I've done something similar to this using ZPL. The macro reads in all the data, then generates a line-by-line merit function with all the tabular values as CONS target values. Your Zemax-generated values are next. The macro then generates DIFF lines between your table values and Zemax-generated values. You're using Zemax as a 2D curve fitting routine, and it can do that very well.
Mike’s method works great.
The overall problem is that OpticStudio expects surface sag as input, and you have curvature data. I can see two approaches to using your data:
Outside of OpticStudio, or using ZPL or ZOS-API, integrate your curvature (power) data to get the surface sag. It’s not a simple integration as curvature uses the first and second derivatives of the sag. It’s possible that the makers of your mapping equipment have such a function.
Inside of OpticStudio, use a series of SCUR operands for each pixel location. Enter the measured curvature as the target of each operand, set a weight of 1, and use a Zernike or other freeform-ish surface to set the surface curvatures at each pixel location.
Remember power at any point is (n2-n1)C. For mirror surfaces n2=-n1 so the power is -2C.