wind prediction request for FM Godsey

[windfind]

Hi Gang,

Mondays are my day off but I received a lot of e-mail saying Claire's forecasts were delayed. Our IT had made some changes to the our publishing system and some west coast and east coast forecasts never made it out to customers. I seems to be fixed now.

[isobars]

Given a conflict, I find the blue line on the dynamic forecast more accurate than the text forecasts. Even when the line's amplitude is low (and other forecasts give us a heads up on that), its shape and timing are very useful.

[koogzah]

[isobars]

[kirk]

Hi Mike
I too would love to see the standard deviation of the wind and the peaks and lulls too as throwing away data seems a waste.

Also, what would it take to get accurate peaks and lulls for SFO? Those graphs are often pretty good for what we get outside at Coyote but all it publishes is the occasional peak wind and some sort of average which I believe is from the free info from SFO or NOA.

thanks for making the constant improvements. I remember when I paid $400 for a scanner so I could get the wind reports to pilots landing at SFO from my home and the shore which was a huge improvement over the report every three hours from the weather radio for predicting what might happen at Coyote....

Kirk out

[allen]

Statistics are nice and a macro that pops up under the windgraphs that allows you to compute the std deviation for a given window of wind would be great but this does not replace local knowledge of your preferred sailing spot.

Take Cabrillo Beach for instance (insert joke). Here the wind on a good day blows from 14 to 24 mph implying an avg of 19 (and a high std deviation) which the Iwindsurf computations pretty much back up. But the avg is not all that meaningful. What really matters is the gust level (duration and velocity) it takes to get on a plane and stay on a plane. For me that means unless the high is 24-25 mph I'm not going to be getting on a plane all that much nor am I going to be going as fast as I'd like to. And when the high reads 20-21 and the low is reported at 15 or 16 (like yesterday)which again implies an avg of 17 or 18 and a low std deviation, well, you're being lied to as Cabrillo is like sailing on gusty inland lake where the low is almost always about half of the peak gust. Thus when the high is 20, the low is most likely 10 or 11 keeping the avg to 15 or so, not the 17 or 18 the iwindsurf meter is reading.

Seal Beach on the other hand, just down the road from Cabrillo, is a different world (steadier, less gusty wind) when it comes to interpreting the numbers. Here, under much tighter wind ranges, say 16-22 with an avg of 18 you're moving much faster and planing more often than you would at Cabrillo with the same readings.

So crunch all the numbers you want but knowing what a meter reading means at a particular beach will tell you more than detailed statistics.

[tomg]

''''''Think about life before iWindsurf, and thank the Heavens it's here! Imagine just driving around to see if it's windy somewhere, especially if you live in the Gorge, or Bay Area! Yikes!
_________________
Spennie the Wind Junkie ''''''


Yep, i remember those days! We had three "wind talkers": @Berkeley, Crissy and Candlestick, if memory serves. And so i would have given up the sport years ago if not for the iW. When my kids were little, i had zero time to drive around SF Bay Area roads looking for wind. And still don't.

For me the forecasts are less about accuracy at a given site then about deciding go/ no go. (though i DO like to see that red on the graph!)

[leverich1]

Standard deviation is only going to be informative if the wind is normally distributed. As allen said above, sometimes the mean is meaningless (ha) and things like the lull and gust are more important. Even if the "average" windspeed is 15, that doesn't mean that it ever blows 15; it might only blow 10 and 20. Standard deviation isn't diagnostic for this situation.

You can beat around the bush as much as you like and present computed summary statistics like SD, median, skew, kurtosis, etc., but its always going to throw out a ton of information, or be hamstrung by assumptions about normality.

What I'd like to see rather than SD or any other computed statistic is a histogram of windspeed and direction. A histogram is compact, retains much of the information in the signal, does not depend on any assumptions about the normality of the signal, is visually pleasing, and its easy for your brain to pattern match regular or unusual shapes (i.e. bathtub vs. normal distribution).

Anyway, the trees are swinging back and forth outside, I guess its time to start packing the car.

[loopless]

One question for Mike.
Does iWindsurf feed the results of their sensors back into their computer models? In theory, that should make their predictions get (much) better over time. But it seems like this is NOT the case.

For example, here in San Diego, the models generally predict a weak N/NE wind at Tourmaline in a building Santa Ana pattern. Inevitably , instead we get a very nice 15-20 NNW side shore. I know that the "un-guided" computer models can never predict this as it is a localized event controlled by local geography. However, if the historical sensor readings were fed back into the computer models then a "guided" model should over time be able to predict this.
But I think most people would agree that the accuracy of iWindSurf predictions has remained pretty static for as long as I can remember, so it seems like the simulations are not guided by any feedback mechanism.

[tomg]

The iW forecasts have been a very important part of my life: I married a woman named Flow, our son is named Surge and our dog goes by the name Eddy.