Abstract: Deep coal seam mining is increasingly threatened by the water damage of underlying limestone. Detecting the water-rich area of coal mining face is a major task to ensure safe production. At present,the mine electrical method is one of the main means of detecting the water-rich area,but there is a problem of inaccurate positioning of water-rich area. In order to solve this problem to improve the detecting accuracy for the water-rich area of the floor rock of coal-mining face,based on the analysis of the advantages and disadvantages of the one-roadway electrical method and two-roadway electrical penetration method,the omni-directional detection method of apparent resistivity was proposed by improving its observation and inversion method. Taking a full-space spherical model as an example,the apparent resis-tivity expression of any pole-dipole system in the outer space of spherical model was given. For the five low-resistivity models at different positions on the inside,edge and outside floor of coal-mining face,the apparent resistivity data were calculated by the omni-directional and two-roadway penetration method respectively. Through the three-dimensional in-version,the rock resistivity images in the floor of coal-mining face were obtained respectively. The comparison of these images shows:① In the horizontal section,the former’s core low-resistance areas are one to one correspondence with the center positions of five models;the latter’s core low-resistance areas are consistent with the center positions of the three models that located in the inside and edge floor of coal-mining face,and have distance error with the center posi-tions of the two models that located in the outside floor of coal-mining face,and the error coefficients are 1. 0 and 2. 0 respectively,which reflects that the farther the model located in the outside of coal-mining face is from the roadway,the larger the error. ② In the vertical section,both of them have various degrees of error in determining the depth of the models,but in contrast,the positioning accuracy for the models from the former is higher than the latter’s,because of its small error. The above analysis results show that the one-roadway sounding data and the two-roadway penetration data can constrain each other to promote the accurate homing of the spherical model with low-resistivity during the om-ni-directional apparent resistivity data inversion. As a consequence,the omni-directional apparent resistivity detection method has greater superiority to the conventional electrical methods,and an engineering practice further demonstrates the accuracy and reliability of the method.