Last week, there were strident warnings of a strong windstorm from the National Weather Service, which were amplified by conventional and social media.
The storm never happened.
Instead, we had a short period of winds gusting to 30 mph or so, producing little damage and few power outages. I received dozens of emails about the errant forecast. Folks were not happy.
As I will describe below, the problem was not the technology of weather prediction, but the poor communication of forecasts and their uncertainty.
The Problematic Prediction
Last Tuesday, the National Weather Service put out a high wind warning for the western Washington lowlands for Wednesday (see below).
According to the official definition, a High Wind Warning means that dangerous winds (sustained 40+ mph or gusts 58+ mph) are imminent or occurring, requiring action.
A warning should only be provided when a forecaster is absolutely sure that the threat will occur. This was NOT the case for this event. Not even close. I describe why below.
This problematic warming was then amplified by the traditional media and further hyped by social media "influencers" and local weather enthusiasts, few of whom have degrees in meteorology. A sampling below:
Below are the maximum winds observed last Wednesday. Most locations had gusts of only about 30 mph. A few in their 30s. Only near the water were higher gusts noted (46 mph near the Tacoma Narrows Bridge and 59 mph on southwestern Whidbey Island).
This time of the year, after the leaves are down and following earlier wind events, there are very few problems with gusts under 30 mph. Very minor problems for 30-40 mph. Only when winds gust to above 50 mph do series power outages become apparent.
No wonder this "event" was a big disappointment for those looking for intense weather.
Are you ready for some inner baseball of the weather forecasting business and to view some state-of-the-science tools?
On Tuesday, of the major international weather prediction centers, only the American was producing a strong, very small-scale low-pressure center moving up the Northwest coast.
To illustrate, here are the 36h forecasts for sea level pressure and surface winds for the American GFS model and the European Center (ECMWF) model, both valid at 4 PM Wednesday. The GFS had an intense mini-low, and the ECMWP had a weak inland low.
But there were more reasons to be worried. The National Weather Service runs ensembles of many forecasts, each a little different, to get a handle on the uncertainties of the forecast. The NWS ensemble forecasts made Tuesday morning for Wednesday were ALL OVER THE PLACE, indicating great uncertainty.
Earlier on Tuesday, the NWS forecast seemed to recognize this uncertainty in a graphic they produced (see below). Very nice.
But later that day, they literally threw caution to the wind and put out the high wind warning:
Next, let me show you a graphic from the Seattle Windwatch site, which my group provides to the city.
This graphic presents the maximum gusts predicted over Seattle by the NWS and several modeling systems for the forecast cycle starting at 4 PM on Tuesday.
The observed winds are shown in black (with a high of around 35 mph). The shading shows the range of forecasts by the US model (GFS).
The NWS forecast wind (yellow color) was at the extreme top of the ensembles, showing they were gambling on a low-probability solution. The UKMET forecast (light red) and Canadian model forecast (dark red), both consistently more accurate than the American model, were almost perfect.
So we had a situation in which the NWS forecasters did not properly consider uncertainty, nor did they recognize that historically better forecasting systems had a very different (and more benign) forecast.
The conventional media, like the Seattle Times (see below), went with the poor NWS forecast, and the weather amateur YouTube channels, Facebook, and X outlets exaggerated the event even more.
Let me be honest, perhaps too honest. A major problem is the deterioration in the forecasts of the National Weather Service.
Their global model, which drives almost their entire forecast system, is inferior to the best (e.g., the European Center). This event is a good illustration of that.
Many of the local forecasters in the Seattle NWS office are relatively new to the region and lack experience with the local meteorology and model performance. They also depend too heavily on the NWS models and don't use ensembles sufficiently. And this is not the only local forecast failure...the December ice storm a few years ago is another one--and there are several more.
A real warning sign is that objective verification of forecast quality has consistently shown that Seattle NWS forecasts (NWS Digital Forecasts) are inferior to totally objective forecasts, such as those from the Weather Channel (see below).
The National Weather Service needs serious reform and reorganization. The American people deserve state-of-the-science weather prediction. They are not getting it now, and last week was a good example of the persistent failure mode.
Cliff: this is spot on. An gency that thinks it will save its budget by predicting disaster and a local press that blames climate change for the increase in the number of potholes unfilled.. NWS needs to be shaken, not stirred.The NW group should be sent back to school to learn the science of uncertainty management and communication. Of course, this class could be funded by multi-millionaire Blethan but his Climate Groupies seem only to want more disasters and dislike hearing that weather varies and that is not always due to carbon changes in Brazilian rainforests.
ReplyDeleteA number of significant December precipitation records were set this month in the NSW Seattle CWA and based on the current forecast it appears that the current totals for most locations will not increase by the end of the month.
ReplyDeleteKBLI recorded 8.05" - its 4th wettest December in 77 years of records and the wettest December since 1979.
