Ventusky weather app

Kerenza said:
Accurate as in I just spent three weeks away, doing my own forecasts while cruising. I don't do speculation.
Not 100% I agree, but enough for me to know if I could sail or had to motor in the right direction..
Click to expand...
Sorry but unless you do substantial logging and graphing of all the models you don't actually know - science has known this for a very long time. Not an easy task, actually very difficult - I gave up.

If you did all this then apologies - perhaps you could post the data?

This the site?
https://www.squid-sailing.com/
Seems to be down.
 
Last edited:
Look,
I'm not here to sell Squid. I simply looked for the best programme I could and chose it.
The results are pretty good, nothing is perfect. I can only report my findings in use.
Million times better than just opening XC and reading the numbers.
Time consuming, might not suit everyone.
 
Kerenza said:
https://www.squid-sailing.com/en/
Click to expand...

Won't open for me :(

But will check it out once it's back up, though from playing around with the likes of predictwind and a lot of logging I'm with Frank, GFS is as good as any.

Does look that you can overlay sat/synop/grib - very interesting feature!
Opencpn's display of gusts as well as wind is a great feature as well, shows how grumpy the fronts are.
 
as I said, not here to sell it, but never had a problem with two windows 10 computers, one at home other on the boat network, two android tablets and sony mobile.
it does grib 1 and 2.
 
just to be accurate, the mobile apps are different to the pc programme.
I've never tried opening an old grib from a different source, but have exported in a different format for other routing programmes, although the philosophy is to use a integrated solution such as Adrena or Expedition.
i will be investigating the possibility of the bandg plotter downloading directly using the squid account.
in case anyone still has any interest in this thread, grib edition 2 was passed for use in 2001, and offers more facilities, including ensemble content and compression options.
 
Last edited:
Kerenza said:
Again I think you are unfamiliar with the power of the programme.
Comparing like with like, i.e. the typical sailor with XC and perhaps one other model, his minor ensemble outcome over two days is likely to be more accurate. Of course when using squid in anger, the ECMWF ensemble will be that much more reliable. (And costly)

I think you must agree to differ with the squid (great circle) experts.
As providers for major international races, used constantly by the top navigators in racing, they have a reputation to uphold and contiuously improve.
To say nothing of the revenue they can command.
Click to expand...

Experts? On what? A competent computer expert can obtain weather model output and process it into an attractive and, no doubt, useful format. They can feed the data into weather routing software. That does not make them weather experts. They can obtain the coding for models such as the WRF, input data from the GFS or elsewhere, run the model but still have little understanding of NWP. Over the years, I have engaged in discussion with four private firms providing weather information. Late last year, one tried to tell me that accurate representation of the jet stream was not important in weather models! A long time ago, one even disputed the need to calculate weather throughout the whole of the atmosphere.

The Squid statement about 3 or 4 NWP models coming up with similar answers being significant demonstrates a clear lack of understanding of NWP and ensembles. If you do not believe what I say at http://weather.mailasail.com/Franks-Weather/Ensembles about ensembles, then read https://www.ecmwf.int/en/about/media-centre/fact-sheet-ensemble-weather-forecasting and https://www.metoffice.gov.uk/research/weather/ensemble-forecasting.

I am not suggesting that you should not use Squid, PredictWind or any other service free or not. I am just saying that statements they make are not always correct. I could talk more about Predictwind than Squid, but the Squid statement about ensembles is echoed by PredictWind. They are wrong.
 
and as i inferred, experts very often agree to differ, or at least in my field they did.
dont overlook the practical sailor at sea, now, with limited resources versus the scientist with retrospective unlimited resources.
its all about making live decisions.
 
Kerenza said:
and as i inferred, experts very often agree to differ, or at least in my field they did.
dont overlook the practical sailor at sea, now, with limited resources versus the scientist with retrospective unlimited resources.
its all about making live decisions.
Click to expand...

