jarcher

PredictWind vs Sail Flow vs Whoever - The sequel

Recommended Posts

Eight years ago the consensus here was that Predict Wind was the better product:

We still have Predict Wind and Sailflow, and also Windy. Do you all still feel Predict Wind is the way to go for near shore buoy racing? I used Predict Wind for Narragansett Bay several years ago and stopped paying because I didn't find it accurate. I looked at Predict Wind last and was not impressed, but that was just one evening. Is it worth a second try?

Share this post


Link to post
Share on other sites

I always thought Predict Wind was more geared for cruising/offshore passage making - bundled with various sat solutions, gives ppl access to weather data if you're planning to be off the grid for more than 5 days.

Not sure why one would use it for near shore buoy racing in any part of the civilized world where 24hr meteorological information is generally free, accessible and pretty accurate.

Share this post


Link to post
Share on other sites
11 minutes ago, jarcher said:

Eight years ago the consensus here was that Predict Wind was the better product:

We still have Predict Wind and Sailflow, and also Windy. Do you all still feel Predict Wind is the way to go for near shore buoy racing? I used Predict Wind for Narragansett Bay several years ago and stopped paying because I didn't find it accurate. I looked at Predict Wind last and was not impressed, but that was just one evening. Is it worth a second try?

What have you been using since ditching Predict Wind?

Seems like almost any service including the free ones are equally good as long as you don't sail somewhere that has a lot of micro climates.

I use PW quite a bit for SF Bay.  I've been happy with it for the most part but they're guessing just as much as any of the other services.  But it does a decent job of forecasting the hundreds of micro climates inside the bay.  Maybe it's just me but at least for our area the NOAA marine forecasts seem to be more accurate in the last 2 or 3 years.  They're still completely worthless beyond 24-48 hours out though.

Share this post


Link to post
Share on other sites

I run a 34' powerboat from Annapolis to Palm Beach and back every year. The 400 miles or so offshore (and the open sounds) requires a good look at the wind, direction and waves. Meteo Earth has been the most reliable and detailed enough where I can avoid small areas of nasty seas. 

For our last bay race sailflow & predict wind were out to lunch

Share this post


Link to post
Share on other sites
1 hour ago, Miffy said:

I always thought Predict Wind was more geared for cruising/offshore passage making - bundled with various sat solutions, gives ppl access to weather data if you're planning to be off the grid for more than 5 days.

Not sure why one would use it for near shore buoy racing in any part of the civilized world where 24hr meteorological information is generally free, accessible and pretty accurate.

Our government weather forecasts on the left left coast (Canada) are consistently incorrect, generally overstating wind speeds by huge margins. PW seems much closer to reality.

A thread:

 

Share this post


Link to post
Share on other sites
12 minutes ago, Ishmael said:

Our government weather forecasts on the left left coast (Canada) are consistently incorrect, generally overstating wind speeds by huge margins. PW seems much closer to reality.

A thread:

 

 

What happens when you use American sources instead of Canadian? It would be an interesting topic to get a meteorologist on the line at the local center - might be a public policy decision to err on the side of caution.

Share this post


Link to post
Share on other sites

This last weekend they ALL blew it.  On SF Bay anyway. Totally missed some weird upper level disturbance that shut off the flow in the north bay

Share this post


Link to post
Share on other sites

I use Sailflow for current conditions because they have the most stations around the PNW and their WRF forecast for "General" conditions especially long term (+4 days)

Predictwind does not seem to have enough stations here. Windy is cool, but hard to figure out what the speed is where you are on a small screen in the sun on the rail. (Is that dark or light green? Maybe light blue? )

Our local university has a very accurate forecast, but the hi-res only goes out 60 hours, only runs twice per day and takes 6-7 hours to run the full forecast. So this mornings forecast hasn't finished all 60h yet and some of it has already happened. 

That said its right at least 85% of the time and most of the rest is often timing. As Ismael stated the general forecasts here are often too high and to wide coverage to make sense . Saw a NWS forecast for Puget sound that said 5-25 becoming 10-15. Really they should just say its 5 100 and call it good. They would be right. 

