The new far‐field low‐frequency electromagnetic method used the electromagnetic ground wave from distant radio transmitters at low frequencies to estimate temporal variations of factors affecting subsurface electrical conductivity averaged along the propagation path between either a transmitter and a receiver or two receivers that are in line with a transmitter. Phase and phase difference between two receivers depend on three factors that influence the changes in electrical conductivity: soil moisture, depth to the groundwater table and soil temperature. This dependence was investigated by simulations and evaluated by an experiment. The measurement layout was based on simulating ground wave propagation over a layered subsurface using the surface impedance method and the Sommerfeld ground wave attenuation function. A three‐layer model for the subsurface was used, which includes a soil layer, an unsaturated vadose zone and a saturated groundwater zone. The results of simulations at a frequency of 77.5 kHz showed that the phase of the ground wave is strongly influenced by natural variation of the above‐mentioned three factors; 77.5 kHz is the carrier frequency of the Normal Time Service Germany (DCF77) in Mainflingen/Germany, that was chosen as a source of the low‐frequency radio waves used in the experiment. Over a 2‐year measurement period, the amplitude and phase of the ground wave were recorded with two receivers, one 70 km and the other 110 km away from the transmitter. Additionally, phase difference between the two receivers was calculated. In situ observations of soil moisture, depth to the groundwater table, and soil temperature along the transects under investigation were used to estimate phase and phase difference dependencies. Multiple regression analysis of the measured phase and phase difference revealed a strong dependence on the depth to the groundwater table and on soil temperature, whereas the impact of soil moisture on the phase and phase difference was found to be very low. Conversely, the relations obtained can be used to estimate the variation of the depth to the groundwater table, if the phase at a given frequency and the soil temperature information are available.