A comparison was made using automatic weather station measurements and conventional observations from November 2002 to February 2003 (temperature, pressure, vapor-pressure, relative humidity and soil temperature of 5 cm to 320 cm in the winter of 2002. The results show that observational errors are relatively small for temperature, station-pressure, vapor-pressure, relative humidity, and soil temperature of 320 cm, and greater for soil temperatures of 5 cm to 160 cm. The chief reasons for observational errors include systematical errors, observational precision, and the sensitivity of the system's responding to changes of weather elements and observational time, as well as some artificial factors.