Accuracy of asymptotic series for the kinematics of high solitary waves

Abstract

The accuracy of several asymptotic series expansions for wave speed and particle velocity under the crest of a solitary wave (on a fluid at rest) up to maximum height is investigated. The very accurate numerical results of Williams (1985) are the measure for our comparisons. The results are based on a scaling of calculated properties of long periodic waves to the case of solitary waves. For wave speeds the classical Boussinesq–Rayleigh expression gives good agreement up to a relative wave height of, say, 0.3. An asymptotic fourth-order expression based on Fenton (1990) can be used up to a relative wave height of 0.7, whereas the corresponding fifth-order expression is slightly less accurate. The Eulerian particle velocity profile under the wave crest is examined using a cnoidal wave expression from Fenton (1990) in the limit of the solitary wave. For low waves a `consistent' (i.e. properly truncated) fifth-order expression and an `inconsistent' ditto both coincide with Williams' results. Beginning at medium high waves, the consistent expression surprisingly exhibits oscillations in the velocity profile, and the oscillations become stronger as the wave gets higher. The inconsistent expression, however, yields the same shape as Williams' profile, but is displaced parallel to this, resulting in slightly larger velocities. For high waves also the inconsistent expression begins to differ in shape from Williams' profile, and asymptotic theory fails. Only for low waves `lowest order theory' gives acceptable results. We show analytically that for the highest wave the particle velocity profile has a horizontal tangent at the water surface; this is corroborated by Williams' numerical results. We also study the particle velocity at the wave crest as a function of wave height. It is shown that the variation has a vertical tangent for the highest wave. Two fifth-order asymptotic series for this velocity, based on the wave speed through the Bernoulli equation, show very good agreement with Williams up to a relative wave height of about 0.6. It is finally shown that it is possible to produce very accurate rational-function approximations to Williams' results for the wave speed as well as for the particle velocity at the wave crest.