A BPNN-based ecologically extended input–output model for virtual water metabolism network management of Kazakhstan

Zhenhao Ma,Jing Liu,Yongping Li,Hao Zhang,Licheng Fang

DOI: https://doi.org/10.1007/s11356-023-25280-6

IF: 5.8

2023-01-20

Environmental Science and Pollution Research

Abstract:AbstractIn this study, a back-propagation-neural-network-based ecologically extended input–output model (abbreviated as BPNN-EIOM) is developed for virtual water metabolism network (VWMN) management. BPNN-EIOM can identify key consumption sectors, simulate performance of VWMN, and predict water consumption. BPNN-EIOM is then applied to analyzing VWMN of Kazakhstan, where multiple scenarios under different gross domestic production (GDP) growth rates, sectoral added values, and final demands are designed for determining the optimal management strategies. The major findings are (i) Kazakhstan typically relies on net virtual water import (reaching 1497.9 × 106 m3 in 2015); (ii) agriculture is the major exporter and advanced manufacture is the major importer; (iii) by 2025, Kazakhstan’s water consumption would increase to [19322, 22016] × 106 m3 under multiple scenarios; (iv) when Kazakhstan’s GDP growth rate, manufacturing’s added value, and final demand are scheduled to 5.5%, 8.5%, and 5.8%, its VWMN can reach the optimum. The findings are useful for decision makers to optimize Kazakhstan’s industrial structure, mitigate the national water scarcity, and promote its socio-economic sustainable development.

environmental sciences