- 1 waves & WWATCH wiki
- 2 Working with wave model results
- 3 Work in progress
waves & WWATCH wiki
The WAVEWATCH III - hereinafter WWATCH - modelling framework was originally developped by Hendrik Tolman. It is now developed and maintained by the WAVEWATCH III development group... which is growing every day. WAVEWATCH III (R) is a registered trade mark of the National Weather Service (NOAA/NWS). Contributions from LOPS particularly include parameterization of physical processes (swell dissipation, bottom friction, wave breaking, wave-ice interactions ...) and A. Roland has contributed most of the numerical schemes on triangle-based meshes. We are using WWATCH for routine wave forecasting in the MARC project (Modelisation et Analyse pour la Recherche Cotiere), and hindcasting.
The goal of this wiki is to facilitate the exchange of information between the different users and developpers of the code, associated tools - and other models -, but also with the users of the model results.
The phase-averaged spectral wave models represent the local sea state - at any point at the surface of the ocean and any point in time - by a discretized spectrum - say N - which has 2 dimensions: N(k,θ). Here k and θ are the wavenumber (k=2π/L where L is the wavelength) and θ is the direction of propagation (beware of wave direction conventions). The spectrum gives the distribution of the energy or some closely related quantities across the different components. Because the variance of any wave-induced disturbance can be related to the variance of the sea surface elevation, the spectrum can thus be used to compute many statistical parameters (bottom agitation, surface slopes …).
The tasks performed by the model are:
- the propagation of the wave energy as a function of the wavenumber k, direction θ and medium properties (water depth D, current profile …)
- the integration of the sources and sinks of energy for each spectral component.
Like any model, it uses many conventions, settings and parameters. Some of these are not obvious. Follow this link for some more details ...
Meetings, Courses, tutorials, training material and forums
All the main pages of the workshops are available here :
- workshop "from surf beats to the Earth's hum": March 12-13 2015
- meeting "ww3 coupling with ocean and atmosphere models" : June 28 2016
All the main pages of the training courses since 2013 are available here :
- Waves short course in Brest, France: Fall 2013
- Waves short course in Brest, France: Spring 2014
- Waves short course in Brest, France: Fall 2014
- Waves short course in Brest, France: Spring 2015
- Waves short course in Brest, France: Spring 2016
- Waves short course in Cape Town, South Africa: Austral summer 2017
- Waves short course in Brest, France: Summer 2017
- Waves short course in Brest, France: Summer 2018
- Waves short course in Brest, France: Summer 2019
If you want to retrieve all the material about the last training course, you can get it by :
- wget -mnH --cut-dirs=3 ftp://ftp.ifremer.fr/ifremer/ww3/COURS/WAVES_SHORT_COURSE
- wget -mnH --cut-dirs=3 ftp://ftp.ifremer.fr/ifremer/ww3/COURS/WAVE_DATA
If you need some help on ww3, fell free to use :
Working with wave model results
Some specific file formats are used for various purposes (data storage, grid definition). here you will find a list and description of these formats . These include directional wave spectra, integrated parameters, and all sorts of maps of different wave parameters (wave heights, wave periods, seismic noise sources ... ).
Wave hindcast configurations
The IFREMER wave hindcasts ran for the IOWAGA project cover the time frame from 1990 to 6 days into the future. These are available on regular grids with varying resolutions (30' to 2') and unstructured grids that have resolutions down to 100 m at the shoreline …
Work in progress here
Wave forecast configurations
The new MARC, "Modélisation et Analyse pour la Recherche Côtière", is running on the Ifremer supercomputer DATARMOR. It goes with a new version of WAVEWATCHIII revision 150X and a multigrid implementation more some unstructured meshes on specific regions. Find more details on the WW3 configuration on DATARMOR.
The previous wave models in Previmer combine WWATCH and other codes in order to cover a wide range of scales from the global ocean to the beach. The results can be seen on www.previmer.org/previsions/vagues . The model results are available via OpenDAP (NetCDF files: please make a request by mail firstname.lastname@example.org : this is the proper way to make sure that the data flows are monitored and that you are warned about system changes. Otherwise all parameters are dumped on the Ifremer ftp updated twice a day at 3 AM and 3PM UTC, and with a clean monthly update around day 10 of each month.
Validation data, tools, and results
Altimeter , SAR and buoy data are routinely used to monitor the quality of the model output.
Data from dedicated field campaigns are also used and we are also playign with Sentinel 2 imagery.
For validation and analysis purposes, various parameters are computed from modelled spectra.
The full set of tools is described here .
The tool box is available by checking out the following svn repos : https://forge.ifremer.fr/svn/ww3tools/trunk/
The documentation for the "automatic" hindcast toolbox is available here
Other useful tools are the NRL testing environment for standard test cases.
The quality of a wave model may be described by a wide range of metrics.
Code evolution and development
Current work at Ifremer and SHOM is based on NCEP's version and associated tools.
There are two projects under version control in:
These tools contains many preprocessing / postprocessing tools … more here
Developments should generally follow the "best practices guide" which are included in the "guide" directory. Developers should be careful that some not-well-optimized algorithms can increase significantly the CPU time. Follow the link for some help on code profiling .
The Ifremer svn server includes a stable development version (now up to revision 8xx) that is very close to the trunk of the NOAA/NWS/NCEP/EMC server. The main differences between the “officially released” version 3.14 of WWATCH and this version (now realeased as a "alpha" test version by NCEP) are:
- WWATCH on triangle-based grids : interface with and inclusion of Aron Roland's (TUD) propagation schemes on triangle-based meshes
- an updated wave breaking term as part of the ST4 switch option (which is now tested to also be able to deal with depth-induced breaking: this is Jean François Filipot's thesis work)
- extra output parameters (surface mean square slope, energy fluxes …)
- direct output to NetCDF files with a ww3_ounf program that can be used instead of ww3_outf and ww3_ounp program that can be used instead of ww3_outp
- possibility to use an iceberg mask on top of the usual ice mask
- coastal reflection
- movable bed bottom friction
- infragravity wave modelling
Some of this is still under development. See the to do list .
Practical information for using wave model output or other wave data
Work in progress
implicit integration of WAE evaluation of the newly developed integration method by Roland and Dutour.
Periodic IOWAGA group meetings
Discussions on the WWATCH model evolution and applications … details here
Configurations, chaînes de calcul, et résultats de Previmer
Les modèles de vagues dans Previmer combinent WWATCH et d'autres codes de calcul afin de couvrir une vaste gamme d'échelles, de l'océan mondial à la plage. Les résultats de ces calculs sont visibles sur le site web Prévimer ainsi que les pages de l'IRD à Nouméa pour zones du Pacifique sud. Les résultats de rejeux dans le cadre du projet IOWAGA sont décrits ici et accessibles sur ce site ftp
Validation du modèle et données associées
Les états de mer simulés sont par ailleurs observés par satellite, observations in situ. et sismographes. en savoir plus …
Couplage avec des modèles de circulation océanique 3D
Le couplage de WWATCH avec des modèles de circulation 3D utilise le coupleur PALM et quelques modifications du code WWATCH, afin de produire - au bon moment - les paramètres nécessaires, et de faire les communications vers le coupleur [VACCI3D ].