By Meteorologist John Cifelli

Yesterday during the 6 o’clock news, I used the words “historic” and “prolific” to describe the forecast. These descriptions proved to be accurate, for all the wrong reasons. The meteorological community is reeling from a cataclysmic failure of forecasting, the worst seen in New Jersey since the disastrous March of 2001 blizzard that wasn’t. I thought we were past this, nearly 15 years and millions of dollars of computer upgrades later. The fact is that we are not, and consequently the biggest storm is the one forecasters are weathering today, and rightly so. One to three inches fell in southwest New Jersey, three to six inches is on the ground in central and western areas, and the rest of the state, from Ocean, Monmouth, Middlesex and north saw five to 10 inches.

This of course begs the question, what happened? It is important to point out what forecasting for New Jersey entails. There are two places in the United States that are favorable for birthing storms. Cyclogenesis occurs on the lee side of the Rocky Mountains, and along the East Coast. Mid-latitude cyclones, the powerful systems that affect us outside of summertime are fueled by the temperature gradient that exists between cold air masses that arrive from Canada and those that exist over the warm Atlantic waters. In the winter, that gradient is strongest. It’s why we don’t get nor’easters in the summertime; that difference in temperature doesn’t exist. Forecasting storm track and intensity is something all meteorologists around the world must do. Those who operate in the mid-Atlantic and northeast must also predict when, where and in what way storms will be born. I submit that this caveat makes forecasting for the East Coast one of the biggest challenges in United States meteorology.

Computer models are not perfect, partly because the initial conditions that are ingested are estimated in places. I frequently mention that the upper level features that impact the development of a surface low pressure center are over data-poor areas such as northern Canada or the Pacific. It is also necessary to simplify complex microscale weather phenomena. Then there is the fact that the physics and calculus equations used to represent atmospheric processes are not always explicitly able to be solved. This is technical stuff, but if you want to know where your snow is, it is lost in the assumptions, rounding errors and subtle misrepresentations of complex atmospheric phenomena. It’s also on Long Island. They racked up one to two feet there. If you go to see it, please bring some back for me.

Different models struggle with different situations, and they do tend to develop biases that can manifest errors depending on the setup. Recognizing what’s noise and what’s a viable solution is a technical process. Although there were models suggesting a track that would not impact New Jersey in the way expected, forecasters across the region (myself included) put our faith in what is usually a very safe place — an agreement between the ECMWF and NAM models. When these two are in step with one another for a short term forecast, it’s usually incredible useful to informing a forecast. With this storm we had this agreement until the last moment. The short-term inaccuracy of the former is unheard of.

I had cautioned leading up to this storm that “Miller B” events, which begin with a primary low pressure center over the Ohio Valley that transfers its energy to a developing coastal low pressure center, are delicate situations that usually have sharp cutoffs in precipitation and carry the possibility of developing too late to impact locations south and west of an anticipated affected area. At 2 p.m. Monday, the storm was 1002mb off the coast of North Carolina. It’s now 988 off the coast of Cape Cod. It was born off our coast yesterday, just a little too east and a little too slowly to have the expected output. Developing low pressure centers tend to develop around clusters of thunderstorms. The rapid ascent of air within thunderstorms create a void of mass, which is essentially what a low pressure center is. Surface winds converge toward this vacancy to fill the void with other air. One could argue that the key to determining storm track then is predicting where the thunderstorms will be. Anyone who has lived a day of summer knows how difficult that is.

There were pattern indicators that were not favorable for a west/stronger track, and I did not weigh them heavily enough. No one did. When there is a consensus of model data suggesting a widespread high-impact snowstorm for New Jersey and the ECMWF is leading the way, any meteorologist would fall for this. There was no way to forecast differently. No deeper investigation, more thorough analysis or harder work would have mattered. It was an incredibly delicate situation that just went the wrong way. It just as easily could still be snowing in Philadelphia right at this moment.

So was this a monumental bust of a forecast? Of course, for New Jersey and New York City. If we view from a larger perspective, East Coast cyclogenisis occurred — the storm produced a wide swath of feet of snow, and followed a relative well-modeled track. With a wide lens, things look different. This is, of course, no consolation to businesses, schools and governments across the state and in the metros. This is the unfortunate reality of forecasting for the the Philadelphia-New York corridor. It is a large population that is hyper aware, wealthy and works at an incredibly fast pace. I anecdotally refer to New Jersey as the center of the world. A forecast that misses by 50 miles here, misses harder than anywhere else on the planet. It’s unfortunate, challenging, but gratifying when done well.

What does this mean for you, the reader? I think it is fair to evaluate your forecaster of choice. Understand that as humans, we have biases and the ability to interpret the raw data available differently from one another. Some meteorologists are risk averse, some are comfortable with taking the big swing. Some have pressures to sensationalize or aggrandize to drive views or clicks. Forecasters may behave differently based on recent successes or failures, or forecast differently based on their known audience. Some prefer to blend in to a consensus and suffer from group think, while another may try to always stand out. This probably does not instill confidence — that there are known errors in models, and in the human element of the men and women who use them. My suggestion to you then: find the meteorologist who speaks the science in their forecasts. Be aware of the context of their forecasting. Trust us to do better. We’re all pushing to do our best.