Regular data sampling makes the creation of geometry much easier for surface representations, however many real-world data sets are irregularly sampled. As mentioned previously, an irregular sampling array becomes a tessellation problem.
There may be a strong temptation to resample the data into a regular grid. This approach can be problematic in many cases, since the data may not change linearly between grid points. False relationships can be created in the resampled data.
Regridding can also lead to a performance degradation, since you will probably end up with a larger number of data points.