Research On Spatio-temporal Variation Characteristics And Quality Control Method For Surface Air Temperature Observations

As a significant forecasting technology in meteorological operation,the core competitiveness of numerical weather prediction lies in the level of data assimilation technology,which requires accurate and complete meteorological data.With the continuous popularization and development of automatic weather station,the amount of surface air temperature observations increases geometrically,consequently,it is more necessary to improve the quality control level of surface air temperature observations.Based on the analysis of spatio-temporal variation characteristics of surface air temperature observations,a statistical algorithm is introduced to establish the quality control model of surface air temperature observations,which is aimed at the feasibility,accuracy and applicability of the quality control method.Finally,the appropriate indicators are selected for conclusion analysis when the corresponding verification and contrast tests are carried out.The main contents are as follows:The problem has been considered that the current surface air temperature analysis methods are insufficiently for data analysis on multiple time scales and in different regions,thus an optimized kernel density estimation algorithm(PA-KDE)is proposed.The timed temperature of 12 stations in Jiangsu Province from 1961 to 2008 are selected as observations to verify the sensitivity of the method at the quarterly,monthly,daily and nightly time scales,and the method is used to estimate the density function of observations in Jiangsu Province,then spatio-temporal variation characteristics are analyzed by region and season.Based on the analysis of surface air temperature observations,in order to meet the basic needs of quality control,taking the methods and conclusions of surface air temperature analysis as a reference,the kernel regression algorithm is introduced into the quality control of surface air temperature observations to solve the limitations of the current quality control methods in different regions and on multiple time scales.Based on the completion of feasibility and applicability tests in different regions,an improved kernel regression(IKR)method optimized by an adaptive algorithm and particle swarm optimization algorithm is applied to the quality control model of surface air temperature observations,which is aimed at the problem that the accuracy and universality of the KR algorithm fail to achieve the expected effect.Surface air temperature of 14 regions in China from 2010 to 2014 are selected as observations,the analysis of quality control effect is performed in terms of the mean absolute error(MAE),root mean square error(RMSE),consistency indicator(IOA),Nash-Sutcliffe model efficiency coefficient(NSC)and error detection rate.The analysis results of multiple groups of tests show that,the PA-KDE algorithm has higher sensitivity at the quarterly,monthly,daily and nightly time scales,the characteristics of spatiotemporal variation and the effects of different influencing factors can be more comprehensively analyzed simultaneously.Thus more attention should be paid to the effects at different time scales,regions and seasons when the characteristics of spatiotemporal variation and different influencing factors are discussed.In the overall quality control model,compared with the traditional IDW and SRT methods,the IKR method has a high error detection rate.Furthermore,the IKR method achieves better predictions and fitting in the national multi-station regression prediction experiment.Therefore,it is more conducive to the research of the quality control of surface air temperature observations.