KOLM recorded 13.42" - its 3rd wettest December in 85 years of records and the wettest December since 2015.
Concrete recorded 19" - its wettest December in 121 years of records.
Diablo Dam recorded 26.23" - its 2nd wettest December in 112 years of records and the wettest December since 1917.
Longmire recorded 24.55" - its 4th wettest December in 117 years of records and the wettest December since 1975.
Mud Mountain Dam recorded more than 14.36" (many days of data are missing) - its wettest December in 87 years of records by a huge margin. The second wettest was December 2015 with 12.74".
Newhalem recorded 24.57" - its 2nd wettest December in 117 years of records and the wettest December since 1933.
Ross Lake recorded 24.08" - its wettest December in 66 years of records by an enormous margin. The 2nd wettest was December 1975 with 18.6".
There are also a number of significant temperatures that will be set this month at man sites in the NWS Seattle CWA as well as precipitation and, especially, temperature records set at sites within the NWS Spokane and Pendleton CWAs which I'll also comment on once more data becomes available. Needless to say, this has been a December for the ages in Washington!
Boom! And that isn't thunder.
ReplyDeleteThat is science.
The times primed and primed this event for all it was worth, towards the end one little drop escaped from the rusty cabin spigot.
ReplyDeleteI've long given up on the 'news' for anything weather related. As an outdoor enthusiast and an advisor for a high school hiking class/group I am on the NWS site(s) constantly and the Forecast Discussion portion of the site is meaningful as you can watch the forecast evolve. Leading into the wind(less) event, you could see the meteorologists struggle literally up to the day as they described the potential of a semi-major event vs. just another windy winter's day. I'd rather them 'overforecast' than not in the event of uncertainty. It's my job to take that and do with what I will in my decision making process and safety. I agree wholeheartedly that that U.S. needs/deserves the world's best forecasting ability.
ReplyDeleteWe had a similar situation of NWS forecasters getting excited and letting their feelings over-rule the data they were seeing on the day of an EF-2 tornado in Michigan last year. After the tornado sliced through the middle of Portage, they issued a Particularly Dangerous Situation for another spin-up further southeast in a less-populated area for a tornado that turns out lifted before the PDS was issued. Also, last month along the front range of the Rockies a PDS was issued for Chinook winds that occur most winters and are always a risk to fan wildfires in the dry climate on the lee of the mountains. I wonder if the increased use of warning words will desenstitise the public to the most dangerous situations that infrequently occur. https://normanjansen.substack.com/i/145321796/conclusion
ReplyDeleteThe bottom line of norman's text is an extreme example of "justified" text. I thought this was eliminated in the 1990s.
DeleteIf "fixing" the NWS means it going private and putting weather alerts behind a paywall, I would rather keep the so-so version we have as a free public service. I do NOT see this administration putting any effort into an agency that can produce climate data that runs counter to policy favoring fossil fuels or the AI build out. So either the NWS stays "meh" or it just gets taken around back and put down. To be replaced by the latest app that bombards the user with adverts in spite of being a pay service and then sells your data.
ReplyDeleteCurrently, if consumers want better weather forecasts, they can buy a sub to any of the private service apps. Or just learn to live without it. Most of us, even if the weather is inclement to the point that going out in it goes against good judgement or even common sense....have to go to work regardless. Even if it kills us. So the weather can be sort of irrelevant given those circumstances...and thus the NWS forecasts can go ahead and suck. Also, perhaps the consumer will one day realize that "influencers" are bottom feeders.
As someone who spent 50 years in broadcast and private sector meteorology, I've never once heard someone suggest ending the NWS and putting private sector forecasts behind a paywall.
DeleteThat said, as Cliff points out, the NWS isn't even competitive in its forecasts any more (see here, https://www.forecastadvisor.com ) and its storm warnings are rapidly deteriorating. Tornado warnings from the NWS, for example, reached maximum accuracy in 2005-2010 and have deteriorated since.
Seems like 2025 has begun and ended with a similar forecast issue - just refer back to your blog about the March 26 non-Severe event.
ReplyDeleteWhere has all the wind gone?
ReplyDeleteAt this time {12/30/25) most of the world's land area has low wind speeds. For the PNW, there is almost no change expected in, at least, the next five days.
Check the Bonneville Power chart. Wind (VER) is the green line in the large line chart.
{I've suggested they redesign the two images to clarify the colors; now a miss-match.}
Precipitation totals during December were also most impressive along the east slopes of the northern Cascades.
ReplyDeleteMazama recorded 11.61" - its wettest December in 76 years of records by a margin of more than 3".
Winthrop recorded 6" - its 4th wettest December in 120 years of records and the wettest December since 1996.
Stehekin recorded 21.04" - its 2nd wettest December in 120 years of records and the wettest December since 1933.
