Interpolation and Inhomogeneity Test of Missing Measurement Records in Long-term Temperature Series at Xi'an Observation Station
Based on the series of daily mean temperature,maximum temperature and minimum temperature in Xi'an observation station from 1971 to 2013,the interpolation experiments are carried out by using standard series method and multiple linear regression method,calculating the average error,average absolute error,root mean square error and the proportion of samples with the error between the interpolation value and the measured value within 0.5℃.The relative advantages and disadvantages of the two interpolation methods are compared and analyzed The experimental results show that the daily temperature series which obtained by the multiple linear regression method is better than the standard series method,and the characteristics of climate trend are more consistent with the actual observed data series.The t-test,the penalized maximal T test(PMTT)and the penalized maximal F test(PMFT)are used to test the homogeneity of the annual mean temperature series in Xi'an observation station from 1951 to 2020.The test results show that according to the t-test conducted on the historical evolution data of the station,only 2 of 6 times changes caused the discontinuity of the annual mean temperature and annual mean maximum temperature series,which were respectively caused by the increase of observation time and the change the instrument type.There were four discontinuities in the annual mean minimum temperature,which were caused by the relocation of stations,the increase of observation time,the replacement of instruments and the interpolation of missing measurement values.The four discontinuity points found in PMTT and PMFT detection are considered to be reasonable discontinuity points because there is no metadata support,and none of the two methods detected discontinuities caused by interpolation of missing measurement values,which indicates to a certain extent that the temperature series of Xi'an observation station from 1951 to 2020 obtained by the interpolation of missing measurement values with multiple linear regression method is relatively reasonable,and the temperature series have good homogeneity.
国家科技期刊平台
NSTL
万方数据
维普
