Differences Between Automatic-observed and Manual-observed Surface Temperature

Based on the parallel comparison of original base stations during the transition from manual observation to automatic observation and the 0 centimeter temperature data from parallel observation of base stations in 2005 in China, the analyses are made on the differences between automatic-observed and manual-observed surface temperature data in time scales of day, month and year. The method of maximum likelihood ratio has been applied to test the homogeneity of the monthly values of 0 centimeter surface temperature. And the homogeneity of surface temperature affected by automatic observation is discussed. The results show that the daily mean value from automatic observation is 0.54 ℃ higher than that from manual observation in 0 centimeter surface temperature of China. The 0 centimeter surface temperature, 0 centimeter maximum and minimum surface temperature have annual comparison values over 0.0 ℃ with the percentage of 80.3%, 58.2%, 92.2%, respectively, and in most stations the automatic-observed annual mean value is higher than manual-observed one. The average daily value difference between auto and manual observation generally decreases gradually from north to south, and the average daily value varies the greatest in the north of Helongjiang Province and the north of Inner Mongolia to 45°N, and most areas in Xinjiang Autonomous Region. In all the time scales including day, month and year, the difference between automatic-observed and manual-observed temperature is greater in cold time period than warm time period, especially in the winter of the north. The main cause is that when there is snow covering the ground in winter, the surface temperature from automatic observation is the temperature under the snow, which obviously is higher than the temperature above snow from manual observation. If without the influence of snow cover, then there is no obvious difference between the two kinds of observation. The instrumental difference and the ground feature both contribute to the observational difference. The inhomogeneity test shows that the inhomogeneity from 0 centimeter surface temperature in the north area lies in the change of observation in automatic observing stations, while in the south, the change of observation does not exert much influence on the inhomogeneity of surface temperature. When there is snow cover in the north, 0 centimeter surface temperature from observation does not present the real situation, which should be taken into consideration while using the data.