The Seattle Times and others have been evaluating Seattle's snow response and in this blog I would like to provide an in-depth analysis from a meteorological perspective. And I would like to make some suggestions of some new ways of dealing with the problem.
There is, of course, a deeper question: can municipal governments, or any government for that matter, be expected to deal with rare disasters? The Feds did poorly with Katrina, Seattle had problems in Dec 2008, and I think all of us could put together a similar list for other major environmental events.
However, I do believe that we can do much better with a little planning and creativity.
Background: Snow in Seattle is a big problem for many reasons. It is a rare event so the population and the various municipal agencies are inexperienced with it. We have hills so a skid can get real serious quickly. Our ground starts out warm and the first snow inevitably turns to slush before it is frozen solid by the usual post-snow "arctic blast." It is a difficult forecast problem because all the elements have to come together in just the right way for serious snow to occur (see my book for explanations of the last two elements). Because of the relative rarity, municipalities don't want to invest the considerable sums required to purchase enough equipment and supplies to clear the city quickly.
This is a very difficult problem and one that takes very smart moves and high-tech to do the essential job effectively and for modest cost.
Report Card:
The City
Mayor McGinn wisely didn't give the city any grades, but admitted that the snow got ahead of them. An intelligent answer. Personally, I give them a "B." First, they were committed to dealing with it and invested in additional equipment and substantial supplies of salt and other deicers. They put together a reasonable plan of using their limited equipment to clear the key arteries of the city (the only thing to d0). They put down large amount of brine solution the day before the big event. I know--I went to the airport the night before and was impressed: spray trucks were seemingly everywhere and the lines of deicer were on all the major roads I traveled on. These folks were working hard. As we shall talk about later, the brine solution was not enough and some of the problems (WSDOT's issues on I5, Metro's bus antics) were not of their own making. Finally, they are working hard to learn from what happened on Nov 22 and to do better next time.
WSDOT
They was the most perplexing thing about this event. WSDOT are masters at keeping roads open during poor conditions and they proved that during December 2008. But on Nov 22, something went wrong. First, they weren't able to keep I5 deiced. Secondly, they failed to open the northbound express lanes of I5, greatly contributing to northbound gridlock in the city.
METRO
They get decidedly mixed reviews. On one hand they acted very proactively before the storm---prior to a single flake they went to snow routes and chained up their buses. A gutsy call and it turns out to be the right one. But then they made some serious errors. First, when things got bad that afternoon, Metro should have pulled all their articulated buses off their routes--they didn't and many got stuck. A Metro bus was a major contributor to the southbound I5 mess. Second, they have to pull their buses off the steep routes where the danger was great. Too many buses remained on unsafe routes. This video shows you the results--thankfully no one was seriously injured. But that bus should not have been on that hill.
http://www.youtube.com/watch?v=rhZCyQ3emQg
But what really was maddening about the METRO folks is their decision to shut off the bustracker software EXACTLY WHEN WE NEEDED IT!! They have a lame excuse about the bustracker locations having errors when the bus go off the regular routes, but this make little sense. For most of the snow routes useful information IS provided. And they certainly could modify the bustracker website to provide what information IS there...when the buses pass certain locations. The truth is that METRO has been very slow in going to a GPS tracking system and have never given sufficient emphasis to giving its patrons good information about where the buses are located. They could be doing much better even with their current old-fashioned location system.
The Forecasters
You don't expect me to be critical of my own profession, do you? The National Weather Service did about as well as the technology allows--a forecast that was good enough to provide warning of potential trouble. The day before they were going 1-3 inches, mainly in the afternoon as a coastal low moved south of the Olympics. Basically, they closely followed their main high-resolution model (the NAM, WRF-NMM). If you read their forecast carefully, they were only going for some flurries or light stuff in the morning, but it turned out to be considerably more (1-2 inches in places). The coastal low was more intense that expected by their model and they had to up the forecast snow amounts during the day. The NWS was correct about cold air coming in later that day. I have to admit, my forecast was not as good regarding snow the night before. I depend heavily on the UW high resolution system, which usually is superior to the NWS NAM model for a number of technical reasons I won't go into here. But this situation was unusual and the NAM was better that morning (I will have a future blog telling why and steps we are taking to fix it). Essentially, the UW system took the coastal low further south and the snow shield only extended to south Seattle. The UW system recovered Monday morning with a very good forecast--in plenty of time to deal with the afternoon threat.
