If the cells of the field are distributed uniformly, and if the height of each cell is weak function of 2D position, then the answer seems simple:
- Rearange the numeration of the cells to MxN=100x10 (use IND2SUB and SUB2IND)
- Measure the Lon&Lat of the bounding cells: (1,1),(100,1),(1,10),(100,10)
- Make regular interp2 over Lon values and Lat values separately like this:
- lon2=interp2(x1,y1,lon1,x2,y2);
- lat2=interp2(x1,y1,lat1,x2,y2);
where
x1=[1,1,10,10] (pay attention that x-position is parallel to j-index!)
y1=[1,100,1,100]
x2=1:10;
y2=1:100;
After the above works, it is also possible to make the results more precise by taking additional measurements at some randomly distributed cells WITHIN the field and executing scattered 2D-interpolation with GRIDDATA at the same manner as with INTERP2.
You can use 'linear' or 'spline' options in both cases, as the extrapolation is not required.