Everyone Struggled with This Weekend’s East Coast Snowstorm
PRO TIP: This winter storm is going to be a perfect candidate for a case study on the verification of numerical weather prediction forecasts.
The flakes started flying across Southern Connecticut and Rhode Island around 7:00pm last night, ushering the start of a well-hyped, late-March snowstorm. By 10:00pm, light snow had pushed north into Massachusetts and by early morning, snowfall reports began filtering into the region’s National Weather Service offices. One thing happened however that no had forecasted…
Some examples of total storm snowfall forecasts from various Boston-area media outlets:
All of the sampled media outlets had forecasted the heaviest snow to fall across Southeastern Massachusetts to the upper part of Cape Cod. Heck, even the National Weather Service was calling for similar:
Here is how the storm actually panned out across the region…
The heaviest swath of snow ended up falling from Southeastern New Hampshire into Eastern Connecticut, a good 70mi+ northwest from where most everyone had forecasted.
Were we duped? Was this because of a lack of forecasting knowledge?
Most likely not, but what was definitely a factor is a building trend of over-relying on deterministic numerical weather prediction during the medium range. The hype build-up for this storm began early last week and continued to gain steam as the weekend approached. The predictability of snowfall is hard enough to nail down three days out but, beyond that time period? Garbage…
Ensemble forecasting, which takes uncertainty into account, is what should have really been communicated for this storm. Wanting to jump to the answer is all too common these days, where communicating uncertainty and low predictability should be a top priority, not disseminating snowfall maps 3+ days out.
No one person is to blame here but I know for a fact that we, as a science-driven community, can do better.