Play by Play Analysis
On Monday morning the temperatures were near freezing as the first pulse of snow moved in. Since it had not been that cold and this was early in the season (remember it was 74F a few weeks before!), the ground and road surfaces in contact with it were relatively warm. Even the elevated structures (e.g., the Alaskan Way viaduct) hadn't cooled below freezing. So when the snow started, particularly with generous amounts of deicer SDOT had spread around, the roads were fine. The morning commute had very little problems. People got into work and school. I even biked in with no hassles on the Burke Gilman Trail.
So far so good.
But then the problems began. The low center along the north coast deepened more rapidly then expected and moved across the southern flanks of the Olympics (see satellite picture below)
As the low moved south, it produced moderate, mild southwesterly flow that moved northward to meet northerly flow pushing south down the Sound (the northerly flow was accelerating toward the lowering pressure over the south Sound). The converging air streams resulted upward motion and bands of snow. But at the same time the northerly flow brought colder and colder air over Seattle starting around 3-4 PM.
Here are the temperature and wind plots from the top of my building at the UW that day. Temps were between 28 and 30F until just after 3 PM and then temperatures started to drop quickly.
At the same time the winds started to become much stronger and gusty (see figure).
Strong winds are important because they provide much more effective removal of heat from the surface. It is like a fan blowing cold air over the ground...much more effective for cooling.
And now it all came together. The deicer put on earlier was increasingly diluted by the larger volumes of snow falling into it until it became of marginal value. The air temperatures were rapidly dropping and the strong winds enhanced cooling. What would ice up first? You know what...the elevated roadways which did not enjoy heat conduction from the relatively warm ground below. The Alaskan Way viaduct iced. The West Seattle Bridge. And since much of I5 is elevated, it glazed over rapidly as well. Icemageddon.
Traffic locked up on major roadways. Plow and sanders couldn't get to where they were needed. Trucks and buses got into trouble. And the rest was history.
What to Do
What I am about to express is opinion...and I am probably missing key points...but I will go ahead anyway.
First, although weather forecasts are getting better, any new system for dealing with snow has to be robust enough not to depend on the forecasts being correct in the timing or amount of snow. It will be 5-10 years before the 12h forecasts are dependable enough for such work. We are quite good in telling you if the cold and winds will come. Or revealing a threat. But getting the exact amount and distribution of snow even 6-12 hr out is not here yet.
The Nov 22 forecast by my colleagues in the NWS were far better than what they could have done 10 years ago. It was good enough to alert the city and various DOT agencies, and to let the general public know that something serious was possible. Certainly KING TV and other stations knew--they were hyping it up like mad the night before. Jim Forman was getting his famous parka pressed and readied.
But if the forecasts are not perfect, what is needed is good nowcasting--a term in my field which means examining current observations carefully and intelligently extrapolating them into the future. There are 50-100 weather stations reporting in real time in the city. With this information, one could determine exactly what the temperatures are and track the changes. As soon as the strong winds and colder air started to push in, it was time to get out there getting rid of snow on the roadways and hit them with lots of salt. Using the weather radar and temperature information it is also possible to determine the snowfall patterns over the city (this will get even better next year with the NWS updating of all local weather radars to dual-polarization and the addition of the new coastal radar). With the information about how much snow is on the ground, where it is falling, and how temperatures are changing, SDOT and WSDOT should be able to do their work far more effectively...particularly in getting ahead of the storm. A software system could be created to pull all this information together...we could call it SNOWWATCH. Furthermore, the city needs someone with meteorological training watching all the observations in real time during these events.
