I am attempting to interpolate the normal strain results in the wall type structure (dimensions:800×500 mm) while having approximately 150 scattered known points in the whole element (find picture for the sensor layout). I attempted using the universal kriging (Pykrige toolkit), but after I compared the interpolated results to FEM results, the lowest error percentage margin I had was 37%, which seems very high. This result is derived while using the Gaussian variogram. I tried creating the custom variogram model but the results seem wrong as the effective range I get is 839.42, sill and nugget effect 0, additionally the error margin is even higher as expected. Am I creating the variogram model incorrectly? Is Kriging useful for this case or should I stick with easier interpolation methods? Python is my first programming language and I never had the connection with Geostatistic so I might have a very simple mistake but any kind of nudge would be appreciated!