I am not decrying services provided by Squid or Predictwind although I would never recommend PredictWind. I can get better information free elsewhere. I do, as a (former) professional meteorologist object to their erroneous statements.
 
i think we have common agreement regarding ensemble forecasting?
a control forecast with multiple reruns using slightly different starting points.
for the sailor on the sea, xc say plus another easily accessed model can give an improved confidence for this afternoon or tomorrow morning.(depending on model run times)
what they cant do is investigate the topographical and local temperature effects, as indicated by a higher definition model, let alone a hd ensemble.
its getting that info easily and adjusting observationally that's important to the sailor.
(I aso persevered with predictwind for some time, until I felt there were enough reasons to look elsewhere)
 
Last edited:
Kerenza said:
dont overlook the practical sailor at sea, now, with limited resources versus the scientist with retrospective unlimited resources.
its all about making live decisions.
Click to expand...

Indeed. Though getting a handle on the accuracy of the data is very difficult. I know I am easily fooled into thinking what I get is the best and someone could sneak onboard, switch from GFS to ECMWF and I'd be none the wiser still seeing what I want to see.

For us non ocean racers I doubt that you'll do much better in the real world beyond GFS & opencpn grib/wfax/sat pics if you can get recent ones. Plus a little experience of what goes on round headlands, sea breeze, coriolis etc.

Probably near impossible to test though :)
 
Yes, agreement that data accuracy is so important.
What comparative grib programmes give you is the ability to interpret the data and observe the surroundings.
So my example yesterday was important because of the differences between land based forecasts, in that case XC and what happened in between the two centres.

Xc and Arome HD both gave 9 knts or so for Cardiff and Weston, directly opposite each other across the Bristol channel.
Arome HD gave 18knts in the channel between them.

Ive now checked observation on flatholm which gave 15knts, half an hour later than predicted.
None of the numbers are exact, the timescale is short, but the feeling was correct.
 
Kerenza said:
Xc and Arome HD both gave 9 knts or so for Cardiff and Weston, directly opposite each other across the Bristol channel.
Arome HD gave 18knts in the channel between them.

Ive now checked observation on flatholm which gave 15knts, half an hour later than predicted.
None of the numbers are exact, the timescale is short, but the feeling was correct.
Click to expand...



One off readings don't tell you much. Arome was hours out and higher than actual. (MPH in the graphs & UTC) .

Great fun having a good dig into the data though :cool:


MkF8x9F.png


skrKMxG.png



Yesterday 6am GFS grib interpolated for flat holm.
BTZTv6V.png
 
Last edited:
Ensembles.

One ensemble type is that used by NOAA/FNMOC. The deterministic forecast is the combination of model formulation plus analysis software scheme tuned to give the best results for the needs of that centre. The ECMWF approach introduces the uncertainties in the model formulation. Ensemble results give an estimate of probability of winds or whatever being within certain limits. They do no more than that.

Short period forecasts.

When it comes to detailed prediction, I am cynical about usefulness of any model. Any grid of points can only define shapes of topography and weather on a scale of about 4 or 5 grid lengths. The finest grids currently used are about 1.3 km. For purely topographic effects that places a limit on calculation of straits and headlands. More importantly, it places limits on the ability to predict detailed weather.

The concept of predictability is a more serious limiting factor. Even if we knew exactly the weather on a 1 km grid, by the time the data had been collected, assimilated into an analysis, the forecast run and results disseminated, small detail defined by the grid would have disappeared. Looking at the Meteo France website, I gather that AROME runs three hourly. That limits usefulness in any deterministic sense.

In some weather situations, topographic effects can be well calculated especially in the very short term. However, it only needs a small forecast error in cloud to affect greatly sea breeze behaviour; a slight change in the wind can greatly impact on the local wind; a shower can give a totally different wind to that predicted.

If Squid or any other system is predicting such detail as what I think you are describing, they will get it right sometimes and wrong at others. Of course, if it is really very short term, then the Mr 1 human eyeball is an invaluable tool. It will see detail that neither models nor human forecasters know about.

Weather and Cruising.

