Share this article

print logo


Don Paul: How I — and other meteorologists — nailed the forecast for this storm

“So, I suppose I’m calling an audible in a print medium.” Those were the words I wrote in the Buffalo News Monday in my “moderate confidence Heads Up” concerning the storm just passed.

On my last workday, Feb. 25, it had become clear a storm would pass by close to our south and then jump off the coast and become a very powerful nor’easter. However, last Sunday’s models suggested we would receive mainly rain from this system as it made its move. Here is what the American GFS model was showing for 1 a.m. Friday.

The Canadian model gave me a little more pause showing rain for the same time period but at least transitioning it over to some light snow by Friday morning.

The European model showed some very modest snow mainly in the Southern Tier, with the storm passing by farther to the south. I chose a conservative route Sunday night and forecast some slush on the hills by Friday. Even the mention of those slushy accumulations was sticking my neck out, but not very far.

On Monday morning, a welcome day off, I decided to go through new model runs, and spend some time thinking about pattern recognition. For one thing, the northwest Atlantic sea surface temperatures have been exceptionally warm for quite some time. Here is a look at the temperature anomalies I’m referring to.

The oranges and red indicate unusually warm anomalies, some of the bigger anomalies on the globe. That extra heat makes the cold air coming off the land much more unstable, or “bubbly.” When air gets unstable, it lifts and pressures fall at the surface. I knew there was excellent model agreement the low pressure system would be vigorous even before it made the jump to these warm waters. Once the jump was made, it was inevitable the lift from the air-sea temperature differential would make explosive intensification of the low inevitable. The laws of physics, particularly thermodynamics, are a part of what is built into the models. While each model handles the many equations, climatology, and transfer of energy and momentum differently, those laws dictate with very energetic lift in the atmosphere comes a dynamic cooling. Since this coastal low was going to be a whopper, I wondered if more dynamic cooling would show up in newer model runs and bring more snow and less rain.

Here is what I found in Monday morning’s GFS.

And the Canadian.

Both models took the surface low off the coast faster but, as I suspected, picked up better on the dynamic cooling back over Western New York.

Game on for snow.

2 feet of snow in Wyoming, more than 18 inches in southern Erie County

I had to wait a bit for the newer run of the European to finish but when it did, it jumped in with both feet as well and introduced heavy snow potential, especially south of the city, which has turned out to be the case.

It was far too early to get into specifics about amounts. That’s just bad science. There is too much uncertainty five days in advance for reliable accumulation forecasts. However, certain things were clear. The marginally cold near surface temperatures would make certain any accumulation would be heavy and water laden.Wind would be a factor because of the rapid deepening of this low and the tightening pressure gradient around it. So, this is what we have today, Friday.

Dominique Nelson, left, and Isaiah Williams shovel the steps at Canalside. (Sharon Cantillon/Buffalo News)

A hurricane-force wind warning is in effect until 11 p.m. Friday for the coasts of Massachusetts and Rhode Island. Seas east of Cape Cod will have wave heights of 33 to 38 feet. Cruise lines have had days of advance warning for this monster, so voyages should have been delayed or cancelled. Precautions have been taken by emergency managers, military and civilian port facilities, and aviation interests. Obviously, the scope of this storm goes far beyond Western New York. On a large scale, more advance warning means lives saved, millions of dollars not lost, and safer travel.

In my experience and that of most meteorologists we know it can be all too easy to rely almost entirely on models. These tools have brought gradual but tremendous improvements in forecast accuracy, even further out in time (when I was in college, the extended outlook was for three days). But pattern recognition must be an important factor in the forecast process.

Another colleague who is quite learned in the use of analogs and pattern recognition was, much to my relief, also raising a red flag on his paid subscription site. I didn’t see it until after I’d submitted my story to the News, but it made me feel better to know I wasn’t out as much on a limb as I might have thought.

Once a forecaster starts to feel a bit smug about a forecast which went well, nature will almost surely come along in a short while and take that forecaster down a peg or 2, back to the inevitable uncertainties in forecasting active weather. I recall in dealing with what was called the “storm of the century,” the Blizzard of March 1993, National Weather Service models and analysts did a standout job eight or nine days in advance in projecting the ferocity of that storm. Some of us, myself included, thought a corner had been turned and we would have a range of computer guidance which could be reliable beyond a week on a regular basis.

It turns out, for the really big storms, there has been some regularity to that, and more than I could have imagined back in college. But we ARE getting closer to regularity on those big ones. It’s the smaller systems which snag us too often.

There are no comments - be the first to comment