| Sustainable water resources assessment is the basis and premise of sustainable water resources management, and indicators system for sustainability assessment of regional water resources use and evaluation methods are two core contents in the process of assessment. This paper conducted researches in three aspects. Firstly, based on the general methods and steps of establishing the indicators system for sustainability assessment of regional water resources use (ISSAWRU), this paper revolved around initialization, optimization and evaluation test those three steps to establish ISSAWRU. In order to decrease the overlapping and interfering information of indicators, correlation analysis and Horafa algorithm for attribute reduction of rough set were combined to optimize the initial ISSAWRU. Moreover, the objective evaluation method coupling grey system theory with attribute importance degree computed by rough set was proposed to examine the optimized result. Nine districts of Fujian province was taken as an example in this research, and the initial ISSAWRU which is composed of 30 indicators selecting from four aspects including water resources condition, utilization status, ecological environment and socio-economic, was established based on the theory of sustainable development. Then the optimized ISSAWRU composed of 12 indicators was obtained through applying correlation analysis and Horafa algorithm. Secondly, this paper put forward RS-SVM model to evaluate the sustainable utilization of regional water resources of ISSAWRS in Fujian province, and compared with single application effect of SVM model. Thirldly, in order to further reveal several evaluation methods of regional water resources sustainable utilization and its application effects, seven evaluation methods including principal component analysis (PCA), analytic hierarchy process (AHP), gray correlation method, improved rank correlation analysis, fuzzy comprehensive evaluation method, BP neural network and SVM were selected to evaluate the optimized ISSAWRU which has been obtained.The results show that:(1) Method combined correlation analysis and Horafa algorithm is reliable to optimize ISSAWRU. Additionally, screening overlapping and interfering information can simplify the complexity of indicators system and reduce the computational complexity of follow-up evaluation. Furthermore, the evaluation result of optimized ISSAWRU which have been screened indicators information can obtained the identical or more reasonable evaluation result than it of initial ISSAWRU. (2) Using optimized ISSAWRU did not reduce the classification accuracy in SVM model, but it reduced the dimensions of input vectors as well as the computing complexity. Besides, compared with SVM model, RS-SVM model can achieve similar or even more reliable result in assessment of regional water resources sustainability. (3) The evaluation results of PCA and AHP method have the highest consistency, and those two methods have the most reliable in comprehensive index type methods. While improved rank correlation analysis method got a higher assessment value of Quanzhou. Moreover, gray correlation method has some shortages that it may obtain smaller range of evaluation value and higher assessment value of some regions, which has lower level of regional water resources sustainable utilization actually.In addition, the evaluation results of fuzzy comprehensive evaluation, BP neural network and SVM are quite different without obvious regularity. In general, traditional methods including PCA, AHP, gray correlation degree and fuzzy comprehensive evaluation method are relatively simple and stable. Although BP neural network and SVM can reduce subjective intervention, but they show poor stability because of the impacts of index ranking standard, training sample, parameter settings or some other factors. Compared with BP neural network, evaluation stability of SVM is better, but the reliability of SVM is greatly affected by division of sample space. (4) Sustainable water resources utilization level of 9 districts ranked in descending order as follows:Sanming, Longyan, Nanping, Ningde, Quanzhou, Zhangzhou, Fuzhou, Putian and Xiamen. In terms of classification results, the regions with Ⅰ degree of sustainability were Sanming, Longyan and Ningde while the region with Ⅱ degree was Nanping, And Quanzhou, Zhangzhou, Fuzhou and Putian were the regions with Ⅲ degree. Xiamen was the only region with IV degree. Overall, The spatial pattern of water resources sustainable use presented decreasing trend from the northwest mountain areas to the southeast coast across Fujian Province. |
PDF Full Text Request