Plain recorded 12.98" - its wettest December in 89 years of records by margin of more than 2".
Once more data becomes available I'll provide more examples of precipitation and temperature records set during December 2025. The temperature records, particularly from the central and eastern portions of the state, are truly exceptional.
Mariners have an endless debate about which model is best for Puget Sound winds. In addition to the Christmas Eve wind nonevent, HRR predicted a big blow the night of Fri Dec 26 that also didn't happen- I know because I was anchored in Andrews Bay on Lake Washington at the time. Do you have any thoughts on which model is best for 24 hour and 48 hour wind prediction on Puget Sound? ECMWF?
ReplyDeleteI agree, the bias is in favor of "crying Wolf" rather than underprediction. Those (like me) who sort of enjoy storms get disappointed. This is especially true in this area regarding snowstorms and thunderstorms. I have learned not to get my hopes up when they predict thunderstorms! Of course, in the NW, a few rumblings constitute a "thunderstorm". But Cliff, from what you have said before about fires, I gather you would rather be safe than sorry and shut off the electric lines rather than risk a forest fire when they expect, but are uncertain about, a summer downslope east wind.
ReplyDeletehey Cliff, I find it curious and interesting that you didn’t mention how the Trump administration has been taking a hatchet to NOAA and thenl NWS...cutting their budget and reducing staff.. Certainly their challenges lately are due in part to being staffed by a bare-bones crew. you get what you pay for
ReplyDeleteRobert...the problem pre-dates Trump....the decline has been occurring for decades. The Seattle Office is fully staffed--so there is a full crew...cliff
DeleteFully staffed but "many of the local forecasters in the Seattle NWS office are relatively new to the region and lack experience with the local meteorology and model performance. "
DeleteWonder how many experienced forecasters took the early buyout?
None. The problem is the system, which has been in place for years. This is not about Trump
DeleteThanks for the great explanation! All I know is I was at Mass "during" the forecasted /predicted horrible wind/rain/calamity 4-6pm and.... ?? The sunshine was streaming through the stained glass windows and it was absolutely calm/gorgeous..... I just thanked God for a wonderful day!
ReplyDeleteReally interesting post. Thank you.
ReplyDeleteI'm seeing colder temperatures next week, talk to us
ReplyDeleteDecember 2025 was, not unsurprisingly, exceptionally dark and sunless. I measured total solar energy for the month at my location in Bellingham of just 48.13MJ/m^2. This is the lowest monthly solar energy that I've measured since December 2019 with 46.39MJ/m^2. For reference, a single typical bright and sunny day around the summer solstice yields ~25MJ/m^2 at my location.
ReplyDeleteThere is a hint of ridging on the horizon starting around Jan 8th. Could we see an extended dry spell in January like we did last year?
ReplyDeleteHi Cliff, as you know I completely agree major changes at needed both at NWS and NOAA.
ReplyDeleteI also suggest considering a subtlety here, "the official definition, a High Wind Warning means that dangerous winds (sustained 40+ mph or gusts 58+ mph) are imminent or occurring, requiring action." If a phenomena (high winds) is already "occurring," the message is a report, not a warning.
An opposite example of yours: I will not name the city but in 2025, 60 mph winds were forecast. 106 mph gusts occurred at one of the official (Class 1) weather stations in the city and 96 was clocked at another shortly after. No one thought this huge gap between forecast and actual conditions was especially unusual. But it is unacceptable and likely would not have occurred 15-20 years ago.
While I believe that the uncertainty was better communicated this time compared to previous events of this nature, throwing caution to the wind and issuing the high wind warning on Tuesday, while there was still quite a bit of uncertainty, was the wrong move for the NWS to make. They should have kept the high wind watch that was in effect going until the forecast could be narrowed down more, while continuing to convey the fact that things were still uncertain. The NWS caused unnecessary stress, panic, angst, and anxiety for many people with their decision at a time of year when people are already overly stressed, and right on the heels of one of the most significant flooding events our region has seen in years. I would love to know what factor(s) led the NWS to issue the high wind warning when they did.
ReplyDeleteAlso, since the NWS forecasts are becoming less reliable, and have been getting worse for years now, I can't help but wonder if the time has come for a private weather service, like weather.com, to start issuing their own NWS style watches, warnings, and advisories. The NWS has long been the one and only entity issuing such warnings, and they are taken very seriously by federal, state, and local officials, even at times when the NWS gets it wrong, such as this one, which is hurting their credibility. Would it be beneficial for these officials, as well as the public, if the NWS wasn't the only entity issuing official warnings? Or would it do more harm than good?
No private company will do that because of liability issues.
DeleteTrue. If they get it wrong like the NWS, they could be sued far more easily than the government can be, and there are people would be inclined to do it in our increasingly litiguous society. That would have to be addressed before any private company could begin issuing such warnings.
