Different definitions for the wave periods are shown
Hi, the issue with the wave period being different between ECMWF and NOAA was relayed to me on Friday (3/3/2017).
I am Jean Bidlot, the main person in charge of wave modelling at ECMWF
I went on the windty website and got the following screen shots
I could not figure out how to plot the wave period as a map but the white text in the middle of the south Atlantic indicates
that ECMWF has a wave period of 9 s and WW3 21 s at the selected point !!!
why is it so different?
It's simple: they are showing two different wave periods.
for ECMWF, they appear to be using the mean wave period of the total sea (more likely MWP in the request below)
but for WW3, they are clearly not! A mean wave period of 23 s is very very unlikely!
We do produce however the peak period and if you compare
for our mean period
for our peak period
you clearly see the difference over the middle of the south Atlantic.
So why such a difference,
mean periods are derived from weight integrals of the wave spectrum (the distribution of wave energy in frequency and direction)
whereas the peak period is the reciprocal of the frequency that correspond to the peak of the spectrum
Over the south atlantic what you see is that the long waves Tim was talking about in the weather discussion, that has affected the Azores and Morocco, have continued their propagation southwards
At that location, they are the dominant wave system, but obviously they are other wave systems and so in the mean the wave period if lower but the peak period is high.
More information on the different wave parameters derived at ECMWF can be found in chapter 10 of
In order to be compared the same definition for the period should be use for both systems.
for surfers, I can see that the peak wave period is more related to the waves that would interesting, but in opinion it is not the ideal quantity to use.
On Monday, I will access the archived data for the 18-19 January to show why.
missing screen shot
@jrbidlot Thank you for your feedback and thoughts. We really miss good meteorologist inside Windytv. We will investigade next week and do something. Unfortunatedlly we have no surfers in company either. I will ask @TomSlavkovsky to get in touch with some surfers that use Windytv do discuss this issue and sort it out.
My previous post shows that ECMWF has different output parameters
that could be used to characterise the swell.
Ideally, the 2d wave spectrum should be used, because it contains the full description of the wave field (in frequency and direction),
but it not very convenient to make maps of it....
The peak wave period only points to the peak in the spectrum. When the spectrum has several peaks, as shown in the example above, it is problematic as it might flip from pointing to one system to the next
Traditionally, this has be addressed by partitioning (splitting) the 2d spectrum into the windsea (the part of the spectrum under the influence of the local wind) and the total swell. The total swell might be composed of several systems with distinct characteristics. Spectral partitioning can be used to split the total swell in different contributions (system). ECMWF employs up to 3 (NOAA many more).
I personally do not like this partitioning as it artificially assumes that the sea state can be divided in clearly distinct systems, rather I have recently introduced a new set of variables which correspond to the wave energy in different frequency bands (expressed in terms of an equivalent wave height). Animate these new parameters and you can clearly see where the long swell is propagating.
All this to say, that users need to be careful when the total sea state is distilled to only a few parameters that are attribute of the total sea state, such as significant wave height (Hs) and peak wave period (Tp) or mean wave period (Tm)
@ivo Tom. Any news on this topic? My data services are telling me that windytv has not yet requested the peak wave period as an extra parameter to add tot he list of wave parameters already requested.
To a minimum the plots between GFS and ECMWF should show the same wave parameter.
TomSlavkovsky last edited by
eonolan last edited by
I am wondering if the new data has been implemented yet? I'm about to set off on a trip down the west coast on an old boat, and it would be good to know how much I can rely on Windy :-)
jswave last edited by
@TomSlavkovsky . Hi. First I'd like to say that windy is great!
Could you confirm that the wave period given by windy for the ECWAM model is the mean wave period Tm (m-1/m0, parameters 232 - MWP, 236 - MPWW, 123 - MWP1, 126 - MWP2, and 129 - MPW3 for total waves, wind waves, swell1, 2, and 3, respectively, see Table 10.1 of http://www.ecmwf.int/sites/default/files/elibrary/2016/17120-part-vii-ecmwf-wave-model.pdf).
Also, is the mean period and Hs of total swell (par. 237 - SHPS, and 239 - MPPS) also available on windy?
And could you indicate whether the peak period Tp of total waves (par 231 - PP1D) is available on windy? (would require a toogle to select either Tm or Tp).
I would also suggest to define somewhere on the website all the parameters given by windy, since the same name could be used to different parameter definitions, such as "wave period". If this is already the case, would you kindly point it to me?
Thanks in advance and keep up the great work!