Abstract: The mining conditions of deep coal mines are becoming increasingly complex, and the issue of mine water disaster poses a serious threat to the safety of coal mine production. Mine direct current electric method has become one of the main means of water disaster detection in the floor of working face because of its high sensitivity and strong anti-interference ability in detecting low resistance. In order to enhance the imaging accuracy of water damage inversion in the working face floor, through the analysis of the one-roadway electrical method, the two-roadway electrical method, and the comprehensive resistivity detection method that integrates both one-roadway and two-roadway method, a novel cross-roadway electrical observation method has been proposed. On the basis of the all-round exploration method of mine resistivity, this study proposes and incorporates the cross-roadway electrical observation mode, forming a set of apparent resistivity observation data including one-roadway electrical method, the two-roadway electrical method, and the cross-roadway electrical method, and enhancing the extraction of the full potential matrix observation information. To verify the practicability of this method, numerical simulations were carried out by designing single and multiple spherical abnormal body models in the full space, and the inversion results were analyzed. The research indicates that: the inversion results of the cross-roadway observation can clearly display the vertical form of the low-resistance abnormal body. With the increase in depth, its inversion results have higher resolution. In the lateral distribution range, the false abnormal phenomena are significantly reduced. The inversion results exhibits superior performance in identifying the boundaries of low-resistivity anomalous bodies, accurately reflecting their shapes and positions, with particularly pronounced inversion results during deep detection; with the increase in the quantity of observation data, the inversion effect has been significantly improved, especially in the identification of low-resistance abnormal areas, it shows higher sensitivity. The engineering case further demonstrates that the accuracy and reliability of this method in detecting water disaster in coal mine floors, forming a new detection approach suitable for precise identification of water channels and areas with high water content in coal mine floors, this provides a foundation for preventing and controlling water disaster in coal mines.