But, I wonder if we are to some extent at cross purposes. I am not usually greatly concerned about small, local short term detail. If that did concern me, then I should not have been where I was. My main concern as a Cruising yachtsman is whether or not I should be going to sea and, if so, where to. In my 50,000+ miles around the Channel, Biscay, Iberia, Mediterranean plus 2000 or so Caribbean charters, I have never felt a need for weather routing. That is so across the Channel and even across Biscay. Basically, I do not go if it seems likely that we will have to beat - gentleman do not. Given the uncertainties inherent in the weather and the consequent uncertainty in boat speed, I can do quite adequately with conventional passage planning.

I am not an ocean sailor, 3 or 4 days has been my limit but I will quote Stan Honey, a winning RTW navigator experienced in the use of GRIB data.. “In my slow 40 footer, I would wait for a weather window of about 3 days to get me clear of land and then take whatever came.” On an ocean voyage, in a slow moving vessel, if a storm has your name on it, it will get you. The vessel will be too slow, the weather system too large and forecasting beyond 5 or 6 days just not good enough.
 
Last edited:
I think some time ago in this thread the different aims were quite apparent. I regularly sail and race in waters that the short term local differences mean the difference between going out or not. Or indeed the race course, with particular effect of tide. Locally we have much folklore about currents related to tide times. My personal goal is to rationalise these traditions using modern technology.
And that's where there is some confusion in the thread.
To confuse science with technology.
Technology is a vehicle of enablement, often ignorant of the science which is its foundation.
While I hope I'm not ignorant of the science, I'm more interested in the technology helping me make decisions.
Although the Bristol channel example was a hyperthetical test, I had no intention of sailing that day, it is what I would have included in my passage plan.
And it was sufficiently different from any simple on line forecast to have made a difference.
I reef at 12knts, and I would have been looking for observational indications in readiness.
From that perspective the single example is far more important than the probability calculations the data might suggest . I'm afraid it's just another tick to support the operational experience.
 
Kerenza said:
I
Although the Bristol channel example was a hyperthetical test, I had no intention of sailing that day, it is what I would have included in my passage plan.
And it was sufficiently different from any simple on line forecast to have made a difference.
Click to expand...

Yet from the actual data that afternoon the GFS model did better than the Arome model. The windspeed were not double in the channel.

Arome hd gives the same forecast on land as XC, but wind speeds double that in the channel.
Click to expand...


Frank has said many times that the wind itself doesn't know what it's doing to within a Beaufort force. That's very likely a more accurate forecast than you'll get from a model :)
 
Well, I started the thread by drawing attention to Ventusky and its ICON-EU updated 3 hourly. On another thread, I have drawn attention to COSMOS-DE and HARMONIE LAMs.

I am sympathetic to the idea of using technology to help decision making. As I have consistently said, i object to the making of spurious claims and statements in order to sell such services. With Squid as with PW, there are statements that are simply incorrect or misleading. I will give one para from https://www.predictwind.com/.

“PredictWind is the only company in the world that runs their own global weather model at 50km resolution with 2 sources. In addition 1km / 8km resolution forecasts are generated for popular regions around the world.”

First, such a global grid length is 2 generation in the past. To make that statement as though it is a big plus is correct but misleading. Secondly, using two different starting points is a 2-member ensemble. Meaningless but implied as a useful tool. Thirdly, as I understand, they use that model to start their 8 and 1 km models. Claims to predict accurate detail without accurate detailed data input are not well founded.

I have no problem in PW selling their service as being easy to use with good graphics etc. It will not greatly mislead as long as you remember that a 50 km model will underestimate strong wind more than a model with a 12 km grid. I would not use it myself because I can get free forecasts produced by more modern models updated 6-hourly, even 3 hourly using Ventusky. PW only updates 12 hourly.
 
Thats all good, just don't tar all programmes with the same brush.
At no point does squid give a forecast in the same way.
Data is presented for interpretation by the user.
That's why these navigators get paid so we'll - until they get it wrong of course :)
 
You must log in or register to reply here.

Other threads that may be of interest

capnsensible
Worlds toughest row. Atlandic Edition.
2
Replies
27
Views
651
ridgy
R
zoidberg
A bit breezy?
Replies
2
Views
309
zoidberg
zoidberg
Cerebus
Why has Guy Cotten search page not listing the old”shopping” icon?
Replies
13
Views
278
bluerm166
B

Share this page

Top