Abstract:

Combining reanalysis data as a reference series and the detailed metadata of each radiosonde station in China, homogeneity test and corrections were carried out for monthly temperature data at mandatory levels at 123 radiosonde stations from 1951 to 2008 in China using a Pairwise method developed by America National Climatic Data Center, PMFT and PMT methods developed by the Environment Center of Canada. The results show that adjusted values of radiosonde temperature data in China and its trends before and after correction have some differences for three different methods. One of the main reasons is that design of the Pairwise method leads to the weak capability for break points detection if majority regional stations of China change their radiation correction method and update the system simultaneously. Another reason is that the PMFT method cannot remove the climate temperature trend because of no reference series, so some break points cannot be detected. The PMT method combines with reanalysis data, which is more suitable for homogeneity test and correction of radiosonde temperature data in China. The adjusted results show that the statistical characteristics of radiosonde temperature are greater than those of the global scale at the low level, and both are consistent at the high level. Two break points and adjusted values between -0.2 ℃ to 0.0 ℃ have a larger proportion in datasets. After adjustment, the troposphere temperature average over China is warming trend. This warming trend weakens gradually with increase of heights and becomes a weak cooling trend till 100 hPa.

Key words: Radiosonde temperature data, Homogeneity test, PMFT test, Pairwise method, PMT method