In an accident of power supply on April 24,2015，Ningbo seismic station found clear interferences of high accuracy thermometer in the same layer observations. We believed that the water temperature observed is a joint result of the actual temperature and the working temperature rise of the sensor. This rise is a dynamic value，depending on the systematic characteristics of the instruments，with an average of 0.01℃ and more than 0.005℃ with interference. The rise can be higher in a parallel observation. Therefore, in an observation with the temperature fluctuation of 10－4℃，the interference should not be ignored which could lead to precursory anomalies. We therefore suggest that no parallel observation of the same layer be needed in wells of small fluctuating temperatures. There should be two sets of temperature detecting devices in a sensor so as to avoid mutual interference but gain mutual confirmation.