Best online weather forecast/app

[malyea]

My fav is PredictWind...

For eastern coastal US, out to an including Bahamas, it’s the best I’ve found...

Do you have a better one?

Thanks

[chuckr]

When we were on that side of the world 1st was Chris Parker, 2nd gribs off the ssb, we also supplemented with windfinder, wind guru and now our favorite is WindyTy -


We look at as many as 4-5 forecasts for any serious sails and a lot of times they do not agree so we have to make our own based on all the inputs - in particular we watch the forecast pressure and try to determine where it is and where it is coming from along with the lo and hi pressure areas and their potential timing and can we get to where we want to be based on what we see


for a day sail we just take a quick look at a couple

[belizesailor]

Ive used Windy a lot over the past few years. Great data visualization tool. It also has access to the ECMWF model which Ive found to be quite good. Mostly used here in the W Carib.

For digging a little deeper, my favorite on line site is WeatherOnline's Expert Charts:

https://www.weatheronline.co.uk/cgi-...ZOOM=0&PERIOD=

Which has access to the largest number of models in one place that I know of, including ECMWF.

[Suijin]

I use Windy for my first swag at weather. It’s important to look at the gust data. Sometimes it will show wind at a lovely 15 knots when it’s actually gusting to 30 fairly regularly.

[sailcrazy]

Windguru is also a good site.

[Paul Howard]

As above, weatheronline.co.uk, clicking on 'sailing' , then the regions on the left margin, it is my fav, as it covers almost the whole world, maps go to good detain and is great for more remote places.

[DMCantor]

Our go-to weather site is passageweather.com. Quite useful visualization of the forecast, and in our experience, usually matches the weather we actually experience.

[Pizzazz]

The best app for me is SailGrib WR. It has advanced weather routing and seamless integration with Iridium Go or any other source for Grib files including direct download when in cell phone coverage areas. It only works on android but you can get a cheap prepaid phone to use as a tablet. I found it better than Predict Wind by a mile and more accurate. The accuracy is due to the multiple routing options where you can adjust the polars to fit your boat/sails combination very well.

Give it a try.

[pippatheskippa]

It seems the question is not what provider, but what source does the data come from? GFS or ECMWF supply most of the popular forecasts, and it’s important to check others - Passage Weather, Windguru and Windy all use GFS, but just present it in different styles - look at where your data comes from

[sailboatboy]

This is not a forecasting tool but it does give a visual representation of what is currently happening. You can then match to gribs in the future and see the predicted motion. clicking on earth, changes the displayed parameters. I have used the predicted rainfall when in Vancouver.

https://earth.nullschool.net/#curren...thographic=200.

[franksingleton]

There is the usual confusion between forecast models and presentations. On the basis of objective monitoring, the best global models are those that start with the best data analyses on the finest grids. On that basis, of those models available, ECMWF with a 0.1-degree grid and the latest start time – about 2 ½ hours later than the others – should be the best for longer term planning. In practice, it may not be much better than the US GFS or the DWD ICON (global) with their grids of about 0.125 degree. As far as I know, the Predictwind global proprietary” model uses a 50 km grid. It cannot be as good as the ECMWF or national weather service models. It may be adequate enough but certainly not better.



The best models are not necessarily the most useful. When it comes to short period forecasts, up to, say 2 or 3 days, the GFS or ICON are more useful as they provide output earlier than ECMWF and they are updated 6-hourly while ECMWF is only updated 12-hourly. I would usually prefer to use models run on later data.



If you want detailed forecasts, remember that small weather derails have short lifetimes. claims of forecast being “accurate” or “precise” on a 1-degree grid are meaningless. A feature of size, say 20-30 km, the size of a large thunderstorm, has a lifetime of some 6 hours absolute maximum. Even national weather service models using high resolution satellite and radar data cannot predict a specific storm. The earliest a limited area model output is available is about 3 1/2 hours after data time. By then, small detail will have disappeared or be near the end of its life.



There may not be much practical difference between “official” and “unofficial” limited area models. In my opinion, it will always be safer to use those that can use these high-resolution data. That comes down to the NAM for US coasts, the Caribbean and the central Pacific. For Europe and the Med, DWD ICON-EU (the only forecast that I know that updates 3-hourly), the various HIRLAM or HARMONIE, the Croatian LAM etc. I am not familiar with JMA detailed models but would probably include them in the models to use. I know of no “unofficial” modellers that can use the high-resolution satellite and radar data. As far as I am aware, all start with global model GRIB data, typically on a 25 km scale. Such a grid can only define weather patterns on a scale of 100km at best. That is a geometrical fact. If anyone can correct me about unofficial models, please do so.



