La Houille Blanche
Number 3, Avril 1966
|Page(s)||321 - 337|
|Published online||24 March 2010|
Analyse et prévision de l'état de la mer méthode D.S.A. 5
Analysing and forecasting states of sea by the D.S.A. 5 method
Ingénieur Civil des Mines.
2 Ingénieur des Travaux Météorologiques.
The French Meteorological Service's method of forecasting states of sea is described. As so many different parameters are involved (coordinates, wind at various points, etc.) a suitable model for routine computer calculations has had to be developed. The theoretical model is described, and an account is given of the initial analysis and forecasting results obtained. The first part of the article shows how a basic differential equation was chosen. The characteristic heights H 1/k for the state of sea are defined, then the relationship between wave energy, standard deviation and height H 1/k (Eqs. 1, 2 and 3). The D.S.A.5 model is one for which sea and swell are considered from their energy aspect ; wave energy changes with direction and period are studied (hence the idea of "spectro-angular density', i.e. French "densités spectro-angulaires', whence 'D.S.A.') Variation of 'D.S.A.' (ρk) at point k is given by əρk/ər and is a function of the three following terms (Eq. 5): (i) An energy propagation term ƒk (Eqs. 6, 7 and 8, Fig. 6) ; (ii) A term ρk for growth of energy with wind and considered 'D.S.A.' period (Eqs. 9, 10, 11 and 12, Fig. 3) ; (iii) A damping term Ψk (Eq. 14). These adequately represent the three phases of wave evolution. The discrete expression of the basic differential equation (15), is then discussed, considering the new approximations made to adapt the calculation to computer treatment, by choice of appropriate wave period and direction bands (Fig. 14), the right calculation grid (Fig. 5), suitable numerical constants, and by practical organisation of the calculation procedure. The second part of the article deals with the analysis results obtained for the Atlantic Ocean during the first three months of 1965. It is found that errors fall within four categories (Table 1), namely avoidable errors due to the human factor, errors due to the model, accidental errors, and unspecified errors. A few examples of state of sea charts (Fig. 13 to 20) are given in the form in which they will shortly be appearing in the facsimile network. The first forecasting results are then examined and it is found that the reliability of the forecasts is about the same as by forecasts based on surface wind fields for the same period. A few improvements are suggested for the model which has already given most promising results as regards both analysis and forecasting.
© Société Hydrotechnique de France, 1966