Share this post


Link to post
Share on other sites
53 minutes ago, Miffy said:

 

What happens when you use American sources instead of Canadian? It would be an interesting topic to get a meteorologist on the line at the local center - might be a public policy decision to err on the side of caution.

American sources are for American weather and they don't go far up the Canadian coast. In addition, frequently we get weather before they do. Also, the direction and strength of incoming winds changes a lot depending on what the highs and lows out in the Pacific are doing. Absolutely they are in CYA mode with the forecasts.

Sometimes major winds come in as a major surprise, apparently to the meteorologists as well as to the public. We had a big one come in with about 15 minutes notice on VHF, sent a few anchored boats and boats on moorings up on the beach and gave us a knockdown out of nowhere. We wondered why the Tayana 37 in front of us was practicing his reefing techniques on such a nice light day. When we saw his mast dip to the horizontal and the top of his keel at the surface we figured out something was afoot. Caught us with full main and a 180 up, big fun.

8ca8713d-bc1e-4a84-9f00-4ff8400bd468.png

https://earth.nullschool.net/#current/wind/surface/level/orthographic=-124.45,49.25,3000

Share this post


Link to post
Share on other sites

I do WxRouting for a living. Predict Wind is an invaluable source of high resolution modeling. I combine the 1km resolution data from PW with real time observations and then take the five model solutions over the previous four-five days worth of runs and throw out the poor performers and tweak those that have best performed into a blend. This information is then combined with personal experience if I have any in the area and the result is usually a pretty accurate forecast and race plan with a good chunk of the data coming from Predict Wind Pro. Just my $.02, your mileage may vary.

  • Like 1

Share this post


Link to post
Share on other sites

I could be wrong here but as I understand it all of these tools utilize the same underlying forecast data/models for a given location (GFS, NAM, etc.). The differences then become in how they interpret the data and graphically display the data.

Share this post


Link to post
Share on other sites
1 hour ago, Carrera said:

I could be wrong here but as I understand it all of these tools utilize the same underlying forecast data/models for a given location (GFS, NAM, etc.). The differences then become in how they interpret the data and graphically display the data.

 
From Rich Clayton-
There's a lot of differences between numerical models, so I'll just (briefly) mention the differences between GFS and ECMWF.

First both are global models, meaning they produce a forecast for the entire world.  There are also regional models like MM5, NAM, and WRF that produce forecasts specific to a region at a high resolution (I'll talk about resolution later).  Regional models tend to seed their forecasts from global models; GFS provides context for MM5.  I'm not sure what regional models use ECMWF, but I'm sure there are quite a few.

GFS and ECMWF use different equations for predicting atmospheric conditions.  GFS is a hydrostatic model and ECMWF is nonhydrostatic.  Hydrostatic models use pressure as vertical coordinates in the forecast (e.g. make forecasts at specific isobars) and essentially make general approximations about topography.  This means it does poorly at higher resolutions in which topography will have greater impact in the overall forecast.  The ECMWF, on the other hand, uses altitude and more accurately accounts for topographic effects on weather systems.  I'm not a numerical meteorologist (I wish I could give you a better answer); just know that ECMWF is generally more accurate (http://en.wikipedia.org/wiki/Atm...).

Next, both models differ in availability.  GFS runs 4 times a day, ECMWF runs twice a day.  This means GFS has the opportunity to recover from bad forecasts (say, due to bad input data) by producing more per day.  ECMWF generally outperforms GFS, but if one of the forecast runs was bad, you've got to wait longer for new data.  The reason for the difference in availability is due to the time it takes to compute the model.  Nonhydrostatic models are more computationally expensive, which is why ECMWF is the only global nonhydrostatic model used for operation meteorology.

GFS is free, ECMWF is not. You can get the latest GFS runs for free from NCEP (http://www.ncep.noaa.gov/).  You can even download the model and run it on your home super computer if you are so inclined (http://www.nco.ncep.noaa.gov/pmb...).  To get ECMWF data, you typically need to be affiliated with some organization that pays for it.