Once the ice forms, you are in trouble. The key is to know when snow is falling and to plow as much of it as possible off the roads and then hitting it with salt (my friends in the business suggested prewetted salt when temps are cold). If the snow keeps up, you have to repeat.
Now at around 3 PM it was clear from the weather observations that bad things were about to happen. The surface chart showed northerly winds starting to pick up and cold air moving in from the north (see plot). The radar showed snow over the region (see plot). Click for big version.
Conclusion: it was about to get colder, winds would increase and the threat of icing was about to increase rapidly. At this point, all equipment needed to be used on elevated structures immediately, and when those were cleared and salted (solid not brine), then the other roads could be dealt with. Metro needed to pull all their articulated buses off the elevated roadways at this point...in fact, they should have replaced with with the non-articulated fleet hours before. Rapid deployment when the weather was clearly going to change seems to me the only viable option.
Question: Do any of the roadway surfaces in the city have real-time temperature measurements? If not, shouldn't this be fixed immediately?
In summary: I am suggesting a combination of the intensive use of all observational assets to provide a real-time view of the weather and road conditions around the city, and the flexibility to use the limited assets of the city and WSDOT to hit the big threats first, guided by these observations. The current state-of-the-art of weather forecasts are good enough to give a heads up, but not good enough today to guide hour by hour actions.
But what really was maddening about the METRO folks is their decision to shut off the bustracker software EXACTLY WHEN WE NEEDED IT!! They have a lame excuse about the bustracker locations having errors when the bus go off the regular routes, but this make little sense. For most of the snow routes useful information IS provided. And they certainly could modify the bustracker website to provide what information IS there...when the buses pass certain locations. The truth is that METRO has been very slow in going to a GPS tracking system and have never given sufficient emphasis to giving its patrons good information about where the buses are located. They could be doing much better even with their current old-fashioned location system.
The Forecasters
You don't expect me to be critical of my own profession, do you? The National Weather Service did about as well as the technology allows--a forecast that was good enough to provide warning of potential trouble. The day before they were going 1-3 inches, mainly in the afternoon as a coastal low moved south of the Olympics. Basically, they closely followed their main high-resolution model (the NAM, WRF-NMM). If you read their forecast carefully, they were only going for some flurries or light stuff in the morning, but it turned out to be considerably more (1-2 inches in places). The coastal low was more intense that expected by their model and they had to up the forecast snow amounts during the day. The NWS was correct about cold air coming in later that day. I have to admit, my forecast was not as good regarding snow the night before. I depend heavily on the UW high resolution system, which usually is superior to the NWS NAM model for a number of technical reasons I won't go into here. But this situation was unusual and the NAM was better that morning (I will have a future blog telling why and steps we are taking to fix it). Essentially, the UW system took the coastal low further south and the snow shield only extended to south Seattle. The UW system recovered Monday morning with a very good forecast--in plenty of time to deal with the afternoon threat.
Play by Play Analysis
On Monday morning the temperatures were near freezing as the first pulse of snow moved in. Since it had not been that cold and this was early in the season (remember it was 74F a few weeks before!), the ground and road surfaces in contact with it were relatively warm. Even the elevated structures (e.g., the Alaskan Way viaduct) hadn't cooled below freezing. So when the snow started, particularly with generous amounts of deicer SDOT had spread around, the roads were fine. The morning commute had very little problems. People got into work and school. I even biked in with no hassles on the Burke Gilman Trail.
So far so good.
But then the problems began. The low center along the north coast deepened more rapidly then expected and moved across the southern flanks of the Olympics (see satellite picture below)
As the low moved south, it produced moderate, mild southwesterly flow that moved northward to meet northerly flow pushing south down the Sound (the northerly flow was accelerating toward the lowering pressure over the south Sound). The converging air streams resulted upward motion and bands of snow. But at the same time the northerly flow brought colder and colder air over Seattle starting around 3-4 PM.
