Pollutant Mean Concentration Estimation Of Yangtze Estuary Based On MSN Theory

Samples based inference is either classical statistics or interpolation,these methods have been widely used in various fields, different methods are based on certain assumptions or the geographical space must meet some certain conditions, but the research object actually does not necessarily meet these conditions.No matter what method is adopted, how to get a more reliable conclusion is the focal point of this work. This paper select the most significant pollution factors of the Yangtze Estuary, the spatial and temporal distribution status of the pollution factors were analysed using statistical method.Based on ArcGIS platform, visualization and data exploratory analysis was carried out. In order to achieve a better mean estimation,a stratification method based on hybrid-distace of the Yangtze Estuary was adopted. Using MATLAB software to implement the stratification algorithm.None-homogeneous area was divided into homogeneous areas using IDW interplation method based on ArcGIS platform. The focus of this research is to estimate the mean value of the two pollutants in the Yangtze Estuary using RStudio software with MSN method, mean value were also calculated by the other methods at the same time, a comparison of different methods from the aspect of estimation error was given finally. The study aims at providing a more accurate inference method for the mean pollutant estimation.The main research contents and results are as follows:1) Exploratory analysis for the distribution of inorganic nitrogen and labile phosphate from 2011 to 2013 was carried out.2) According to the heterogeneity of the Yangtze Estuary, a partiton for the purpose of mean estimation was put forward, that is using a combination of spatial distance and attribute distance based SOFM clustering method to determine the stratification of Yangtze Estuary, and the heterogeneity was tested to make fundation for improving the accuracy of mean estimation.3) Global and local variogram models of two pollutants for different sub-areas were fitted, and the fitting degree of the models were tested by cross-validation method.Taking autocorrelation and heterogeneity of the Yangtze Estuary into account,this article presented the application of MSN to make unbiased and optimal mean estimation of the Yangtze Estuary pollutants.MSN refers to mean of surface with non-homogeneity.4) Simple random statistical inference, stratified statistical inference, block kriging mean value were estimated and the estimation error is calculated.From the aspect of estimation error, MSN mean estimation method was compared with the other methods.Compared with stratified statistical inference,MSN has reduced 9%,11%,48% of the standard deviation of estimation error respectively from 2011 to 2013 for inorganic nitrogen.The corresponding reduced percentage is 30%,27%,46% for labile phosphate.Compared with block, MSN has reduced 57%,53%,56% of the standard deviation of estimation error for inorganic nitrogen and 41%,54%,60% for that of labile phosphate.The result showed that adopting heterogeneous surface average estimation can reduce the mean estimation error effectively,this can provide a more reliable and scientific basis for government policy-making.