The electrical resistivity of the soil is not a fixed value, instead, it varies along its depth and lateral extent. Soil resistivity measurements are typically carried out using the Wenner method by placing four-pin electrodes inside the soil and passing a current through it. Measurements are repeated with different electrode spacings. These electrical resistivity values at different spacings are used to develop the soil model which is then used for earthing calculations.

SafeGrid Earthing Software Version 7.0 was used for the modelling.

Example calculation

Consider the measured electrical resistivity values for a site, at an increasing spacing of 1m to 50m, using the Wenner method, as follows. 

The graph of electrical resistivity versus electrode spacing is as shown below.

From the above graph and table, it is observed that the value of electrical resistivity 466.62 Ω-m at 25mm spacing (R7) is very high compared to the rest of the data. It indicates that this value is an outlier, probably a result of an error in measurement or recording.

This can be further verified by considering the soil modelling results in the SafeGrid Soil Modelling calculator, first with all the datasets considered, and second, by omitting the outlier R7 from the dataset.

Case 1: All data points are considered (including the probable outlier R7).

The soil model was calculated for four layers.   It is observed that most of the data points lie outside the fitted curve and Root Mean Square Error (RMSE) of the model is 81.88% which is very high.

Case 2: Outlier R7 is omitted from the calculation.

The soil model was calculated again for four layers, but now electrical resistivity at 25 mm spacing (R7) is neglected. The Root Mean Square Error (RMSE) of the model was reduced to 16.97% and the resulting curve is a better fit along the range of data points.

Conclusion

Outliers are introduced in soil resistivity measurements because of electrode spacing errors and electrode contact resistance during measurement. The soil modelling calculations are sensitive to outliers in the data, causing the soil model to be incorrect.  In most cases, outliers are easy to spot. However, checking the calculated RMSE of the soil model, both with and without outliers, could help the user make an informed decision about whether to include or omit certain measurements.