Here are the temperature and wind plots from the top of my building at the UW that day. Temps were between 28 and 30F until just after 3 PM and then temperatures started to drop quickly.
At the same time the winds started to become much stronger and gusty (see figure).
Strong winds are important because they provide much more effective removal of heat from the surface. It is like a fan blowing cold air over the ground...much more effective for cooling.And now it all came together. The deicer put on earlier was increasingly diluted by the larger volumes of snow falling into it until it became of marginal value. The air temperatures were rapidly dropping and the strong winds enhanced cooling. What would ice up first? You know what...the elevated roadways which did not enjoy heat conduction from the relatively warm ground below. The Alaskan Way viaduct iced. The West Seattle Bridge. And since much of I5 is elevated, it glazed over rapidly as well. Icemageddon.
Traffic locked up on major roadways. Plow and sanders couldn't get to where they were needed. Trucks and buses got into trouble. And the rest was history.
What to Do
What I am about to express is opinion...and I am probably missing key points...but I will go ahead anyway.
First, although weather forecasts are getting better, any new system for dealing with snow has to be robust enough not to depend on the forecasts being correct in the timing or amount of snow. It will be 5-10 years before the 12h forecasts are dependable enough for such work. We are quite good in telling you if the cold and winds will come. Or revealing a threat. But getting the exact amount and distribution of snow even 6-12 hr out is not here yet.
The Nov 22 forecast by my colleagues in the NWS were far better than what they could have done 10 years ago. It was good enough to alert the city and various DOT agencies, and to let the general public know that something serious was possible. Certainly KING TV and other stations knew--they were hyping it up like mad the night before. Jim Forman was getting his famous parka pressed and readied.
But if the forecasts are not perfect, what is needed is good nowcasting--a term in my field which means examining current observations carefully and intelligently extrapolating them into the future. There are 50-100 weather stations reporting in real time in the city. With this information, one could determine exactly what the temperatures are and track the changes. As soon as the strong winds and colder air started to push in, it was time to get out there getting rid of snow on the roadways and hit them with lots of salt. Using the weather radar and temperature information it is also possible to determine the snowfall patterns over the city (this will get even better next year with the NWS updating of all local weather radars to dual-polarization and the addition of the new coastal radar). With the information about how much snow is on the ground, where it is falling, and how temperatures are changing, SDOT and WSDOT should be able to do their work far more effectively...particularly in getting ahead of the storm. A software system could be created to pull all this information together...we could call it SNOWWATCH. Furthermore, the city needs someone with meteorological training watching all the observations in real time during these events.
Once the ice forms, you are in trouble. The key is to know when snow is falling and to plow as much of it as possible off the roads and then hitting it with salt (my friends in the business suggested prewetted salt when temps are cold). If the snow keeps up, you have to repeat.
Now at around 3 PM it was clear from the weather observations that bad things were about to happen. The surface chart showed northerly winds starting to pick up and cold air moving in from the north (see plot). The radar showed snow over the region (see plot). Click for big version.
3PM surface observations
Conclusion: it was about to get colder, winds would increase and the threat of icing was about to increase rapidly. At this point, all equipment needed to be used on elevated structures immediately, and when those were cleared and salted (solid not brine), then the other roads could be dealt with. Metro needed to pull all their articulated buses off the elevated roadways at this point...in fact, they should have replaced with with the non-articulated fleet hours before. Rapid deployment when the weather was clearly going to change seems to me the only viable option.
Question: Do any of the roadway surfaces in the city have real-time temperature measurements? If not, shouldn't this be fixed immediately?
In summary: I am suggesting a combination of the intensive use of all observational assets to provide a real-time view of the weather and road conditions around the city, and the flexibility to use the limited assets of the city and WSDOT to hit the big threats first, guided by these observations. The current state-of-the-art of weather forecasts are good enough to give a heads up, but not good enough today to guide hour by hour actions.
