It wasn’t the storm of the century. Heck, it wasn’t even the storm of the week.
What was advertised to me one of the most destructive storms since the Great Hanukkah Eve Storm of 2006 was nothing more than a breezy day for most people around the Pacific Northwest. When the National Weather Service mentioned that this could be a storm that would “be long remembered,” this isn’t what they had in mind.
The Lackluster Bluster is a perfect case-study of the repercussions of forecasting a far more severe scenario than the one that plays out. People put their faith in the National Weather Service and the local and national media, cancelling important events and making preparations for a “historic” storm. Though I thought that the NWS did a great job with such a difficult, rapidly changing forecast, there’s no doubt that we need to implement changes in our approach to forecasting major storms, particularly when the track and intensity of these storms has a high element of uncertainty. All it takes is one bad forecast to decrease people’s confidence in meteorologists as a whole, meaning that when we DO finally get our “storm of the century,” people may not heed the warnings of the National Weather Service and local media.
Because there is so much to discuss, I’m going to write this blog in several parts. First, I will give a synopsis of just how difficult this forecast was, and what went wrong at the very end. In later blogs, I will discuss what we can do to improve our responses to these types of situations in the future.
Shown below is the UW WRF-GFS model, which is used by many local meteorologists for forecasting the weather around our region. I have created GIFs showing how the 5 pm Saturday forecast changed throughout the past 14 model runs ending on Saturday evening’s run, when it was clear that that we would not be impacted by a major storm.
All the major global models were on board with sending the remnants of Super Typhoon Songda across the Pacific and developing them into a major extratropical storm off the Pacific Northwest as far back as a week in advance. Some, like the crème de- le crème European model, brought a sub-960 hPa low right into Southern Vancouver Island, which would be a major, regionwide windstorm for nearly everywhere in Western Washington and Oregon. Others, like the GFS, sent the low further northward, thrashing the coast with 100 mph gusts and 40 foot waves, but sparing the Puget Sound region from high winds. Before Thursday, when the models were in agreement that a major storm with a central pressure of 960 hPa would take a dangerously close pass at Western Washington and Oregon, meteorologists were very concerned about the most damaging storm to hit the Pacific Northwest since the Columbus Day Storm of 1962.
The forecast changed dramatically Thursday morning, as models began to develop a system on Friday ahead of Saturday’s low. Because they were relatively close in space and time, Friday’s low was able to tap into some of the tropical moisture from Songda, weakening Saturday’s low in the process and making it much smaller and more compact – around 970 hPa instead of 955. However, the models also took Saturday’s windstorm right across the Olympic Peninsula, setting up the stage for a major windstorm across the Puget Sound region like the devastating Inauguration Day and Hanukkah Eve windstorms of 1993 and 2006, respectively. Models were showing this path and track even Saturday morning, with the potential for 50-60 mph gusts. When you combine that with leafy trees, saturated soils, and weakened branches from Friday’s windstorm, forecasts calling for the most destructive windstorm in 10 years still seemed justified.
But as Saturday morning went on, it became clear that the storm was further west than modeled and that the region of maximum pressure gradients would reside over the ocean, not even coming into contact with coastal communities along the Oregon and Washington Coasts. The low was also a tiny bit stronger than was modeled the previous day, as a buoy near Tattoosh Island just off the northwestern tip of Washington recorded a minimum pressure reading of 967 hPa. According to Pacific Northwest windstorm expert Wolf Read, the Tatoosh Island buoy dropped a staggering 8.4 hPa in one hour, which is among the largest hourly pressure drops ever recorded at a station during a Pacific Northwest windstorm. Additionally, there was an extremely rapid pressure rise after this drop, with the pressure rising approximately 6 hPa per hour in the two hours following the drop. This evidence supports the idea of an extremely compact storm tracking very close to Tatoosh Island. It’s a shame that the anemometer at the Tatoosh Island station is down – I’m sure it would have measured some extremely powerful winds.
Though there was undoubtedly a large amount of disagreement in the models for many days prior to the storm and the storm tracked slightly further west than the models were predicting (well within the margin of error), this was NOT, in my opinion, a forecast model failure. Rather, this is a failure in communicating changes in the forecast as the week went on and the compact nature of Saturday’s low.
I am planning to meet with the folks at the Seattle National Weather Service office soon (no timeline yet, but hopefully within a couple weeks) to discuss ways to better communicate uncertainty to the public. If you have any suggestions on how we can all improve at our craft, please leave them in the comments below! Moreover, if the National Weather Service decides to host a broader, community feedback session, I will make an event on the Facebook page and lead a trip down there with anybody who is interested.
Thank you for staying with me throughout the storm, and I promise to improve my forecasts in the future!
Charlie
2 Comments