DeleteCliff, your assessment about the newer forecasters over reliance on forecast models and not local knowledge or local expertise is spot on. The problem goes well beyond that. I recently retired from a PAC NW office, and over the last several years there has been more emphasis on the "load and go" mentality. The forecast database gets preloaded with the NBM via a program called the "forecast builder." There is a NWS forecast office in the Pac NW which will not be named and is being told not to make any changes to what is loaded or face potential consequences. There are other offices that follow similar rules, but there are some forecasters that will follow the "trust but verify" concept and they will make improvements. However there are some that will choose not to "ruffle feathers" and just go along with what gets loaded, even if the forecast does not make sense given what the operational models and ensembles are showing. There have been several cases where I chatted with the forecast office that does not alter what the NBM loads to question a forecast. Their response was in the order of "were comfortable with what the NBM has" even though operational models and ensembles tell a different story. It's common knowledge in the realms of the field offices that the NBM does not perform well in Pac NW and mountain west forecasting. This is where the forecaster with local expertise can add value to that is loaded. Furthermore, in the not too distant future, the forecast is going to be more centralized where WPC (Water Prediction Center) will take over the local forecast responsibilities (excluding issuing watches,warnings ect) that the WFO's currently do. Additionally, there is less emphasis on the accuracy of the NDFD (official forecast), meaning it's not important. Part of the reason is there has been less attention given to the general public and more towards core partners (think D.O.T. 's Emergency managers ect) through DSS (decision support services). This is something better served by the private sector. After all, it's the general public that helps fund the NWS. Most of my former colleagues do not understand this. I look at this from a different lens because prior to working at the NWS, I worked for a very successful private weather company in the Seattle area. A few of my former colleagues at the time were former students of yours and their knowledge of the local area and expertise allowed me to become a much better mountain west and Pac NW forecaster. The knowledge they shared was invaluable. This was at a time where I was "green as grass" and had little or no experience for this area. Also in the NWS if the forecast is off or wrong, there is no accountability. In the private sector, you better produce quality and accurate forecasts, otherwise paying clients will move on, resulting in loss of revenue and possibly loss of jobs.
ReplyDeleteCliff, your assessment about the newer forecasters over reliance on forecast models and not local knowledge or local expertise is spot on. The problem goes well beyond that. I recently retired from a PAC NW office, and over the last several years there has been more emphasis on the "load and go" mentality. The forecast database gets preloaded with the NBM via a program called the "forecast builder." There is a NWS forecast office in the Pac NW which will not be named and is being told not to make any changes to what is loaded or face potential consequences. There are other offices that follow similar rules, but there are some forecasters that will follow the "trust but verify" concept and they will make improvements. However there are some that will choose not to "ruffle feathers" and just go along with what gets loaded, even if the forecast does not make sense given what the operational models and ensembles are showing. There have been several cases where I chatted with the forecast office that does not alter what the NBM loads to question a forecast. Their response was in the order of "were comfortable with what the NBM has" even though operational models and ensembles tell a different story. It's common knowledge in the realms of the field offices that the NBM does not perform well in Pac NW and mountain west forecasting. This is where the forecaster with local expertise can add value to that is loaded. Furthermore, in the not too distant future, the forecast is going to be more centralized where WPC (Water Prediction Center) will take over the local forecast responsibilities (excluding issuing watches, warnings ect.) that the WFO's currently do. Additionally, there is less emphasis on the accuracy of the NDFD (official forecast), meaning it's not important. Part of the reason is there has been less attention given to the general public and more towards core partners (think D.O.T. 's Emergency managers ect) through DSS (decision support services). This is something better served by the private sector. After all, it's the general public that helps fund the NWS. Most of my former colleagues do not understand this. I look at this from a different lens because prior to working at the NWS, I worked for a very successful private weather company in the Seattle area. A few of my former colleagues at the time were former students of yours and their knowledge of the local area and expertise allowed me to become a much better mountain west and Pac NW forecaster. The knowledge they shared was invaluable. This was at a time where I was "green as grass" and had little or no experience for this area. Also in the NWS if the forecast is off or wrong, there is no accountability. In the private sector, you better produce quality and accurate forecasts, otherwise paying clients will move on, resulting in loss of revenue and possibly loss of jobs.
ReplyDeleteCliff, we chatted last in late December 2022, after the extreme low pressure event (28.71)over the PNW on the 27th, coupled with the king tide which flooded thousands of homes (including mine). I was relieved to hear that in your expert opinion, you deemed it a 100-year event. Well, we got as close as we ever have in the last three years at 6:29am this morning. We had a king tide and severe low pressure (29.32) effectively added over 12”+ to the tide tables. Thankfully, right at the peak high, the wind velocity dropped to near zero! The water was just 9” from my living room.
ReplyDelete