical temperature profile from Chatham, Mass., from four historic snowfalls: two of the most recent large Northeast blizzards (Feb. 8, 2013, and Jan. 27, 2015) and two extreme snowfalls from the winter of 1978 (Jan. 20, 1978, and Feb. 7, 1978). 1978 was the mid-year of the three coldest winters recorded for the United States, and when global average temperature was colder than it is today.

If today’s atmosphere can hold more moisture during snowstorms because of recent warming, we would expect that the atmospheric column during snowstorms should be warmer today than in the past. Of the four storms, the high atmosphere was by far the coldest during the Feb. 7, 1978 storm. This is supportive of the idea that today’s snowstorms are possibly warmer and moister, leading to greater snowfalls than in the past. But the issue is complex; this storm produced all-time record snowfalls that are still in place for cities like Providence, R.I., demonstrating temperature is just one factor in producing extreme snowfalls.

By contrast, the temperature profile for the Jan. 20, 1978, snowstorm is not that much different from the vertical profiles of the two most recent snowstorms. This similarity may suggest that, regardless of the average climate, the vertical temperature profiles need to be on the warm side to support extreme snowfall.

Our analysis suggests that the contribution of a warmer climate to greater snowfalls is not as straightforward as a “warmer atmosphere holds more moisture” argument. Extreme snowfalls are occurring more frequently now than in the past, but the upper limit of the snowfall totals remains unchanged. The influence of climate change may be that the warming brings the atmosphere closer to the “optimal” profile to generate heavy snowfalls, more so in the present than in the past. More rigorous analyses are warranted to determine the relative roles of the dynamics versus thermodynamics in creating these extreme events.

Judah Cohen is an atmospheric scientist at AER, developing long-range forecast products for some of the largest investment firms in the United States. Cohen has a Research Affiliate appointment in the Civil Engineering Department at Massachusetts Institute of Technology and has published over two dozen articles on seasonal forecasting. Most recently, Cohen was appointed associate editor of the American Meteorological Society’s Journal of Climate.

Jason Furtado is an atmospheric scientist at AER, providing medium- and long-range weather analysis for clients and assisting with seasonal forecasts. Furtado received his PhD from the Georgia Institute of Technology in 2010 and is a member of the American Meteorological Society, the National Weather Association and the American Geophysical Union.