Both models produce different products. You will probably be able to get the typical forecast products (surface, 850MB, 700MB, 500MB, etc.), but there are also some unique products produced by both (it's been 6 years since I've used them, so I can't remember what they are anymore).
  • Like 3

Share this post


Link to post
Share on other sites
20 hours ago, Hugh Jorgan said:

What have you been using since ditching Predict Wind?

Seems like almost any service including the free ones are equally good as long as you don't sail somewhere that has a lot of micro climates.

I use PW quite a bit for SF Bay.  I've been happy with it for the most part but they're guessing just as much as any of the other services.  But it does a decent job of forecasting the hundreds of micro climates inside the bay.  Maybe it's just me but at least for our area the NOAA marine forecasts seem to be more accurate in the last 2 or 3 years.  They're still completely worthless beyond 24-48 hours out though.

I have just been using generic forecasts which are great for telling me if I should think about bringing my rain gear or if I need to be on the watch for a sudden T-Storm. 

I'm not sure exactly what a micro-climate is, but it sounds like something we would have a lot of in Narragansett Bay. 

Share this post


Link to post
Share on other sites
18 hours ago, DryArmour said:

I do WxRouting for a living. Predict Wind is an invaluable source of high resolution modeling. I combine the 1km resolution data from PW with real time observations and then take the five model solutions over the previous four-five days worth of runs and throw out the poor performers and tweak those that have best performed into a blend. This information is then combined with personal experience if I have any in the area and the result is usually a pretty accurate forecast and race plan with a good chunk of the data coming from Predict Wind Pro. Just my $.02, your mileage may vary.

Okay great, but I'm mostly interested in what is likely to happen, say, this Saturday afternoon? What will the velocity be, is it expected to build or fall, will it go left or right? That kind of thing. I understand of course that we have to track the shifts while we're out there. My question is, can any of these tools be helpful for local racing where I'm not circumnavigating the world?

Share this post


Link to post
Share on other sites
2 minutes ago, jarcher said:

Okay great, but I'm mostly interested in what is likely to happen, say, this Saturday afternoon? What will the velocity be, is it expected to build or fall, will it go left or right? That kind of thing. I understand of course that we have to track the shifts while we're out there. My question is, can any of these tools be helpful for local racing where I'm not circumnavigating the world?

I used Predict Wind on the Great Lakes last summer, and was not particularly impressed. The PW models almost always understated the wind speed. The ECMWF model overstated regularly, and the timing was out of sync. The GFS model seems to be the most accurate for my region, and is available for free from several sources.

To answer your question, no weather forecast will be accurate enough to win you a buoy race. I have not sailed in Narragansett Bay, but I would guess that the influence of surrounding land is going to make a larger difference on what wind you experience than any major pressure system changes within the hour or two you are racing.

For something like a Wednesday night buoy race, I usually look at the chart attached in the morning. Then check your favorite weather station all day, and note changes.  If you have an all-day Saturday race, then watching a GRIB can be helpful for being aware those changes are coming - but only for the big picture. As far as which one is most accurate - watch and compare regularly for days or weeks. 

92fndfd.gif

Share this post


Link to post
Share on other sites

I would at least like the model to help me to determine which rig setting I should use (pinned rig)

Just looking out 24 hours, the various models within 1 app, sailflow, are calling for peaks of 12-14 or peaks of 26+.  For 1 forecasting location.

Ah well.

Share this post


Link to post
Share on other sites
21 minutes ago, Blizzard said:

To answer your question, no weather forecast will be accurate enough to win you a buoy race. I have not sailed in Narragansett Bay, but I would guess that the influence of surrounding land is going to make a larger difference on what wind you experience than any major pressure system changes within the hour or two you are racing.

Of course not, it's just another tool. And yes, land effects are complex, hence my desire for some tool to help. I'm not at all a weather expert, far from it, but as a computer guy I do understand the difficulties of modeling complex systems. Perhaps this area is just too complex to model reliably.

