AROME 1km vs ECMWF 8km wind maps Irish Sea

[fifer]

So unfortunately my planned launch on Monday has been cancelled due to forecasted wind at the marina. The worst forecast is saying 25kts with gusts considerably higher so fair enough.

I've been poking around the available predictwind forecasts for the weekend and found the comparison shown in the link. I understand that AROME models on a much smaller scale and therefore captures local detail - I'm particularly impressed by the relation of wind to the local topography. But I can't explain the patches of low wind speed in the Irish sea between Holyhead and Dublin. Doesn't seem to follow any fluid dynamic behaviour I can think of. Anyone care to give me an education?

Wind map

 

[KompetentKrew]

 

[KompetentKrew]

I'm not an expert, but there's an occluded front coming through and breaking up.

It's not every day that you see a red dotted line like that in the second two charts.

• https://www.mwis.org.uk/blog/post/528-line-snow
• The 528 line

 

[Roberto]

Wild guess follows
Could it be the two models assimilate sea surface temperature differently?
Such marked boundaries in wind speed only are often due to sharp changes in SST , hence change in air stability -notable example the North Wall effect on the Gulf Stream-
in this case the opposite: cold air from NNW meeting colder water which increases stability and reduces mixing at sea level --> less wind
If the patches move, disappear or are modified during the following hours, it might as well be due sst changes due to tidal flows?
There seem to be sst differences up to a few degC in the Irish sea (hist. map)
MODIS sea surface temperature image of the Irish Sea and St. George's...

 

[franksingleton]

So unfortunately my planned launch on Monday has been cancelled due to forecasted wind at the marina. The worst forecast is saying 25kts with gusts considerably higher so fair enough.

I've been poking around the available predictwind forecasts for the weekend and found the comparison shown in the link. I understand that AROME models on a much smaller scale and therefore captures local detail - I'm particularly impressed by the relation of wind to the local topography. But I can't explain the patches of low wind speed in the Irish sea between Holyhead and Dublin. Doesn't seem to follow any fluid dynamic behaviour I can think of. Anyone care to give me an education?

My only Wind map

First, a nit-pick. As far as I know, AROME has a northern boundary to the south of Anglesey. That is confirmed by the XyGrib output and the Weatheronline charts. I do not know how PW provided the chart that you show,
AROME uses a 1.3km grid, ECMWF uses a 0.1 degree lat/lon grid. That is about 11 km north-south and around 63 km east-west. Obviously, I hope, these define the ability of a model to define topography. Therefore, all things being equal AROME should beat ECMWF for topographical effects.
I cannot say much about the example you give. I do note that the wind direction is roughly ESE. So, some shadowing effect in the lee of the land is to be expected. The AROME pattern does look as though there might be a wave effect like you get with mountain waves. That could be one reason for the lighter winds.

Detailed models are fine for topographic effects as seen here – as long as the overall pattern is well predicted. They can predict some detailed sea breezes. Some years ago, a Met Office scientist told me that their detailed model could predict the sea breeze coming up both arms of the Solent. But, it might not always do so.
Where detailed models fail is in small weather detail such as showers – even with their highly detailed analysis software. Detailed models run by non-official bodies do not start with detailed analyses. They may be OK for topographical effects but not as reliably so as a national weather service. They will be less reliable than a national service when it comes to detailed, non-topographic effects.

 

[fifer]

First, a nit-pick. As far as I know, AROME has a northern boundary to the south of Anglesey. That is confirmed by the XyGrib output and the Weatheronline charts. I do not know how PW provided the chart that you show,
AROME uses a 1.3km grid, ECMWF uses a 0.1 degree lat/lon grid. That is about 11 km north-south and around 63 km east-west. Obviously, I hope, these define the ability of a model to model topography. Therefore, all things being equal AROME should bear ECMWF for topographical effects.
I cannot say much about the example you give. I do note that the wind direction is roughly ESE. So, some shadowing effect in the lee of the land is to be expected. The AROME pattern does look as though there might be a wave effect like you get with mountain waves. That could be one reason for the lighter winds.

Detailed models are fine for topographic effects as seen here – as long as the overall pattern is well predicted. They can predict some detailed sea breezes. Some years ago, a Met Office scientist told me that their detailed model could predict the sea breeze coming up both arms of the Solent. But, it might not always do so.
Where detailed models fail is in small weather detail such as showers – even with their highly detailed analysis software. Detailed models run by non-official bodies do not start with detailed analyses. They may be OK for topographical effects but not as reliably so as a national weather service. They will be less reliable than a national service when it comes to detailed, non-topographic effects.

So Meteologix is consistent with Predictwind, but as you say Weatheronline disagrees. Not sure what to make of that.

 

[fifer]

Wild guess follows
Could it be the two models assimilate sea surface temperature differently?
Such marked boundaries in wind speed only are often due to sharp changes in SST , hence change in air stability -notable example the North Wall effect on the Gulf Stream-
in this case the opposite: cold air from NNW meeting colder water which increases stability and reduces mixing at sea level --> less wind
If the patches move, disappear or are modified during the following hours, it might as well be due sst changes due to tidal flows?
There seem to be sst differences up to a few degC in the Irish sea (hist. map)
MODIS sea surface temperature image of the Irish Sea and St. George's...

The 'boundary line' does progress SE over the subsequent hrs so it could be a tidal flow effect as you say

 

[franksingleton]

So Meteologix is consistent with Predictwind, but as you say Weatheronline disagrees. Not sure what to make of that.

Nor do I. My other source is Meteociel, Meteociel - Modèle Numérique AROME Meteo-France pour la France

 

[DFL1010]

"All models are wrong, but some are useful", as some bloke once said.

Of course, the trick is working out which are useful, when and why.

 

[franksingleton]

Apologies. In #5, I must have pulled up the wrong chart - or had an senior moment.
However, I still think that a topographic explanation is the more likely. I doubt the AROME is as clever as that and, even if it were, it would not know the sea temperatures on that scale.
DFL1010 is correct. No weather prediction is ever precise, as some would have you believe. They are usually helpful.

 

[franksingleton]

YBW is jumping around on my iPad. It just repeated my last post.
What I am trying to post is that AROME runs in two configurations. One using a 1.3 km grid is intended to predict just for France. The Northern boundary is south of Anglesey. They also run AROME PE using a 2.5 km grid for Europe. The N boundary is just about the N of Ireland and the Scottish borders, So, whether AROME 1.3 on Windy and PW is really the 2.5km version or an amalgam of the two, I do not know. It is a reminder that with third party providers you are never really sure what you are getting.

PS. This is AROME PE, Meteociel - Modèle Numérique ensembliste PE-AROME pour la France

 

[franksingleton]

Deleted