关键词: 虚拟水流动格局, 影响因素, 神经网络, 决策与试验评价实验室法

Abstract: The virtual water has become one of the hot spots in the research field about water recourses and water safety. How to predict and control the virtual water flow pattern has become an important precondition for implementing the virtual water strategy with Chinese characteristics, so identifying the key influence factors of virtual water flow pattern is very important. Virtual water interregional flows for major crops in China were explored and virtual water flow during 2008-2010 was estimated. It is found that the virtual water flow for major crops is still from North China to South China. In order to quantificationally research the influence factors of the virtual water flow pattern, a BP-DEMATEL model is built, which is commonly used by the Decision Making Trial and Evaluation Laboratory in the field of recognizing the influential factors. And the BP-Neural Network is introduced, which can replace the expert scoring to obtain the direct correlation matrix. The method is used to analyze the main influence factors of China's virtual water flow of the major crops and to reveal the mechanisms of each influence factor. It is necessary to further divide the influence factors into four types, i.e., strongly driven, driven, strong feature type and characteristics, which provides a theoretical support for scientifically and rationally adjusting the virtual water flow pattern. The influence factors are further divided into four types, i.e., strongly driven, driven, strong feature type and characteristics, which provides a theoretical support for scientifically and rationally adjusting the virtual water flow pattern. The influence factors of the virtual water flow pattern of the major crops in China are analyzed from the following six aspects: population, agricultural resources, agricultural ecological environment, economy, technology and transportation and logistics. The results show that population factors, agricultural resources factors and technology factors are the principal influence factors; fertilizer input is the strong driving factors, the proportion of the first industry accounted for the GDP is the strong feature type factors. Finally, the following three measures are suggested to be taken: (1) optimizing and adjusting the population structure and intensifying the favorable policies; (2) transforming the mode of agricultural development and developing recycle agriculture; (3) improving the level of agricultural technology and developing high-efficiency agriculture.

Key words: virtual water flow pattern, the influence factors, BP Neural Network, DEMATEL