Share this post


Link to post
Share on other sites
27 minutes ago, jarcher said:

Perhaps this area is just too complex to model reliably.

This is my opinion of almost all bouy racing locations, as we tend to race in protected waters. The basic idea is that hot air rises over land, because the hot sun warms up dirt or concrete or whatever. This rising air disrupts the pressure gradient wind. Because the pressure gradient is different every day, the effect of disruption will also differ. 

Predict Wind claims their algorithm accounts for land influence, but the size of our bouy courses are too small for a world-wide algorithm to figure out.

Check out Wind Strategy by David Houghton. It's a short book with lots of illustrations, and the rules described seem pretty accurate.

Share this post


Link to post
Share on other sites
2 hours ago, DryArmour said:
 
From Rich Clayton-
 
There's a lot of differences between numerical models, so I'll just (briefly) mention the differences between GFS and ECMWF.

First both are global models, meaning they produce a forecast for the entire world.  There are also regional models like MM5, NAM, and WRF that produce forecasts specific to a region at a high resolution (I'll talk about resolution later).  Regional models tend to seed their forecasts from global models; GFS provides context for MM5.  I'm not sure what regional models use ECMWF, but I'm sure there are quite a few.

GFS and ECMWF use different equations for predicting atmospheric conditions.  GFS is a hydrostatic model and ECMWF is nonhydrostatic.  Hydrostatic models use pressure as vertical coordinates in the forecast (e.g. make forecasts at specific isobars) and essentially make general approximations about topography.  This means it does poorly at higher resolutions in which topography will have greater impact in the overall forecast.  The ECMWF, on the other hand, uses altitude and more accurately accounts for topographic effects on weather systems.  I'm not a numerical meteorologist (I wish I could give you a better answer); just know that ECMWF is generally more accurate (http://en.wikipedia.org/wiki/Atm...).

Next, both models differ in availability.  GFS runs 4 times a day, ECMWF runs twice a day.  This means GFS has the opportunity to recover from bad forecasts (say, due to bad input data) by producing more per day.  ECMWF generally outperforms GFS, but if one of the forecast runs was bad, you've got to wait longer for new data.  The reason for the difference in availability is due to the time it takes to compute the model.  Nonhydrostatic models are more computationally expensive, which is why ECMWF is the only global nonhydrostatic model used for operation meteorology.

GFS is free, ECMWF is not. You can get the latest GFS runs for free from NCEP (http://www.ncep.noaa.gov/).  You can even download the model and run it on your home super computer if you are so inclined (http://www.nco.ncep.noaa.gov/pmb...).  To get ECMWF data, you typically need to be affiliated with some organization that pays for it.

Both models produce different products. You will probably be able to get the typical forecast products (surface, 850MB, 700MB, 500MB, etc.), but there are also some unique products produced by both (it's been 6 years since I've used them, so I can't remember what they are anymore).

With SF you can actually select which of the national or global models to use in the prediction. However the default is the "Quicklook." which is Sailfow's model. On the Web version they also overlay the NWS forecast as a grey box so you can see how they forecasts, and actual conditions relate the NWS.

BUT if you use any of the other models they are pretty limited in time span. GFS 0.5 on SF only goes 3 days. whereas Quicklook goes 10 days. Around here in the PNW any forecast beyond 3 days is guesswork at best. 

the other thing I have noticed on SF forecasts is they under forecast the wind and over forecast the gusts. True wind is in the middle somewhere. 

 

Last evening the local forecast called for 10k building to 15 for 1-2 hours and then back to 10. Timing was 1.5 hours off so we had to wait, but velocity direction and location were spot on. SF showed 5k gusting to 18 which I would interpret to 11-13 which was close. We had a solid 17-18 for a while backing to 12 then back down to 10. Following the SF average on Quicklook was pretty close, but the actual forecast was too wide range. Even looking back at the graphs in SF you can see the difference BTW the Quicklook and reality. 

So I stand by using them with the Caveats of averaging the forecast BTW the wind and gusts. They also have a much better station distribution so we have solid wind data at the very local level in the places we sail. (Premium station right at the end of our breakwater and non-premium 2 miles away.)

