Abstract: Objective The frequency-domain induced polarization (FDIP) method has emerged as a pivotal direction in the advancement of water detection technology for coal mines. This is attributed to its capability to measure critical electrical parameters, such as chargeability and complex resistivity of geological bodies, and to mitigate the ambiguity in interpreting electrical anomalies. However, the current implementation of FDIP primarily focuses on data along the tunnel axis, constraining its capacity to capture geoelectric information beyond the tunnel's immediate path. This limitation leads to practical challenges, including the unclear identification of the orientation of water-bearing anomaly bodies. Methods To address these challenges, we propose a three-directional FDIP parameter observation method tailored to the specific conditions of mine tunnels. We modeled an infinite plate-shaped conductor in a full-space to represent a water-bearing fault ahead of the tunnel excavation and derived the response expressions for the threedirectional FDIP parameters. Subsequently, we analyzed the variation characteristics of these parameters with respect to model azimuth, dip angle, and distance from the source using numerical calculations and theoretical analysis. Results Our findings reveal that: (1) the apparent chargeability and apparent complex resistivity curves along the tunnel axis exhibit K-type (low-high-low) and H-type (high-low-high) patterns, respectively, and are largely invariant to model parameter changes, with electrical anomalies consistently displaying 'low resistivity and high chargeability' characteristics; (2) the apparent chargeability curve perpendicular to the tunnel sides is K-type when the model is directly ahead of the tunnel, and inverse function type otherwise, while the apparent complex resistivity curve is Ktype when the model is offset to the left and H-type when offset to the right; (3) for the direction perpendicular to the tunnel ceiling and floor, the apparent chargeability curve is K-type when the model is upright, and inverse function type otherwise, with the apparent complex resistivity curve being K-type when the model is inclined forward and H-type when inclined backward; (4) the anomaly amplitude and detection ranges of the three-directional FDIP parameters are significantly influenced by model parameters, particularly the detection range at extreme or step points, which varies markedly with the model's distance from the field source. Conclusions The three-directional FDIP parameters demonstrate a pronounced anisotropic response to a plate-shaped water-bearing body model. The sensitivity of FDIP parameters along the tunnel axis to the model's attitude is relatively low, contributing to the current challenges in accurately identifying the orientation of electrical anomalies during exploration. In contrast, FDIP parameters perpendicular to the tunnel sides are sensitive to the model's azimuth, and those perpendicular to the tunnel's roof and floor are sensitive to the model's dip. This three-directional observation method provides richer electrical information for detecting water bodies ahead of tunnel excavation compared to existing methods, thereby enhancing the spatial positioning accuracy of water bodies.