A complex question, the answer to which will depend upon the application.
If your objective is to define the contour of the kerb, then the most cost effective solution would be to survey the kerb directly. That idea is not necessarily very helpful, for reasons given later, but is useful because it highlights the fact that a kerb does not have a frequency response, as such. The "frequency content" of a kerb profile occurs only when a tyre is passes over the kerb at a given horizontal velocity [putting it crudely, dz/dt = (dz/dx)*(dx/dt)].
Having said that, the accuracy of results obtained from a mathematical model of a vehicle driven by a surveyed kerb profile will depend upon the accuracy of the tyre model (amongst other things), and experience suggests that this becomes increasingly problematical as the ratio of profile wavelength to tyre diameter reduces. This is because the forces transmitted to the hubs depend not only on the vertical tyre stiffness and damping properties, but also on the ability of the tyre to "wrap itself around" an obstacle (something Tony Matthews memorably proposed calling the tyre "cuddle factor", I recall).
The above is a preamble to stating that, if vehicle responses are used to reconstruct a kerb profile, the result (however comprehensive and accurate the measurements) will not normally be the actual kerb profile, but will be an amalgam of the profile and the vehicle model used in the reconstruction process. That is again a useful notion to keep in mind, because it can be helpful when trying to understand why model responses might no longer reflect actual responses if/when changes are made to vehicle parameters.
The next general point to be made is that springs and dampers are multi-functional. Greg provided an example of a vehicle (presumably a road vehicle) negotiating a "chuck hole". If he executed a spring, damper and unsprung mass parameter sweep, & his objective was to minimize the loads transmitted to the vehicle, then he would probably conclude that lower rate springs & bars were better, all damping should be in rebound (with no bump damping), & increased unsprung mass would be beneficial. The resulting set-up would probably not survive a subjective assessment because other ride and handling requirements would be adversely affected. In other words, discrete single inputs are just one of several conflicting requirements that must be satisfied when deciding upon an adequate suspension set-up.
Greg made an important point in his last paragraph: Understanding the effect of discrete inputs on vehicle response, and how that response can be manipulated, is probably more important than the ability to reconstruct a specific kerbing event (which would, in any case, be expected to vary from one lap to another). That idea might lead you to the conclusion that an idealized discrete input might be sufficient for your needs. If you agree, then you might decide that a
"haversine" function of appropriate frequency, perhaps "phased" correctly between front & rear axles, would be suitable for your modelling requirements. The frequency of the function could be adjusted to excite, variously, sprung & unsprung mass modes (perhaps with an emphasis on sprung mass modes).