How do you choose which forecast model on Predictwind?

[CharlieO.]

Sometimes the different forecast models show quite a different outlook. How do I choose? Are some better or more reliable for a certain geographical areas?

Thanks

[ScottRhodes13]

I go to the "Tables" page where I can look at all models at once. Then I can judge what might be the most likely scenario and what might be the worst case.

[tozz]

Originally Posted by CharlieO.

Sometimes the different forecast models show quite a different outlook. How do I choose? Are some better or more reliable for a certain geographical areas?

Thanks

Comparing to actuals and looking for the consensus guidance seems to work pretty well for me in the PNW. There are different models with different resolution and short/med/long focus. ECMWF is more accurate in my opinion here in the PNW. As you get into very short timeframes the NAM and HRRR are good as well and seem to converge pretty nicely with others making outliers easy to spot and discount... but no more than 3 days out. I found this poking around a while back and it helped my understanding a lot:

https://blog.weather.us/model-mania-...hey-different/

-tozz

[CharlieO.]

Originally Posted by ScottRhodes13

I go to the "Tables" page where I can look at all models at once. Then I can judge what might be the most likely scenario and what might be the worst case.

Thanks Scott,

I get anywhere from 12knots NW to 4knots SE. That can make a big difference where I want to anchor. Never have I pulled up the tables tab before though, it looks to be helpful.

[tiltrider1]

It is my experience when most the models agree there is a high chance for accuracy. The more divergent the models become the more unreliable the forecast is. As to which one is more accurate tends be related as to what you are looking for. Since I use it to look for light wind crossings In the PNW, I find the PWG model to be the most accurate. If I was looking for rain or looking at off shore wind I might find a different model more accurate.

[rslifkin]

Originally Posted by tiltrider1

It is my experience when most the models agree there is a high chance for accuracy. The more divergent the models become the more unreliable the forecast is. As to which one is more accurate tends be related as to what you are looking for. Since I use it to look for light wind crossings In the PNW, I find the PWG model to be the most accurate. If I was looking for rain or looking at off shore wind I might find a different model more accurate.

I also look at how the forecasts change over time as you approach the time you're looking at. If they're changing with every update, they're likely less reliable than if they've been consistent for a couple of days.

[Cigatoo]

Look at your models then read and compare with the “Forecast Discussion” on your NOAA weather app. They are looking at the same models however their interpretation skills are most likely better. Certainly far better than mine.

https://forecast.weather.gov/product...&highlight=off

[BandB]

Originally Posted by CharlieO.

Sometimes the different forecast models show quite a different outlook. How do I choose? Are some better or more reliable for a certain geographical areas?

Thanks

Choose to check them all. That's the benefit of different models. No one knows for sure and you should consider worst case scenarios. Even the historically least accurate may be the most accurate this time. I would never settle for just a single model.

[Bacchus]

Originally Posted by BandB

Choose to check them all. That's the benefit of different models. No one knows for sure and you should consider worst case scenarios. Even the historically least accurate may be the most accurate this time. I would never settle for just a single model.

That is exactly what I recommend to others using Wndy.
Different models have different strengths & weaknesses - besides using different predictions and computer systems they have different grid sizes and update timing. Some are better at local details while others are better at larger area systems.
In many cases the models predict similar system movement but have different timing. Rather than favor a model that provides an acceptable answer I say look at them and be aware of the worst case that can happen as a planning guide.

[mvweebles]

As tiltrider1 states, best guess is when the models converge. But it really depends on what you're doing. Crossing a large body of water, or heading 70-miles up the ICW?

1. PredictWind has their own algorithm that interpolates each model. It's fairly well respected for offshore passage makers.

2. Both the ECMWF and the GFS core models focus on 3-5 day forecasts, not 1-3 day which is of much greater interest to the average recreational boater. As such, while either will generate a short term forecast, they are most useful for longer range passages or departure planning.

3. PredictWind, FastSeas (which uses Windy data) and other prediction and route planning are reasonably accurate - with many caveats. 72-hour forecast remains the limit for reasonably reliable (80% accuracy or so) forecasts.

4. If you plan to make a run of more than 24-hours or so, it really helps to spend a little bit of time digesting the synoptic charts issued by NOAA/NWS. These have human interpretations about predicted movements and intensities of high and low pressures that are invaluable. It gives you about 80% of the data you need to make an reliably informed decision, and to understand why Windy or PredictWind says what they say. Picking up buoy reports fills in a bunch more data. .

https://www.weather.gov/marine/radiofax_charts

Peter

[NCheaven]

Windy, Town Dock (Oriental for NC upper coast), and Weather Channel (land, not marine, but indicative) Albemarle Sound, Pamlico Sound and River, Neuse River. New to me sources note in prior posts - thanks TF members - I'll check them out.