| Home > Research > Ifsdocs > WAVES > |
Chapter 2. The kinematic part of the energy balance equation
Chapter 3. Parametrization of source terms and the energy balance in a growing wind sea
Chapter 4. An optimal interpolation scheme for assimilating altimeter data into the WAM model
Chapter 5 Numerical scheme
Chapter 6 The WAM-model software
Chapter 7 Wind-wave interaction at ECMWF
REFERENCES
Previous Section
6.6 ECMWF Post-Processing
Although, the ECMWF version of WAM is basically following the same structure as the original version, there are also important differences to be noted. In particular, our version takes full advantage of grib coding and decoding both for the integrated parameters and the two dimensional spectrum. The advantages of grib coding are that the fields are archived in a platform independent form and that the size of the fields reduces by a considerable factor. For example the size of an integrated parameter field reduces by a factor of 3, while the size of a spectral fields reduces by a factor of 9. The large reduction in the size of spectral fields is accomplished by archiving the logarithm of the spectrum, thereby reducing the range of the values considerably. Furthermore, rather than archiving one spectrum per grid point, which would result in spectral fields of a large size, ECMWF archives a particular frequency-direction bin as one global field. Thus, the global spectral field is splitted up in
fields, where
is the number of frequencies and 
is the number
of directions of the spectrum.Because IO is relatively slow it is advantageous to minimise the amount of IO. This is accomplished at initial time by transferring grib coded information from disk to one of the PE's and by transfering the initial data to one of the other PE's where it is decoded. Next the decoded data is is distributed over all the other PE's. Since spectral data have been splitted up, the reading of the initial conditions may be performed in a balanced manner. To that end the spectral file is read on PE 1, who distributes the fields per frequency and direction to all other PE's where it is decoded. Writing output is accomplished in a balanced manner by collecting on the first PE the data for the first field from all other PE's, by coding it and by transfering it to disk, while at the same time the second PE is doing the same task for the second field etc.
Finally, the information written to disk is temporarily stored in a sophisticated Fields Data Base (FDB), where it is picked up by archiving tasks that store the information in the MARS archive. The full list of products that is being archived is given in Table 6.1. Post-processing may now be accomplished in various manners. One way is by running programs that read and plot analysed and forecast wave parameters. A more popular method nowadays is to do post-processing in interactive mode using METVIEW.
Next Section
Previous Section
 |
23.04.2002 |
 |
© ECMWF |
