Those wishing to model a surface from data in the form of z(x,y) from scattered or semi-scattered data have had few options in matlab – mainly griddata.
Griddata is a valuable tool for interpolation of scattered data. However it fails when there are replicates or when the data has many collinear points. Griddata is also unable to extrapolate beyond the convex hull of the data unless the ‘v4’ option is used, which is slow.
Gridfit solves all of these problems, although it is not an interpolant. It builds a surface over a complete lattice, extrapolating smoothly into the corners. You have control of the amount of smoothing done, as well as interpolation methods, which solver to use, etc.