As is obvious from the replies that I have seen so far that the main deciding factor about which service you use is ease of use and access. There is nothing wrong with that. But, please do not claim higher accuracy.

[belizesailor]

Originally Posted by franksingleton

There is the usual confusion between forecast models and presentations. On the basis of objective monitoring, the best global models are those that start with the best data analyses on the finest grids. On that basis, of those models available, ECMWF with a 0.1-degree grid and the latest start time – about 2 ½ hours later than the others – should be the best for longer term planning. In practice, it may not be much better than the US GFS or the DWD ICON (global) with their grids of about 0.125 degree. As far as I know, the Predictwind global proprietary” model uses a 50 km grid. It cannot be as good as the ECMWF or national weather service models. It may be adequate enough but certainly not better.



The best models are not necessarily the most useful. When it comes to short period forecasts, up to, say 2 or 3 days, the GFS or ICON are more useful as they provide output earlier than ECMWF and they are updated 6-hourly while ECMWF is only updated 12-hourly. I would usually prefer to use models run on later data.



If you want detailed forecasts, remember that small weather derails have short lifetimes. claims of forecast being “accurate” or “precise” on a 1-degree grid are meaningless. A feature of size, say 20-30 km, the size of a large thunderstorm, has a lifetime of some 6 hours absolute maximum. Even national weather service models using high resolution satellite and radar data cannot predict a specific storm. The earliest a limited area model output is available is about 3 1/2 hours after data time. By then, small detail will have disappeared or be near the end of its life.



There may not be much practical difference between “official” and “unofficial” limited area models. In my opinion, it will always be safer to use those that can use these high-resolution data. That comes down to the NAM for US coasts, the Caribbean and the central Pacific. For Europe and the Med, DWD ICON-EU (the only forecast that I know that updates 3-hourly), the various HIRLAM or HARMONIE, the Croatian LAM etc. I am not familiar with JMA detailed models but would probably include them in the models to use. I know of no “unofficial” modellers that can use the high-resolution satellite and radar data. As far as I am aware, all start with global model GRIB data, typically on a 25 km scale. Such a grid can only define weather patterns on a scale of 100km at best. That is a geometrical fact. If anyone can correct me about unofficial models, please do so.



As is obvious from the replies that I have seen so far that the main deciding factor about which service you use is ease of use and access. There is nothing wrong with that. But, please do not claim higher accuracy.

Yes, it is common for many not to understand the difference between data source and presentation. There are always those who chime in that one presentation service is more accurate than another...without knowing that they both just present the same data source differently.

[Hobiehobie]

Currently we use windy t for daily local, add Passage weather for offshore and Chris Parker or commanders bespoke forecasts for ocean passages. Then supplement with Ocens satellite weather app to download 5 or 7 day gribs out in the ocean. If that fails, SSB forecasts for the area we are in.

If that fails call up a ship on the horizon! If that fails use a wetted finger.

[did-g]

Originally Posted by franksingleton

The best models are not necessarily the most useful. When it comes to short period forecasts, up to, say 2 or 3 days, the GFS or ICON are more useful as they provide output earlier than ECMWF and they are updated 6-hourly while ECMWF is only updated 12-hourly. I would usually prefer to use models run on later data.

If you want detailed forecasts, remember that small weather derails have short lifetimes. claims of forecast being “accurate” or “precise” on a 1-degree grid are meaningless. A feature of size, say 20-30 km, the size of a large thunderstorm, has a lifetime of some 6 hours absolute maximum. Even national weather service models using high resolution satellite and radar data cannot predict a specific storm. The earliest a limited area model output is available is about 3 1/2 hours after data time. By then, small detail will have disappeared or be near the end of its life.

I don't know if they are available online but at least NOAA and Meteo France produce a regional model update every hour with around two hours lag, ie 21H run at 23H


NOAA CONU 15 minutes step 0.025° grid. 24 hours forecast
and MF West Europe 15 minutes step 0.025° and 0.01° grid 23h hours forecast


Subset of HRR:

[pippatheskippa]

This thread is very interesting - I asked recently about what satellite comma to get for ocean passages - Iridium Go with PredictWind seemed to be the consensus - I’m now thinking that PredictWind may not be the way to go....