Matthew Gregory
*
Jul 1, 2006, 2:26 PM
Post #1 of 3
(2534 views)
Shortcut
|
|
Garbage In - Garbage Out
|
Log-In to Post/Reply
|
|
The following letter was in Scuttlebutt 2126 for which I would like to comment:
* From David Brayshaw: The just-completed Newport-Bermuda Race saw almost all of the "A" list (big, professionally crewed and/or navigated) of yachts go well east of the rhumbline and come up way short of the (mostly) amateurs who went of the rhumbline.
I have applied the wind and current data to do routing/ tactical analysis, illustrated by the track of Bella Mente, first-to-finish. From this perspective, the result wasn't all that surprising. Your readership may be interested in analysis and graphics found at http://www.goflow.com/nb06race.htm
I would like to add some commentary to David Brayshaw’s comments on his routing solution for the Bermuda Race this year. I taught a new North U series of courses this winter on instrumentation and the Expedition software package. The concept of “Garbage in gives Garbage out” was a driving theme of that course. David’s posed routing solution is generated with ‘perfect data’. If after the race, one takes the actual wind observations from a race, the winning route will be produced, by definition. However, predicting the wind, before the race, can be phenomenally complicated, particularly this year, with the variable weather scenarios that each strategist had to decipher during the race. Each had to ask themselves, based on the computer modeling and their own intuition; Would the model over or under predict the speed of the high as it traveled through the course? Would it take a more northerly or southerly route? What would the shape of the isobars be as the fleet neared and then crossed the high? and How accurate are my current predictions? Computers aren’t that smart. The only thing they do better than us is crunch numbers. Thus routing programs have not replaced the strategic sailing brain that the modern day navigator brings to the race. They are algorithms that amalgamate wind, current, course and boat performance polars to derive situation analyses based on those assumed inputs. They do not give a 4 day long “yellow brick road that leads to OZ (Bermuda in this case)” guide for the team to sail the boat along. Instead they give strategists the computational power to play out all of the ‘what if’ scenarios that they can dream up.
The route is only as good as the accuracy of the inputs and this was one of the KEY elements in the routing solution for the Bermuda race this year. Because of the high pressure system, the timing of each boats entry into the gulf stream and the current driven “vortex of victory and despair” (as we called it on our boat) that lay on the rhumb line, the router really plays the roll of ‘scenario checker’. The ‘pro navigators’ that David refers to that went East probably ran ten to twenty wind and current scenario’s through their ‘router’ and then weighed the risk that that team wanted to take. Was finishing first worth the risk of finishing last? And how much leverage do you want to give your main competitors? To illustrate this point, I let out a big “OH NO” when I looked at the position reports and realized that some of the boats in our class were not slowed down by the light winds to the west, as I anticipated, and were able to get through the northerly flowing currents on the west side of the course. On day 3, I thought that those boats had a great chance of winning our class if the center of the high moved quick enough to allow a westerly component to the wind in the last 24 hours. However it was a boat on “our side” of the course, to the east, that had the last laugh and pulled off a victory since they had a better angle to Bermuda in the light south easterly winds in the last 24 hours.
Matthew Gregory
http://www.matt-gregory.com
|
SCUTTLEBUTT
SEARCH
WHO'S 
LOG IN