Study On Station Network Homogeneity Adjustment For Climatic Data Of China And Its Application

Climatic data collect directly from the meteorological observation, which is the foundation of atmosphere general circulation and climate analysis. Distribution heterogeneity of observation station network impacts seriously on the accuracy of the results from the element field time series analysis. Based on sphere geometric knowledge, the two parameters station domain area di and network density mi for measuring heterogeneous station network are defined, and their calculation method is given, which are universal. The mi and di of the network from 160 stations are calculated with this method, which may objectively indicate the spatial uniformity of this station network. By use of station domain area, the first weight function 1wi is constructed, and the national mean seasonal and annual air temperature series in the past 60 years (1951～2010) are calculated. The results show that the linear warming rate of the national annual mean temperature is 1.4℃/60a, the rate of winter (2.2℃/60a) is apparently higher than that of summer (0.7℃/60a). The maximum value of the mean square deviation in the national mean seasonal temperature is in winter (1.62℃) and its square deviation has a high ratio 58.2% in four season's total anomaly, while the minimum value is in summer (0.75℃) and its square deviation has a small ratio 10.7% in total anomaly. Using the station domain area, the second weight function 2wi is constructed and the adjusted EOF analysis (AEOF) method is proposed. By use of AEOF analysis, anomalous summer and winter air temperature series for 1951～2010 are analyzed and the results show that:results from the two aspects about variance contribution and space modes with AEOF analysis are superior to those of EOF. No matter how in summer or in winter, the variance contribution rates of the leading three orthogonal modes with AEOF analysis are both higher than 10% and the cumulated variance rates of all modes are greater than those of EOF analysis. The total variance field of the field series reconstructed by the leading three modes with AEOF analysis is more similar with the total variance field of the anomalous summer and winter air temperature time series, and its cumulated variance fitting rate reaches 56.6% and 73.4% respectively, which are significantly higher than the rate 52.2%(in summer) and 71.1%(in winter) corresponding to EOF analysis. Furthermore, the typical fields and time coefficients of the main modes in summer and winter AEOF analysis focus much more on reflecting the main features of anomalous spatial and temporal variations of the anomalous summer and winter air temperature. They are more reasonable than the available results and analysis methods.