Abstract: Climate data is the basis for studying climate change. In addition to the regional climate change factors, single station climate data can also carry non-uniform information due to the human factors such as urbanization and land use,and other factors like station migration and the instrument change,resulting in conflicting results of climate change studies. By using the difference sequences,trend value,and t-test analysis,and based on the observational data of Turpan and reference stations(Toksun and Dongkan Station)from 1974 to 2014,this paper first analyzed the regional coherence and uniformity between the Turpan and reference stations. Then,the difference and continuity between new and old station data were analyzed based on the observation data from 2013 to 2014. Finally,by comparing the observation data from Turpan Station and other six moved plain stations which already had one-year comparative observations data,this paper analyzed the special difference of arid region station information before and after migration. Results showed that the air temperature,precipitation,relative humidity,and wind speed have poor consistency between Turpan Station and the reference stations,of which the annual average temperature and precipitation showed bigger rising trend in Turpan station than that in the reference stations,relative humidity and wind speed showed bigger decreasing trend in Turpan station than that in the reference stations. Because of urbanization,the temperature rising rate reached 0.24 ℃·(10 a)^-1,and breakpoints appeared in 1982 and 1998 for temperature,relative humidity,precipitation and wind speed according to the homogeneity test. Comparative analysis showed the observation data were quite different due to the station environment between the old and new station. Compared with the old station,the average temperature and the lowest temperature were 0.9 ℃ and 1.5 ℃ higher,relative humidity was 8.1% lower,and the annual average wind speed was 1.4 m·s^-1 larger. comparison results between the Turpan Station and other six moved stations showed the annual mean temperature,minimum temperature,relative humidity,the difference values and standard deviation between new and the old stations in Turpan station were twice larger than most stations,the consistent rate of wind direction between the new and old Turpan stations are the smallest with 32.29%. Thus in arid area,big environmental difference can be caused when the new station are without vegetation,and further leading to larger difference between old and new observations,especially when the new station underlying surface is Gobi. Therefore,it is recommended to choose new station in grass region or in rural area,and try to avoid Gobi area as far as possible.

Key words: station migration, difference analysis, comparison station, trend, homogeneity