Self-potential gradient inversion based on a hybrid weighting strategy

Abstract

Self-potential (SP) signals serve as intuitive indicators for identifying subsurface natural current sources. However, this simplicity is offset by the significant challenge of source imaging. Sensitivity-based model weighting approach has limited capability to prevent the skin effect, and measurement errors arising from inappropriate reference positions can lead to distorted source interpretations. To address these challenges, we develop an SP gradient data inversion method based on a hybrid weighting strategy. Three synthetic examples indicate that the hybrid weighting method generally outperforms the individual sensitivity and depth weighting strategies under varying media and source conditions. The underlying mechanisms are analyzed in detail by comparing the model weights, initial models, and integrated sensitivities within the existing focusing inversion framework. Furthermore, we investigate how variations in SP polarity impact source interpretations, reveal the potential distortions they may introduce, and prove that such effects can be effectively mitigated through the SP gradient inversion. The proposed approach is also validated by a laboratory experiment of a naturally polarized CuFe sphere. Both the synthetic and laboratory results demonstrate the superiority of the proposed method in SP source imaging and highlight its potential to deliver valuable insights for reevaluating SP anomalies in environmental and engineering applications.