Homogenization and analysis of China radiosonde temperature data from 1979 to 2012

Using the PMTred (penalized maximum t test) method developed by the Environment Canada with the ERA-interim reanalysis data as the reference series and first using the detailed metadata of each radiosonde station in China as the judging basis, the inhomogeneity test and corrections are carried out on the monthly temperature data at mandatory levels at the 125 radiosonde stations in China. And combining with the detailed metadata, the main reasons causing the homogenization problem of the radiosonde temperature data in China have been analysed. The result shows that there are break points due to human factors in the radiosonde temperature data of China. With the height increasing, a general trend is presented, that is, both the number of the inhomogeneous stations and break points show less in lower levels and more in high levels. Also, the mean magnitude of correction increases with height and the negative correction values have the high proportion in the whole dataset, implying that the radiosonde temperature data in China is systematically higher. After adjustment, the temperature trend on 200 hPa level becomes close to zero. The cooling trend on 100 hPa level weakens after adjustment, but the warming trend strengthens on 300-700 hPa. The update of the sounding systems in 2000 and 2001 including the radiation correction method is the main reason of the inhomogeneity in Chinese radiosonde temperature data. Another reason is the shifts of the radiosonde instrument. This two consecutive metadata changes caused the following monthly mean temperature data appears systematically lower. This is also the main reason why there are some differences in the temperature trend between before and after adjustment.