To further improve the effectiveness of tunnel resistivity ahead prospecting, a novel tunnel resistivity ahead prospecting inversion method constrained by observation data features was proposed. To address the underutilization of known information and the low accuracy in conventional resistivity inversion, the response characteristics of water-bearing structures were analyzed through numerical simulations. The resistivity of the surrounding rock, the initial position of the anomalous body, and its lateral offset direction were respectively extracted from the tail of the apparent resistivity curve, the zero-crossing point of the apparent conductivity''s first-order derivative, and the response differences among varying measurement points. Based on these extracted features, a data weighting function was introduced and the depth weighting function was modified. The proposed method effectively utilized the available information to enhance inversion accuracy. Finally, comparative analyses with the conventional resistivity inversion method were conducted via numerical simulations. The proposed method was applied to the Xianglushan Tunnel of the Central Yunnan Water Diversion Project, where the actual location was merely about 2 m away from the predicted initial position of the anomaly with highly consistent spatial orientations. The results show that the proposed method significantly improves the positioning accuracy and type identification capability of water-bearing structures in deep-buried long tunnels, providing technical support for the advance assessment of water inrush risks in major water diversion tunnels.