28
R. Grimshaw and C. Yuan
the formation of a solitary wave train ahead of the undular bore. This solitary wave
train can be described by the reduced set of modulation equations (18, 19, 21), and
there is a region where the rear of the solitary wave train interacts with the undular
bore, forming a two-phase wave interaction. The rear part of the undular bore retains
its shape, where pertinently we note that in the similarity solution for the constant
coefficient case, 𝛼 occurs only in the variable x∕𝛼U 0 𝜏, implying that the effect of a
variable 𝛼 is in some sense equivalent to adjusting the time scale. This leading solitary wave train can be described by a similarity solution of the modulation equations
(18, 19, 21),
d = 0 , a =
3𝛼 1∕3 X
𝜒
, X m (𝜏) < X < X M (𝜏) , 𝜒 =
𝜏
∫
0
𝛼
4∕3
(𝜏
′
) d𝜏
′
.
(22)
At the head X = X M (𝜏) of the wave train, the amplitude is a M = 3𝛼 1∕3 X M ∕𝜒, while
at the rear X = X m (𝜏) > 0 of the wave train, the amplitude is a m = 3𝛼
1∕3 X m ∕𝜒. The
speed of each solitary wave is V = 𝛼a∕3 (17) which yields the asymptotic position
of each wave as X = C𝜒 where C is a “constant” which depends on the initial position of each solitary wave. This must be matched to the following undular bore to
determine X M,m (𝜏), and the details of that are described by [5]. But note that if it
is assumed that the leading wave in the following undular bore has amplitude 2U 0
then we may expect that a m ≈ 2U 0 and a M ≈ 2U 0 𝛼
1∕3 , assuming that initially 𝛼 = 1.
The dynamics of an undular bore as it passes through a critical point was examined
recently by [16]. Before the critical point, the behaviour of the leading solitary wave
train can again be described by the modulation theory as above, where now 𝛼 > 0
decreases towards zero. The behaviour after the critical point is described below.
Numerical Simulations of a Process Model
Numerical simulation of an internal undular bore propagating up a slope were
reported by [16] using the transformed vKdv equation (13). Here we review and
supplement those results. For the numerical simulation we use a pseudo-spectral
method where the nonlinear term is evaluated in physical space.
We consider a process model to detect interesting dynamics, and choose the coefficient 𝛼 in the transformed equation (13) to model internal waves propagating up a
slope, in two cases, one when the nonlinear coefficient 𝛼 increases, and one when it
decreases with change of sign, causing a polarity change. Note there is no loss of generality in choosing the initial value of 𝛼 > 0, as otherwise one one can make the transformation U → −U. Thus we set 𝛼 = 𝛼(𝜏) varying monotonically from 𝛼 = 1, 𝜏 = 0
to some constant value 𝛼 = 𝛼 a , 𝜏 ≥ 𝜏 a . Then there are two cases, either 𝛼 a > 1, or
𝛼 a < 0 (a change of polarity). Specifically the coefficient 𝛼 is given by
Précédent

- 34/610

Suivant