En:using ww3:2013 fall waves school:2013 Fall Projects

De WAVEWATCH-III Wiki

Wave data analysis or model validation

Even if you have not run your own model, you can certainly look at NCEP's or Ifremer's hindcast and altimeter or buoy data, and do some comparisons.

Running your model and using the results

There is no unique choice of model, configuration, parameters ... that will satisfy everybody. Modelling is about making choices, and these require some expertise that takes time to build up. Jump in, try different things and you will go forward. In practice we would all love to have a model that is

  • easy and quick to set up and use
  • runs very fast
  • gives very accurate results

ease of use

Now, in practice there will be trade-offs. If you are reading this page it probably means that you are willing to make some efforts towards a not-so-easy and not-so-quick model set-up (otherwise SWAN may be a better option than WAVEWATCH III).

CPU time

  • choice of numerics (implicit vs explicit for the UNST grids)
  • choice of spatial resolution
  • choice of spectral resolution
  • choice of parameterizations (ST4 is more expensive than ST3, ST2 or ST1)

see the talk about optimisation and profiling to know what is the bottleneck in your case

Accuracy

DO NOT FORGET TO ADJUST THE PARAMETERS TO YOUR WIND FIELD (ECMWF or NCEP or ... )

For wave heights

  • Global ocean: ST4 default, except for east coasts: use TEST570 settings:
& Unified breaking term by Filipot & Ardhuin (JGR 2012), Leckler et al. (OM 2013) : 
  &SIN4 TAUWSHELTER= 0. / 
  &SDS4 SDSC1 = 1.0, SDSC2 = 0.0, SDSBCK =0.185, SDSHCK = 1.5 /
  &DB1  BJALFA = 0. /

Note that in some regions, ST3, ST2 or ST1 can be just as good for Hs (see slides about parameterization impact)

  • Enclosed seas (Med, Channel ...? ). In Previmer we are using this -with ECMWF winds:
 &SIN4 BETAMAX = 1.50, ZALP=0.006, ZWND = 5., 
 Z0MAX = 0.0020, TAUWSHELTER = 0.0 /
 &SDS4 SDSBR = 0.00085, 
 SDSC2 = -2.2E-5, SDSCUM = 0.000, FXFM3= 2.5 /
 &SNL1 NLPROP = 2.7E7 /

With recent increase in ECMWF resolution, this may lead to a bias a bit too high for Hs ... will increase ZWND towards 10 in the future.

  • Shallow coastal areas:

Namelist parameters are defined in ww3_grid.ftn Most of them are documented in the manual and ww3_grid.inp but not all. Look into ww3_grid.ftn for the full story.

For high moments of the spectrum: Tm02, Stokes drift, mss

For seismic noise

Projects

Example of previous work