Research On The Spatial-temporal Distribution Characteristics And Quality Control Methods For Surface Precipitation Data

Precipitation is a climatic variable that reflects the dynamic and thermal processes of the earth’s atmosphere,and is also the basic driving factor for natural disasters and extreme events.The main source of ground precipitation observation data is the regular observations of automatic weather stations.A large amount of observation data contains random errors,systematic errors and gross errors.The use of these data often leads to erroneous results in precipitation prediction and related scientific research.Therefore,it is of great significance to perform quality control on them.Corresponding quality control methods have been difficult points in research.According to the characteristics of precipitation and its data quality control methods,this paper has carried out corresponding research.The main work is as follows:The spatiotemporal distribution characteristics of precipitation in Jiangsu Province for many years and the abrupt change characteristics of various solar precipitation in summer were analyzed.Regarding the multi-year precipitation in Jiangsu area,the linear trend and moving average analysis are first performed,and then the empirical orthogonal function is used to decompose it.According to the variance contribution rate of eigenvalues,Jiangsu is divided into three types of precipitation characteristics,and its spatial distribution is analyzed.Then the wavelet analysis method is used to analyze its periodic scale and periodic oscillation.Finally,the annual precipitation scale was reduced to the scales of various solar terms in summer,and the changes and abrupt changes in the average precipitation anomaly of the solar terms were analyzed by using linear tendency estimation and M-K test.From the perspective of the spatial characteristics of precipitation,three commonly used spatial interpolation models based on spatial distance,inverse distance weighting(IDW),ordinary kriging(OK),and coordinated kriging(CK)considering elevation are established under complex terrain and use three model evaluation indicators to compare and analyze them.In addition,the sensitivity of the three interpolation methods to spatial interpolation samples in the case of sparse sites simulated by random sampling is compared.The results show that CK has the best effect among the above three interpolation methods,but the accuracy decreases significantly as the number of sites decreases.Considering that the above-mentioned spatial interpolation method is limited by the number of spatial stations and the integrity of the stations data.From the perspective of single station precipitation quality control,firstly the slow forcing analysis(SFA)is used to extract the external forcing factor information of the precipitation time series,and then embed the additional forcing factor information into the chaotic weighted zero-order local prediction(WZLM)model.Finally a prediction model(SFA-WZLM)that includes external forcing factors is formed.By comparing and analyzing the weighted first-order local prediction model(WFLM)and its external forcing mode(SFA-WFLM),which are also chaotic local prediction methods,a single station precipitation quality control method based on SFA-WZLM is proposed.The experimental results show that the prediction results of the model are better than the three spatial interpolation models involved in this paper,and the experimental results at six stations across the country have verified its good accuracy and applicability.The method also makes up for the deficiency of single station precipitation quality control.