Research On Quality Control Method Of Surface Temperature Observation For Single Station

Nowdays,with the improvement of living standards,people are increasingly concerned about the accuracy of weather forecasting.Data assimilation technology is an important means to improve the level of weather forecasting and its primary task is to provide effective quality control of surface meteorological observations.Because of China's vast territory,complex terrain climate,uneven distribution of meteorological stations and other factors,resulting in surface meteorological observations is poor,and only a few of them can be used to the data assimilation system,which is difficult to meet the demand of current meteorological services.Based on this,also takeing the distribution characteristics of surface temperature observation data in time dimension as the starting point,and considering the distribution density of meteorological stations and the completeness of temperature observations,this paper establishes the corresponding single station quality control model,and conducts in-depth analysis on the generalization performance and stability of the model.The main contents are as follows:The trend analysis method and multi-fractal trend-wave analysis technology are used to analyze the surface temperature data from the time dimension for long-range correlation analysis and multi-fractal characteristics analysis,which provide a theoretical basis for the construction of quality control model.Based on the long-range correlation and considering the white noise in the temperature observations,the EEMD-CES single station quality control method is constructed by using the empirical mode decomposition(EEMD)and the cubic exponential smoothing(CES).Based on the multifractal characteristics,considering the poor completeness of the observation data in low-density areas,the FSO-FI-based single station quality control method is constructed by using fractal interpolation(FI)method.Meanwhile,the parameters are optimized and the interpolation points are optimally selected.The analysis of multiple experiments shows that the surface temperature observation data not only has strong long-term correlation,but also has complex multifractal characteristics.Based on the distribution characteristics,the EEMD-CES single-station quality control algorithm shows high error detection ability in both high-density and low-density observation sites,and can solve the quality control problems of most observation stations.For the meteorological stations with low distribution density of adjacent stations or serious missing observation data,the PSO-FI single station quality control method uses the fractal interpolation method to interpolate the data according to the fractal characteristics of the temperature observations.The quality control of observation data is realized from another perspective,and compared with the traditional single station quality control method,it has the remarkable quality control effect.