Scholarlink

<p>Water resources are vital to Africa’s sustainable development. Six machine learning models were employed to examine the socio-economic drivers and to forecast water use across ten major African countries from 1987 to 2052. Three scenarios, i.e., business as usual (BAU), general water-saving, and enhanced water-saving, have been developed to simulate water demand driven by demographic, socio-economic, and industrial factors. Artificial neural network (ANN) demonstrated high accuracy, with R² ranging from 0.816 to 0.973 for training and 0.866 to 0.967 for testing in the ten countries. The scenario analysis illustrates the potential influence of policy interventions on future water demand. Under the BAU scenario, total water consumption in the ten countries is projected to increase by approximately 120% from 2022 to 2050. However, the projected increase declines to about 80% under the general water-saving scenario and further decreases to around 21% under the enhanced water-saving. Clustering analysis revealed distinct patterns. Egypt, South Africa, and Nigeria initially belonged to the same cluster but diverged over time. South Africa’s economic growth reduced water use, whereas Nigeria’s population surge drove up water demand. Countries like Kenya, Ghana, Algeria, and Angola improved their water efficiency and economic performance, and could transfer to a lower demand cluster with water-saving strategies. These findings underline the transformative impact of potential water-saving measures and data-driven strategies in achieving sustainable water use in Africa.</p>

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

Application of machine learning to water use forecasting and water-saving scenarios in African countries

  • 期刊论文

  • 发表日期：

    2026年5月15日

查看全文
  • Edwin Kipkirui,
  • Jianfu Zhao,
  • Yi Xu,
  • Tao Wang

摘要

Water resources are vital to Africa’s sustainable development. Six machine learning models were employed to examine the socio-economic drivers and to forecast water use across ten major African countries from 1987 to 2052. Three scenarios, i.e., business as usual (BAU), general water-saving, and enhanced water-saving, have been developed to simulate water demand driven by demographic, socio-economic, and industrial factors. Artificial neural network (ANN) demonstrated high accuracy, with R² ranging from 0.816 to 0.973 for training and 0.866 to 0.967 for testing in the ten countries. The scenario analysis illustrates the potential influence of policy interventions on future water demand. Under the BAU scenario, total water consumption in the ten countries is projected to increase by approximately 120% from 2022 to 2050. However, the projected increase declines to about 80% under the general water-saving scenario and further decreases to around 21% under the enhanced water-saving. Clustering analysis revealed distinct patterns. Egypt, South Africa, and Nigeria initially belonged to the same cluster but diverged over time. South Africa’s economic growth reduced water use, whereas Nigeria’s population surge drove up water demand. Countries like Kenya, Ghana, Algeria, and Angola improved their water efficiency and economic performance, and could transfer to a lower demand cluster with water-saving strategies. These findings underline the transformative impact of potential water-saving measures and data-driven strategies in achieving sustainable water use in Africa.