Predict Wind


Patrick McAneny
 

I bought Predict Wind a few months ago and  a Iridium Go to download offshore. I found it to under report winds with TS Gordon by a large margin,like 50%. Now with the Hurricane Florence and sustained winds of Cat4 140 mph reported for the last two days ,Predict Wind is showing from 68mph on PWG to 102 mph on GFS. Since this is a well reported storm I don't understand the under reported wind speed . Is PW generally this far off, can it be relied on? I asked PW about this and basically they said , that their models are"not geared to predicting hurricanes." I do not understand this explanation , a hurricane is simply a low pressure system, their models are not geared to predict low pressure systems ? I have been looking at their models and comparing them with known conditions. So far it is not reassuring , has it been found to be reliable ?

Thanks,

Pat

SM#123


greatketch@...
 

Pat,

PredictWind has four different models that it presents.  The first two are the GFS and ECMWF which are the two major worldwide weather models.  Nothing about these models is special to PredictWind, these are exactly the same two models that virtually EVERY weather forecast agency all over the world uses. ECMWF tends not to show in many "free" weather programs because you have to buy the results from the European Center for Mid-range Weather Forecasting.  The output from the GFS model is free to all from NOAA.

The other two models are the PWG and PWE are the PredictWind's "enhanced" versions of the output from the GFS and ECMWF models. I have not found them to be significantly more helpful than the actual model outputs.  I do not understand exactly what PW does to create this output from the major models.

In comparing model outputs to the "real world" data, there are many things that can make them look different.  Starting with the fact that the most violent parts of a very large hurricane are actually a rather small weather feature on a global scale. 

PredictWind averages conditions over significant areas, and this can significantly impact the numbers.  In the highest resolution data available from PredictWind it still averages data over 50km.  That is a very large area when you are talking about a hurricane, where the highest winds are in a very narrow band around the eye.  To take this to an extreme example, PredictWind can not tell you about the wind speed in a tornado, it is too far small.  It's the same with a hurricane, although it obviously does better.

The GFS model itself only works internally to a resolution of 28km.  ECMWF model works at a higher resolution internally (9 or 10km, I think). So the models actually calculate a LOT more data than PredictWind presents, but they are still an approximation--at best--of what a real hurricane looks like.

Another issue is that it is not always clear about what wind number we are actually talking about.  Is the the 1 minute average?  the 3 minute average?  the surface wind?  The wind at 10 meter elevation?  All these numbers can be very different.  I don't know which of these numbers PredictWind presents.  I don't really care, for the way I use the program.  Many times the media presents the biggest number they can find because they sell more advertising that way.

I always caution people when they look at GRIB files created by computer programs. They look so PRECISE, it is so tempting to trust them.  They have none of the caveats that and weasel words that come from weather forecasters.  But these are NOT weather forecasts, and they are NOT reporting current conditions... they are the inputs that are used to create weather forecasts.  

Bill Kinney
SM160, Harmonie
Annapolis, MD, USA

---In amelyachtowners@..., <sailw32@...> wrote :

I bought Predict Wind a few months ago and  a Iridium Go to download offshore. I found it to under report winds with TS Gordon by a large margin,like 50%. Now with the Hurricane Florence and sustained winds of Cat4 140 mph reported for the last two days ,Predict Wind is showing from 68mph on PWG to 102 mph on GFS. Since this is a well reported storm I don't understand the under reported wind speed . Is PW generally this far off, can it be relied on? I asked PW about this and basically they said , that their models are"not geared to predicting hurricanes." I do not understand this explanation , a hurricane is simply a low pressure system, their models are not geared to predict low pressure systems ? I have been looking at their models and comparing them with known conditions. So far it is not reassuring , has it been found to be reliable ?

Thanks,

Pat

SM#123


Alan Leslie
 

As Bill says, you have to remember that this data GFS PWG ECMWF PWE is all theoretical. Models are just that they are not forecasts....they are algorithm generated numerical models
A good forecaster looks at the models but makes assessments based on many other factors to provide a forecast.
GRIB files (which is what these are) can give you an "idea" of what might be happening in the near future.
In my experience even when they all agree, this data is only valid for about 3-4 days.
In the Pacific the models are notoriously inaccurate especially around land forms, isolated islands etc. 
Predict Wind claim that their enhanced models take this into account, but in my experience other than when there is a steady trade wind flow, they're not that accurate.
...The wind speeds are frequently underestimated by PWG and overestimated by PWE in my locale. About the only thing you can almost depend on is the direction...but close to landforms that not always true either....We have seen PWE and PWG give completely different wind directions at the same time.
Even the top forecasters don;t always get it right.
BUT in terms of cyclones / hurricanes, as Bill says, the models just can't deal with those highly concentrated weather events.
One needs to look at all the available information, models, weather charts, forecasts ( from forecasters...not buoy weather etc) and try to decide what is most likely and make plans on that.
One of things that happens with all this scrutiny here in the S Pacific when heading S ahead of the cyclone season is what they call "Analysis Paralysis"...so much information and conflicting data that no-one does anything !
We look for what's mostly likely suitable weather and then go....you get what you get and you should be able to deal with it....hurricanes aside.....

Cheers
Alan
Elyse SM437
Noumea