Share this post


Link to post
Share on other sites

i lick my finger and then stick it in the air.............

Share this post


Link to post
Share on other sites
1 minute ago, dacapo said:

i lick my finger and then stick it in the air.............

Probably just as accurate. But do you wash your hands first?

Share this post


Link to post
Share on other sites

iwindsurf/kitesurf gives  real time data from navaids coastal and offshore.

The forecasting models seem to be within a 60 min accuracy timeframe.  Have used it for years...

 

Share this post


Link to post
Share on other sites
2 hours ago, jarcher said:

I'm not sure exactly what a micro-climate is, but it sounds like something we would have a lot of in Narragansett Bay. 

Micro climates:

a good example are the wine region valleys that run off of SFBay. Napa/Sonoma in the North, Santa Clara in the South, Livermore to the East. 

the gradient wind comes through the gate, and as it flows over land, warms up going away from the bay. Carneros to Calistoga is about a degree F per mile, so Burgundy near the bay, and Bordeaux farther, going to Spanish/Italian varietals in the extreme. 

Depending on whether you are in a wind shadow (coastal ranges that form valleys are 2000-4000 ft) you get funneled winds.  

On the Alameda/Oakland Estuary you get both funneled winds, Von Karman Streets off hills and buildings and never get Fog, while only 6 miles away you have SF with the fog chasing the tides. 

In San Jose we get NW or SE (latter if we have a cyclonic low offshore) and extremely rarely E or W. 

 

Share this post


Link to post
Share on other sites

I use SailFlow largely, but tempered with the larger picture.  There is too much chaos at the local level near shore to take any of the forecasts as gospel.  Learn the geography.  Watch the clouds and understand what they tell you.

Get the surface analysis and surface forcasts from a much larger view than the local venue to get an idea of the most probable trend that will be coming.  And anticipate if those trends are happening sooner or later than the initial forecast.  They almost never happen on time, but if you watch the environment you will get an idea of when the general trends are happening relative to the forecast.

Share this post


Link to post
Share on other sites

I use all of them and start looking several days out to see what the trend is for the local forecast while trying to identify the influence the trend of the more hemispheric events are having upon the trend of that local forecast.

Sometimes those local forecasts are 100% correct but sometimes they are 100% wrong as far as wind direction and strength, even on the day of the forecast or while it is happening.

I have not been able to identify situations which correctly presage that a local forecast will be wrong; anybody have any luck in that area?

Share this post


Link to post
Share on other sites
On 5/24/2018 at 1:47 PM, Ishmael said:

American sources are for American weather and they don't go far up the Canadian coast. In addition, frequently we get weather before they do. Also, the direction and strength of incoming winds changes a lot depending on what the highs and lows out in the Pacific are doing. Absolutely they are in CYA mode with the forecasts.

Sometimes major winds come in as a major surprise, apparently to the meteorologists as well as to the public. We had a big one come in with about 15 minutes notice on VHF, sent a few anchored boats and boats on moorings up on the beach and gave us a knockdown out of nowhere. We wondered why the Tayana 37 in front of us was practicing his reefing techniques on such a nice light day. When we saw his mast dip to the horizontal and the top of his keel at the surface we figured out something was afoot. Caught us with full main and a 180 up, big fun.

8ca8713d-bc1e-4a84-9f00-4ff8400bd468.png

https://earth.nullschool.net/#current/wind/surface/level/orthographic=-124.45,49.25,3000

You have to understand that they are Canadian government employees and do things by committee. They use models that nobody else does. Never figured that out although I have actually worked with some of them for Southern Straits etc. And I know CBC's Johanna Wagstaffe personally:http://www.cbc.ca/vancouverweather/bio/ But most of the work by these people are all Environment Canada. I have worked with all of this and my good to go models are always Predict Wind. I contacted them as it wasn't working for our local outflow/Squamish thermals and they did fix it. Racing on the Riptide 44 "Dark Star" McKee and the boys have used Expedition real-time localized forecasts. But you pay for that.    

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now