Robust Analysis Of Spatial Variable Data In Water Environment Simulation

The dissertation brought in advanced achievements of Robust Statistics, dealt with the outliers and did the robust analysis of the monitoring data of spatial variables, which had filled up the short of traditional geostatistics and made the processing mode of the data more scientific and rationalization.With the data got from the experiment and scene, this page forecast the estimated accuracy and analyzed the impact factors, and then by the Influencing Coefficient Method and the Estimating Neighborhood Method, we recognized the outliers and then did some justice. The result showed the estimation effect after processed by the both methods had advantages over those not processed. To those data with high spatial ability, the Influencing Coefficient Method had advantages over the Estimating Neighborhood Method. We got the variation function which had high spatial constitutive property through the computational method of Robust variation function, and the robust calculation effect of the data after outlier processed is more obvious. The article brought in Indicator Kriging of non-parametric geostatistics to analyze the data which had skew distribution and nonstationarity. The result showed it had more robust than the Ordinary Kriging.The calculation and analysis showed the methods of Robust Statistics can be brought into the environmental science, which had high application and learned value to strengthen the spatial variability and the basic study of Robust variation function.