A set of time histories of sea wave elevation can be found in the WAFO MATLAB software package, which you can easily download for free (along with this statistical suite).
Referring to Seyed Mehdi Enderami's comments, that should be the most simplistic and practical way to get a time series of an irregular wave. Further, you can add random phases also. For real data for any particular sea, you need to get hold of some measured data from some reliable sources.
A 'real' wave is much more than surface elevation at one point (wave length, wave direction; orbital motions in the horizontal plane).
If one arbitrary 20 minutes 1 s sample of surface elevation at one point is sufficient for the purpose, I would prefer a synthetic sample from a stationary long- crested seastate, represented by a wave spectrum (surface elevation variance distribution over frequency).
If nonlinearity in the waves is the issue, then I think faster sampling, longer duration, and direction spreading measurement, in addition to the surface elevation, is required.