Soil Modelling

Overview

The Soil Modelling module uses a powerful optimisation algorithm that can interpret field measurement data and deduce the soil structure which best matches the measurements.

How to use

You should enter your field measurement data in order of ascending electrode spacing. If you have multiple measurement datasets (traverses) it is better to write the data to Excel and import it. Enter the desired number of layers, start from 3 layers, and calculate and review the RMSE error. Exclude measurement points that appear to be outliers and recalculate the model to see if a better (lower) RMSE can be achieved.
Increase the number of layers and recalculate the model.

The goal should be to achieve both a low RMSE and a minimal number of soil layers.

Measurements plot

This plot displays the soil resistivity field measurements (crosses), the soil resistivity model (black continuous line), and the model as a step plot (blue stepped line).

Soil model results

The results from the calculation of the soil model are displayed in this field. Model Parameters – Returns the soil resistivity model parameters; layer resistivity values and associated thicknesses.

An important measure of the goodness of fit of the model with the data is the Error (RMSE) of the model.

Generally an error below 15 % means the model fits well with the data. If the RMSE is unacceptably high then the number of layers can be increased.

Field measurements table

The field measurement data can be entered into the table by pressing the Add Row button and clicking on the fields in the table which contain <Click>. Field measurements can be removed by highlighting the row of the table and pressing Remove. All measurements in the table are removed by pressing the Reset button. Data can be imported from an Excel file by pressing the Import From File button. All measurements in the table are saved to the csv file by pressing the Export Data button.

Multiple datasets (traverses) can be analysed by pressing the Add Traverse button. Each traverse can be given a unique name. The user can switch between datasets using the left and right arrows.

Multiple traverses can be grouped using the dropdown, which combines the measurements to obtain a single soil model.

For Wenner measurements, the Spacing is the distance between any pair of adjacent pins.

Import Data - Data which is saved in Excel should be in the format contained in the templates.

Calculate - Pressing this button will automatically determine the optimal soil model to match the field measurement data. The soil model which is calculated will have the number of layers as determined by the automatic setting or can be entered by the user. If the number of soil layers specified by the user cannot be found then a model with a lower number of layers will be returned.

When pressing Calculate again for the same data set that the model returned by the algorithm will always be slightly different from the previous and this is because in general there are multiple global solutions to finding an optimal model and that each time a calculation is performed the first model which satisfies the criteria will be selected and displayed.

Model settings

The soil model can be determined automatically or the desired number of horizontal soil layers can be specified by the user. The user can select any number of layers however the algorithm will only fit a model up to a maximum number of layers which it determines based on the data. For example, if you enter Number of layers as equal to 5 and the algorithm determines the maximum number of layers is 3 then this is the model that is returned.

It is typical to use a soil model consisting of between 3 and up to 7 layers.

The standard modelling equations are valid for electrode spacings much larger than electrode length (or burial depth, if spherical sources are used). When this is not the case however, like when the spacing between the measurement electrodes is small, the driven depth of the electrodes should be considered. Selecting Probe depth correction to be True may improve the Error (RMSE) of the model.