Outlier Detection in Subsurface Modeling of 2D Electrical Resistivity Imaging by Using Boxplot

The 2D electrical resistivity method has huge applications in environmental, engineering, and shallow subsurface investigations. This electrical resistivity imaging (ERI) survey obtains the subsurface distribution by injecting current into the ground using two current electrodes (C1 and C2) while another two-potential electrode (P1 and P2) is injected to measure the resulting voltage difference. The resistivity value is calculated from the current and voltage differences obtained from this survey. For 2D electrical resistivity imaging, a large set of data is required to effectively map the complex resistivity distribution of the subsurface structure. However, due to the nature of the measurements, noise is detected present in this 2D ERI survey. This noise may affect the quality of the data obtained and will contribute to the quality of the model. A good quality model must have good quality acquired data and must have minimum impact from the presence of the noise. This work aims to conduct noise detection mechanism by using statistical tool, called boxplot. Noise detected by the boxplot was removed. 2D electrical resistivity imaging (ERI) survey was replicated by using Geotomo software to generate synthetic data that is used in developing the forward and inverse models. The developed models were analyzed by comparing their respective Root Mean Square (RMS) values before and after the removal of the noise. The subsurface model after noise removal has shown higher RMS value if compared to the model without noise detection as the outlier is replaced. This indicates that the proposed noise detection mechanism has managed to improve the current practice of manually removing the outliers. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.