The forecast for Wednesday’s snowfall in Washington was agonizing. Just 18 hours before we were expecting the first flakes, model predictions ranged from 2 to 12 inches. Ultimately, 4.1 inches fell at Reagan National Airport (Washington’s official weather site), the most this year and the most so late in March since 1964.
This amount fell right inside Capital Weather Gang’s most likely snowfall amount of 2 to 5 inches in areas from downtown Washington southward.
Around the region, 3 to 6 inches fell generally, although a pocket of 6 to 8 inches set up in our eastern suburbs. Amounts increased to 6 to 12 inches in northern Maryland and decreased to 2 to 4 inches in north-central Virginia.
Dulles Airport posted 5.3 inches, and BWI recorded 4.7 inches.
Evaluating Capital Weather Gang’s forecast
The forecast for the storm was really difficult. I know we say that about a lot of storms — because forecasting snow in Washington is never easy — but this one was particularly hard.
Model forecasts were bouncing around for days. The European model led the charge in calling for a significant event, as early as one week ago. We were skeptical but recognized that a snow event was a realistic possibility given the overall weather pattern — which we had said in late February would be favorable for snowstorms during March. We wrote:
The bottom line is that the pattern seems to offer at least one more small chance for a winter storm before we turn to spring. More likely than not, it will not materialize into a blockbuster, but the European model suggests that we should pay close attention.
By Friday and Saturday, we concluded that a significant storm was likely and offered a range of scenarios. But we stressed that models were jumping around on the details, which we would not be able to narrow down until Monday.
- On Friday, we wrote: “Will this fourth in the parade of March coastal storms finally be the one to exert a major impact on the Mid-Atlantic? It has that kind of potential but is far from a lock.”
- On Saturday, we wrote: “We still don’t know whether the storm is likely to be a mostly rain event with a little wet, inconsequential snow or a storm that produces moderate to heavy snow.”
On Sunday, we figured out that the storm would probably come in waves, the first rainier (Monday night and Tuesday) and the second snowier (Wednesday) around Washington. We issued a snowfall forecast map for the first phase of the event, which reasonably approximated what happened.
By Monday, as we realized the first part of the event would mostly be a rainstorm around Washington, despite some mixed precipitation north and west of town (which mostly did not stick to roads), we turned our attention to the snow potential on Wednesday. “[T]here is potential to bring widespread accumulating snow if it comes together,” I wrote. But I cautioned that forecast surprises were possible, quoting our winter weather expert Wes Junker, who called the pattern “a convoluted mess.”
We released our first snowfall forecast map for Tuesday night and Wednesday that day, calling for 1 to 3 inches in the immediate area and 3 to 6 inches for our northern suburbs. This was a conservative forecast and on the low side, but not terrible 36 hours ahead of a complicated storm. We correctly cautioned that Wednesday’s morning commute could be problematic.
Forecasting took a dramatic turn on Tuesday, on the eve of the storm. We were taking a slow and steady approach to forecasting this storm, and our prediction of snow amounts was lower than pretty much every outlet in the region. We realized the predicted evolution of weather features was nearly textbook for a significant snowstorm in Washington but, as experienced forecasters, knew that it is really hard for more than about 3 to 6 inches of snow to fall at this time of year.
That said, we did bump up our forecast amounts for the storm because models had come into reasonably good consensus that we’d see at least moderate amounts and the overall setup for snow was favorable. We released the map below:
The amounts on this map were still lower than what most other government and media outlets in the region had forecast, but we felt comfortable with it. That is, until the European model came out a couple of hours later.
The European model suggested that the region would see widespread double-digit totals. We went into a little bit of a panic mode because, this winter, it had been the most accurate model for precipitation forecasts at short range.
Our team had a lot of internal discussion and deliberation and, after dozens of emails and multiple phone calls, concluded that the European model forecast was probably bogus. We decided we would not increase our snowfall unless models later that afternoon supported the European model — which they did not.
So we stuck with the above forecast for most likely amounts, while expanding our “boom scenarios” as a small nod to the European model (even though we were skeptical about it).
It was the right call. If you compare the forecast map above with what actually fell, it is not a bad match. Yes, our totals are a little on the low side in northern Maryland and some of our southern and eastern suburbs. But, generally, the idea of 2 to 5 inches from downtown Washington south and 4 to 8 inches to the north worked out well.
A lot of people get caught up on inches, but, more important in my view, we correctly described how the event would unfold and how big of an impact it would have on the region. While the storm got off to a slightly slower start than we thought, we accurately said that conditions would be worst in the morning, when snow would fall heavily at times, and that they would slowly improve in the afternoon because the snow would have a difficult time accumulating as temperatures rose.
We welcome your thoughts as to how we handled this forecast.