Scholarlink

Comprehensively identifying key nodes in virtual waterVirtual water network and predicting variations in virtual waterVirtual water characteristics remains a significant challenge. In this study, an input–output analysisInput-output analysis CNN-LSTM (IOACL) model is developed to identify the sources of virtual waterVirtual water consumption within the industrial chain and predict future water usage. IOACL is applied to Ningxia, one of the most water-scarce provinces in China. The major findings are as follows: (i) agriculture, food & tobacco and construction have high virtual waterVirtual water consumption, collectively accounting for 70% of the total virtual waterVirtual water consumption; in terms of physical water consumption, agriculture, construction and metal smelting are major contributors, together representing 94% of the total physical water consumption; (ii) food & tobacco, textiles and gas supply exhibit higher virtual waterVirtual water consumption compared to physical water consumption, which indicates that their supply chains contain large amounts of virtual waterVirtual water; (iii) compared to LSTM, the CNN-LSTM model demonstrates superior performance in predicting water consumption in Ningxia, achieving an R2 value of over 0.9.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Input–output Analysis CNN-LSTM Model for Analyzing the Virtual Water Consumption

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shizhuo Li,
  • Yongping Li,
  • Guohe Huang,
  • Panpan Wang,
  • Yanfeng Li

摘要

Comprehensively identifying key nodes in virtual waterVirtual water network and predicting variations in virtual waterVirtual water characteristics remains a significant challenge. In this study, an input–output analysisInput-output analysis CNN-LSTM (IOACL) model is developed to identify the sources of virtual waterVirtual water consumption within the industrial chain and predict future water usage. IOACL is applied to Ningxia, one of the most water-scarce provinces in China. The major findings are as follows: (i) agriculture, food & tobacco and construction have high virtual waterVirtual water consumption, collectively accounting for 70% of the total virtual waterVirtual water consumption; in terms of physical water consumption, agriculture, construction and metal smelting are major contributors, together representing 94% of the total physical water consumption; (ii) food & tobacco, textiles and gas supply exhibit higher virtual waterVirtual water consumption compared to physical water consumption, which indicates that their supply chains contain large amounts of virtual waterVirtual water; (iii) compared to LSTM, the CNN-LSTM model demonstrates superior performance in predicting water consumption in Ningxia, achieving an R2 value of over 0.9.