The previous studies regarding types of temperature changes have been mainly related to the concept of the coefficient of similarity.During the computing of the coefficients of similarity in the past,the area weight of all the sites was thought to be equal,and the information obtained from the sites was believed to be uniformly distributed in the space domain.The grid and site climate data were found to generally meet the requirements of temporal uniformity.However,the data were notable to strictly meet the requirements of spatial uniformity (for example,the area represented by a single grid or site was different),which may have led to distortions in the statistical results involving the spatial regions.Therefore,in order to analyze the temperature variation similarities,the key was to process the network density in homogeneity.In this study,based on a scheme of network homogeneity correction proposed by Wang Panxing,a new calculation method for the coefficient of similarity was presented.This new method considered the weight of each area of a site.Therefore,the coefficients of similarity were determined to be more accurate.
Two schemes were used in this study to choose similar cases.The first scheme was as follows:1)Two fields were found to be similar before the correction.However,they were observed to be not similar or reversed after the correction.Alternatively,two fields were not similar or opposite before the correction,but were found to be similar after the correction;2)The absolute values of the differences between the coefficients of similarity before and after the correction tended to be the maximum.
The second scheme used in this study was as follows:1)The first condition was the same as 1)in the first scheme;2)The coefficients of similarity before and after the correction tended to be the maximum.
Also,according to the aforementioned two schemes for selecting similar cases which were utilized in this study,the inter-annual similarity features of the air temperature anomalies at 160 stations in China during the four seasons ranging from 1951 to 2010 were examined.The results showed the following:1)According to the first scheme,if a pair of air temperature anomalous fields were more similar in the eastern region,and less similar in the western part,then the adjusted national similarity coefficient was observed to be significantly reduced.Otherwise,if a pair of air temperature anomaly deviation fields were less similar in the eastern region,and more similar in the western part,then the adjusted national similarity coefficient was found to be significantly increased;2)According to the second scheme,if the coefficient of similarity of a pair of air temperature anomalous fields was significantly larger before the correction,then the pair of fields were observed to be very similar in the eastern region,and the adjusted national coefficient of similarity was significantly reduced.Otherwise,if the coefficient of similarity of a pair of air temperature anomalous fields was significantly larger after the correction,then these two anomalous fields were definitely very similar in the western region,and the adjusted national similarity coefficient was observed to be significantly increased.The two schemes showed that the network homogeneity adjustments led to the weight of the eastern region becoming reduced,and the weight of the western region increasing in the national average,which more objectively reflected the similarity degree of the inter-annual air temperature anomaly.