Study On Coastal Water Environmental Capacity

With the rapid economic development and the population booming in coastal zone, the coastal water quality was deteriorated. In order to control the coastal water pollution effectively, its characters and regulations must be mastered, based on which the coastal water environmental capacity could be calculated. The studies before which mainly concerned the mechanisms had taken little use of the environmental monitoring data, so they were inapplicable to management in the thinking of economics and practicability. In this dissertation, through the integrated using of many theories and methods in different fields, the prediction, simulation and assessment of the water quality in Tianjin coastal marine were studied totally based on the environmental monitoring data. Furthermore, a new calculation method for coastal water environmental capacity was proposed, which could supply the decision support for coastal pollution total amount control and environmental management. Firstly, the association analysis among monitoring points was taken based on grey system theory. In this section, the B type association degrees between points were calculated and then the association degree matrix and its sort order matrix were set up. According to the information in the above two matrix, the monitoring points were clustered by k-means method. What's more, the concept of point sensitivity was purposed. Then, the studied area was divided into three management subareas and the idea of coastal optimal management was presented. Secondly, according to the information about the inlets, the COD values of each monitoring point were predicted by NN method successfully, during which the over-fitting problem was solved by Bayesian automatic regularization technology. And, the monitoring data serial of each point was also fitted by Fourier serial whose parameters were decided by robust weighted nonlinear least squares algorism, after which the Fourier serial was also used for prediction. Finally, the integrated prediction was taken based on the above two prediction results, whose weights were calculated by its log-logistic probability density. This method was demonstrated to be effective to reduce the prediction error. Thirdly, based on the batch version of SOM algorism and SOM toolbox in MATLAB environment, the coastal water quality assessment software was compiled. Through the training of SOM, the monitoring data with high dimension could be abstracted and displayed in a two-dimension figure. Then, based on some research findings in environmental science, monitoring data was clustered into 5 pollution kinds. Furthermore, the data trajectory tracking and automatic classification were also realized in figure mode. So the water quality assessment could be simplified and improved. At last, for the first time, the spatial variation of contaminant concentration in the coastal marine was analyzed using geostatistical theory. Three theoretical models were used to fit the experimental variogram of each period, based on which the ordinary Kriging interpolation method was employed to estimate the COD concentration in the whole studied area. After that, the worst area was located by raster GIS maps algebraic functions. Finally, the water environmental capacity under the worst condition was calculated according to the conservation of mass.