- The data you are trying to grid is collinear (that is, data all in a line). Add some additional data points off the existing line of points, or choose a different method.
- If you don't have enough data points, you may not be able to go as high as third order (cubic). Try a quadratic polynomial.
- There is a high discrepancy between the X and Y data values and the Z data values. For example, your X data and Y data is in the range of a hundred thousand or million range (ie. UTM or State Plane values), but your Z data is roughly around 0.01 or some other very small number. This causes the polynomial equation coefficients to be extremely small, which introduces instabilities into the fit equation and illustrates the limitations of computer precision.
If the issue is caused by #3 above, you can work around the problem by subtracting a base value from the X and Y values and then creating the polynomial grid. For example, create new data columns in your data file. If your X and Y data are on the order of 500,000 and 6,000,000 respectively, subtract 500,000 from the X data and 6,000,000 from the Y (to reduce the size of the XY values) and put the new data into these columns. Then grid the data using these XY columns with Polynomial Regression. After the grid is created, add the subtracted values back into the XY positions using the Grids | Calculate | Transform command with an the Offset operation.
Updated September 12, 2017