Homogeneity Study Of The Temperature And Precipitation Data Over China Using Multi-methods

Homogeneous long climate series is the basis for conducting and researching the characteristics of climate change, especially the researching with regard to the distribution of climate and extreme climate change. However, the inevitable changes of the location of stations, observation instruments, the surrounding environment, the calculation methods will lead to non-homogeneity in observation series. The non-uniform in the series not only has an impact on the climate and its trends but also the extreme. Using the latest improved PMFT (penalized maximal F test) and PMT (penalized maximal t test), combined with the two-phase regression test and the metadata of the migration of the stations, the homogeneity of the monthly average temperature, the monthly minimum temperature and monthly maximum temperature of the433stations is tested and revised. Besides, using the PMFT and PMT test, the precipitation data of the1725observation stations in china is tested and revised. The trends of the adjusted temperature and precipitation series are analyzed. Besides, the changes and trends of the adjusted annual and seasonal temperature of the five provinces in northwest China are analyzed. At present, the work has made the following conclusions:(1) Using the penalized maximal F test, the penalized maximal t test, two-phase regression test and the metadata, the homogeneity of the433stations of monthly temperature is tested. There are224non-homogeneous stations and264non-homogeneous points, accounted for51.8%of the total number of average temperature stations. There are191non-homogeneous stations and246non-homogeneous points, accounted for44.1%of the total number of maximum temperature stations. There are233non-homogeneous stations and313non-homogeneous points, accounted for53.8%of the total number of minimum temperature stations. In the non-homogeneous stations, the large distance change of location is one of the main causes for the non-homogeneity of some stations. There is largest number of non-homogeneous stations in the minimum temperature, followed by the average temperature, at least for the highest temperature. This may be related to the atmospheric stability.(2) After homogenization revised, the regression coefficients and variance distribution of the original temperature series have been significantly improved. The before and after adjusted maximum annual temperature are both17.1℃. The trend of the annual average temperature before the adjustment is0.023℃/a but it turns to0.027℃/a and0.030℃/a after difference and QM adjustment. After the two adjustment, the trend decreases about0.04℃/10a and0.07℃ /10a. The before and after adjusted minimum annual temperature are both6.2℃.The trend is both0.04℃/a and has not changed after the adjustment.(3) Based on the results of the test, the variation of the annual and seasonal temperature data of the142meteorological observation stations in northwest China has been analyzed. The results show that:there is a clear upward trend in the series of the average, maximum and minimum, annual and seasonal temperature in this region both before and after the adjustment. The warming rate after the difference or QM adjustment is different from the warming rate before adjustment, but the difference is small. The adjustment range is about0.01℃/10a.The increased rate of the average, maximum and minimum annual temperature are about0.33℃/10a,0.28℃/10a and0.42℃/10a. For the average, maximum and minimum seasonal temperature, the winter has the largest increased rate, followed by spring and autumn and the summer has the smallest increased rate. Using the moving t test, we also find:there is a significant mutation in the year of1997which is from cold to warm in the average winter temperature series of this region but the mutation becomes not significant after the QM adjustment. Besides, there are also two significant mutations in the year of1996and2003which are from cold to warm in the minimum spring temperature series but the mutation in the year of1996disappeared after the adjustment. Except the two mutations mentioned above, the other mutations are same before and after the adjustment.(4) Using the Rhtest, the homogeneity of the precipitation of the1725stations is tested. The results show that the homogeneity detection method works well for the examination of the precipitation data over China and has a high prospect. It is more beneficial to the selection of the reference series for using2400stations monthly precipitation data which has the higher density and it is more reasonable to detect the homogeneity of the series, using both the mathematical statistical analysis and the subjective judgments with the metadata. Only about2.1percent records are non-homogeneous among the detected stations which indicates the homogeneity of the precipitation is good in most areas of China. In the non-homogeneous stations, the change of location is one of the main causes for the non-homogeneity of some stations. Applying the method of Rhtest, the series which have breakpoints have been adjusted and the trend of the series has been analyzed. We find that the yearly precipitation trend has been changed obviously and the homogeneity has been improved well.