Rotate the contour for the Gaussian integral through angle ; a small positive damping factor justifies the deformation and can then be removed. After scaling by , the oscillatory Gaussian integral with negative phase gives
Similarly, first scale and then rotate the positive ray through angle . Since , the cubic oscillatory Gamma integral gives
For fixed , consider the complex phase
Its unique interior stationary point is
where
The endpoint contributions are , while the one-dimensional stationary-phase formula has order . Taking its real part proves the stationary-phase asymptotic for the generalized Bessel cosine integral:
For the phase is . Its stationary point has merged with the endpoint , where
Use the cubic endpoint stationary-phase scaling . The other endpoint and the region away from zero contribute only , so
Therefore
Put with fixed . The relevant stationary point will have , so the assumption allows the long-wave approximation
Using , the phase has no stationary point for . The other phase is
with
and
The stationary part of the integral is therefore
By the one-dimensional stationary-phase formula, this is asymptotic to
Since
we obtain
Using , this has the requested form
where
The Fourier modes evolve according to the dispersion relation found in part (a), so
For , the stationary-point equation is
It has the two real solutions
Define
The one-dimensional stationary-phase formula gives one contribution from each stationary point:
Because the initial field is real, its Fourier transform has the conjugate symmetry . Hence the same result can be written
Along , if , the stationary wavenumber obeys . The one-dimensional stationary-phase formula then gives
Only the integrated initial velocity survives in this long-wave far field.