4.7 Article

Soft computing assessment of current and future groundwater resources under CMIP6 scenarios in northwestern Iran

期刊

AGRICULTURAL WATER MANAGEMENT
卷 285, 期 -, 页码 -

出版社

ELSEVIER
DOI: 10.1016/j.agwat.2023.108369

关键词

Aquifer; Climate change; Groundwater management; Machine learning; Groundwater level prediction

向作者/读者索取更多资源

The excessive use of water resources along with climate change is threatening to significantly decrease the groundwater level in arid and semiarid regions. A study was conducted to assess the impact of climate change on the groundwater level in the Dehgolan Aquifer in Iran. Soft computing techniques were applied to estimate the future groundwater level drop by 2045, using climate projections and machine learning models. The findings highlight the importance of implementing effective management strategies to prevent further depletion of the aquifer.
Excessive use of water resources in combination with climate change threaten to significantly reduce ground-water in arid and semiarid regions. We studied the effects of climate change on the groundwater level for the important Dehgolan Aquifer in northwestern Iran. The water level in this aquifer has dropped by about 35 m during the last 30 years. Soft computing techniques were used together with climate projections in three methodological steps to estimate the groundwater level drop by 2045. Firstly, MODFLOW was used to simulate groundwater flow and movement. Secondly, simulation results, support vector regression (SVR), and least -squares SVR (LSSVR) machine learning models were used to predict groundwater levels for the future 20-year period (2026-2045). The whale optimization algorithm (WOA) was used to improve the prediction results by optimizing the SVR parameters. Thirdly, three climate models of CMIP6 (ACCESS-CM2, BCC-CSM2-MR, and CMCC-ESM2) were used to predict the changes in precipitation for the future period (2026-2045) using SSP 2.6 and SSP 8.5 scenarios. The results showed that the MODFLOW-LSSVR model predicted the groundwater level more accurately than MODFLOW-SVR and MODFLOW-SVR-WOA. The calculation scenario containing previous month groundwater level, monthly aquifer withdrawal, and monthly precipitation had the highest performance in predicting groundwater level with root mean square error (RMSE), mean absolute percentage error (MAPE), and Nash Sutcliffe efficiency (NSE) equal to 0.305 m, 0.014 m, and 0.998, respectively. The results indicate that precipitation may decrease in the future period for the SSP 8.5 scenario (about 6% compared to the reference period 1987-2005). This decrease, along with the continuation of the current aquifer withdrawal, will cause a drop of about 36 m (during 28 years) of the groundwater level (1.3 m per year). The results reveal that the drop could be reduced to 12 m by adopting a 25% reduction in the current aquifer withdrawal. The findings show the necessity of providing a suitable management approach to prevent future aquifer exhaustion due to the continuation of the current withdrawal situation in the region.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.7
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据