Whistler Heli-Decision-Making Dilemna
Hey guys…interesting decision to be made yesterday by Whistler Heli-Skiing.
60+ customers jacked about deep Kootenays-style powder on a day where a 500mb shortwave trough (SW) was forecast to pass overhead of the Whistler area sometime around mid-day. There was high confidence that skies would clear but the critical question: exactly when? WH needs a four-hour window ending 90 minutes prior to sunset. In late January, the hard close of that window is about 3:30pm. As it turned out, WH bailed when flurries got REALLY HEAVY at about 11am. Skies did not go SCT-BKN until about 2pm so that was a very good call. (I spoke to PSPC Supervisor Chris Gibbons at 11amish and he suggested mid-afternoon)
Here’s the question: Could that decision be made with confidence earlier? Say at 6:30am or perhaps by 8am?
Have a look at this video:
Do you think the forecast position/timing of the SW suggested more of a mid-afternoon clearing vs a mid-day clearing?
Is the negative tilt of the SW a clue to the timing or the manner of the clearing? i.e had the SW a more positive tilt, would there have been a more rapid flush with subsidence immediately behind the SW and therefore a ‘cleaner, faster’ clearing?
I noted that the 4am HRDPS SpotWx forecast for Blackcomb Peak kept the flurries going for about 2 hours longer - until 1-2pm rather than 11-12pm as per the 10pm run from the night before. Significant?
One of the guides asked me what resource I would ‘go to’ in a scenario like this…here are my thoughts. Feel free to trample on them.
Being so close to T+0, (short-of some rejected upstream balloon data on the 12Z model update) and calibrating the sat imagery to the position of the SW, we ought to have super high confidence in the model’s timing. However,
is resolution a problem? The model 500mb domain is coarse. EVEN if the model SW forecast was PERFECT in every way, does the resolution of the model allow us to sharpen the timing of SW passage to the degree required? If this SW was crossing the flat, barren Prairies, might we have more confidence in the timing?!!
Could we look for subtle wind shifts at 700mb to sharpen the timing?
It was interesting to have one of the Boston-based clients approach me (when he realized that I was an MT) and show me a Weather Channel animation on his iPhone that led him to believe the flurries were going to persist and thus the probability of a cancellation was high. This is what we (MTs) are now up against: motivated and sophisticated clients drilling into public weather resources to make decisions. I happen to think that is brilliant. It just means that MTs need to up their game and figure out a way to get their expertise out there to help these clients.
Consider this for a moment: How can an MT add value with clients like Boston-guy or WH, who believe they are making the right decision 95% of the time ON THEIR OWN?!! I think the only way MTs can help that guy from Boston make a more informed decision is through context and through real-time engagement/discussion…e.g. how have the models been performing lately? How does THIS model perform in THIS scenario? What about ensemble techniques and assessments?
I am not sure that anyone can help WH move to 96% success, but I am open to argument. As MTs, we need to answer the bell and FIGURE IT OUT, lest we lose our livelihoods in the same manner as truck-drivers.
dj
On 28 January 2018 at 13:52, Reg Dunkley wrote:
Hi Dave.
Glad to hear that you are still passionate about the art and science of forecasting and I enjoyed checking out your blog. In answer to the question
"One of the guides asked me what resource I would ‘go to’ in a scenario like this…"
Here is my simplistic info strategy keeping in mind that I am an old-fashioned forecaster who is not current with the plethora of advanced info tools now available:
1) In the interests of expectation management, I would remind the Heli Ski guide that the precision of his needs may exceed the current ability of the science.
3) Encourage the guide to compare the graphics from this site to corresponding satellite cloud imagery.
I examine the Astronomy Cloud Forecast product every day and although not perfect I find that it does a remarkable job. I have used it to successfully predict frontal passages and wind shifts within an hour with a lead time of over 40 hours.
It just so happens that on Saturday afternoon in celebration of their 150th anniversary the Royal Astronomical Society of Canada was holding a national event that through the internet connected 29 centres across Canada. Given the extensive period of wet weather preceding the event, even a fleeting glimpse of the Sun or Moon would be worthy of celebration. In this case with a lead time of 84 hours the GDPS models
suggested that there maybe clearing in the wake of a vigorous low-pressure system that would traverse the coast near mid day Saturday. The attached graphics show a good agreement between the observed visual satellite imagery and the 34-hour Astronomy Cloud forecast!
In this case the forecast encouraged us to deploy specialized solar telescopes that allow the public to view solar prominences arcing from the limb of the Sun. As it would happen only centres in Newfoundland and Victoria witnessed the Sun that day and the remaining centres were shrouded in cloud. (It would appear that astronomers are lousy climatologists).
Even as recent as the mid nineties lead times of 84 hours and reasonably accurate predictions of cloud amounts 34 hours in advance would be astonishing. I think that some time between ~1995 and 2005 there was a remarkable improvement in model initialization, but I do not think that it has been identified as to what contributed to this advance. Let's not forget, however, the great progress that has been achieved. There are applications that generate similar cloud amount forecasts from a mind-boggling array of other models, but the half life of my memory is too short to conduct an inter comparison and I adhere to a simplistic single or double model strategy.
Notice that I have not mentioned Mesoscale forecast products. Their output is very convincing and seductive but in the past, they have seemed to be noisy and have led me astray and so I gave up on them. Should I give them another try?
I have given weather forecasting presentations to the Astronomical Community and championed the use of the Astronomy Cloud Forecasting product to optimize observing opportunities. One feedback that I did receive is that even an relatively scientifically advanced community finds the conversation from UTC to LST a sufficient speed bump to discourage the use of this product.
Take care and I hope that you are enjoying retirement Reg As Arthur C Clarke said: Any sufficiently advanced technology is indistinguishable from magic. We live in magical times
PS…Regarding the performance of mesoscale models, way back when, they seemed to be quite vulnerable to the initialization and possibly amplify noise embedded in the Pacific data void. The output, however, was so visually compelling it was difficult to dismiss. Things have improved greatly over the past few years. It would be interesting to learn what others have experienced. Perhaps another topic for your blog.
Comments
Post a Comment