Optimal geostatistics model and its application of suspended sediment in Taihu Lake, China

Suspended sediment (SS) is an important parameter to indicate pollution status of aquatic environment. Most of the published papers to date concerning SS pollution are focused on the degree of SS concentration, and there is little literature concentrating on the spatial diversity of it in surface water bodies. Study on spatial diversity of the SS concentration can help authorities to make more appropriate policies and take more effective measures. In the research of water pollution of inland lakes, the emphases focus on not only SS concentration but also its spatial diversity in lakes. Geostatistics method, which I widely used to analyze the spatial structure and distribution of regionalized variables, can be applied to simulate the spatial diversity of SS in lakes. The key procedure during the application of geostatistics method is optimizing model parameters and choosing the optimal model. Based on the sampled data of SS concentration in Taihu Lake, the parameters of four geostatistics models (spherical model, linear model with sill, exponential model and gauss model) are optimized. Though model comparison, the optimal one is selected. With the calculated variogram and the method of 2-Stage Least Squares, the optimal parameters of spherical model are: nugget 100.11, sill 135.22-and range 33.83; those of linear model with sill are: nugget 100.16, sill 150.19 and range 44.89; the optimum values of exponential model are: nugget 90.48, sill 148.01 and range 51.33; and those of gauss model are: nugget 104.82, sill 155.19 and range 2.01. Based on the optimized parameters of these models, the residual sum of square (Q), standard error (S) and coefficient of determination (R-2) of these four models are calculated. Through comparison of Q, S and R-2 of these four geostatistics models, the result indicates that Q and S of spherical model are the,minimum (375.62 and 6.85 respectively). R-2 of spherical model is the maximum (0.75). By comparison of the four geostatistics models, spherical model of SS concentration in Taihu Lake is regarded as the optimal one. At the end of this paper, the spatial diversity of SS concentration in Taihu Lake is simulated by the optimal model, i.e. spherical model.