Optimizing geophysical anomaly detection using combined electrode arrays through image processing

Abstract

Abstract The use of a single electrode array in geophysical surveys can lead to increased uncertainty in anomaly detection. To address this, we took advantage of image processing to produce a wide range of data from a limited one and subsequently enhance data manipulation capabilities. Three geophysical anomalies—rectangle block, dyke, and fault—were modelled using four typical electrode arrays (Wenner, Pole-Pole, Dipole-Dipole, and Schlumberger) with RES2DMOD software. The synthetic data were then inverted using RES2DINV software. Python software facilitated the conversion of inverted resistivity models from RGB to real resistivity values, enabling the application of various statistical approaches. We plotted the resistivity sections using the reconstructed data to ensure that our conversion procedure was correct. Next, the resulting models were evaluated using mean resistivity value (MRV), mean absolute error (MAE), and mean absolute percentage error (MAPE) criteria. Our study revealed that combined models outperformed individual arrays in detecting underground anomalies. However, increasing the number of electrode arrays combined does not necessarily give rise to an ideal result. By identifying and eliminating an electrode array that significantly deviated from real resistivity values, new combined models were plotted. In all three models, we can see that three combined electrode arrays provided more accurate results than the four conventional ones if the less relevant array was properly recognized.