<p>During coal mining operations, water-rich aquifers within the height of developed water-conducting fracture zones pose safety risks. Identifying the extent of these water-rich zones in advance provides a scientific basis for implementing water control measures. The transient electromagnetic (TEM) method is commonly used to detect these water-rich zones. However, due to the effect of turn-off time, the inversion accuracy of TEM data is reduced, hindering precise delineation of boundaries of these zones. To more accurately reflect actual water storage properties of aquifers and achieve high-precision inversion results, this study investigated laterally constrained inversion (LCI) processes of the time-domain TEM data. Specifically, the ramp step within the TEM turn-off time was divided into a series of vertical pulse responses, and the response formula of TEM-induced electromotive force was derived and analyzed. By establishing a terrestrial electricity model, forward modeling allowed us to analyze the characteristics of induced electromotive force response curves for different TEM turn-off times and the effect of turn-off time on early resistivity. Based on pseudo-two-dimensional (2D) inversion of TEM data, full-time domain LCI of TEM data was carried out. The study showed that the longer the turn-off time, the greater the effect on early TEM signals. When the turn-off time was fixed, considering it in TEM inversion yielded results that more closely reflected actual geological conditions. The results of conventional LCI inversion, full-time domain LCI inversion, and Interpex inversion of TEM data were compared using practical engineering data; the results of full-time domain LCI inversion more accurately reflected the actual geological conditions. This work enhances the accuracy of detecting water-rich zones in coal mines, improving safety and water control measures.</p>