I am investigating methods to reconstruct the strain field within a wall subjected to a concentrated top load. Scattered strain data is available from DFOS (distributed fiber optic sensing) measurements.

The objective is to interpolate the data to create a continuous strain field across the entire wall structure. Universal Kriging was initially used with spherical, exponential, and Gaussian variograms but resulted in significant validation errors. While default parameters for nugget effect and sill were used, their potential impact on improving results is unclear for me.

What is the recommended approach for constructing an accurate normal strain field based on scattered DFOS measurements (1 dimensional tangential strain) and a known loading scenario?