Abstract: In geological surveys, the unmanned aerial vehicle (UAV)-borne time-domain electromagnetic detection system with a flexible pod structure (also referred to as "flexibly connected payload") faces multiple challenges, including unstable attitude, fluctuating altitude, and increased data noise during conventional level flight and terrain-following flight. Based on contour-derived terrain data, this study developed a highest-point algorithm for terrain-following navigation of UAVs with a flexibly connected payload. By introducing constraints on the payload attitude ceiling and the UAV flight ceiling, the proposed algorithm enabled terrain-following navigation of UAVs with a minimum number of waypoints while ensuring flight safety, thereby significantly alleviating the fluctuations of flight systems. Based on this, a matched terrain-following navigation software system was developed, consisting primarily of a geospatial data processing module, a flight route planning module, and a Blender 3D integration module. This software system allows for terrain data import, automatic flight route generation, and 3D visualization. Field tests conducted in the Gandel Mountain area in Wuhai City, Inner Mongolia, demonstrate that the proposed algorithm supported accurate terrain-following flight and enhanced signal quality and subsurface detection capability. This study provides a novel technical approach for mineral exploration and geological hazard monitoring in complex mountainous areas while also extending the application of UAV terrain-following navigation in geophysical exploration.