Scholarlink

<p>Precise water quality forecasting is crucial for sustainable resource management, especially in areas facing rising salinity and hydrological strain. This research utilizes the Convolutional-LSTM with Attention model to forecast essential water quality parameters(Cl<sup>−</sup>, Mg<sup>2+</sup>, SO<sub>4</sub><sup>2−</sup> HCO<sub>3</sub><sup>−</sup>, and EC)and hydro-meteorological parameters (rainfall and water volume) in the Hammam Boughrara dam, Northwest Algeria. The model demonstrated remarkable predictive accuracy, attaining R² values surpassing 0.98 for all parameters. The RMSE values for Cl<sup>−</sup> (0.09&#xa0;mg/L), Mg<sup>2+</sup> (0.08&#xa0;mg/L), SO<sub>4</sub><sup>2−</sup> (0.12&#xa0;mg/L), EC (0.131 µS/cm), and HCO<sub>3</sub><sup>−</sup> (0.123&#xa0;mg/L) indicate its dependability in reflecting hydrochemical fluctuations. Furthermore, the estimates for rainfall and water volumedam demonstrated highperformance, exhibiting RMSE values of 0.087&#xa0;mm and 0.108 hm³, respectively. For the period 2023–2043, forecasts indicate a 12–18% decline in annual rainfall, a reduction in reservoir storage from 160 to below 25 hm³ (~ 85%), EC levels approaching 2,500 µS cm⁻¹, Cl⁻ stabilizing near 600&#xa0;mg L⁻¹, HCO₃⁻ rising to ~ 500&#xa0;mg L⁻¹, and SO₄²⁻ decreasing below 50&#xa0;mg L⁻¹, highlighting the model’s efficacy in predicting long-term hydrological trends. Bias correction was applied to temperature projections to enhance predictive accuracy. Trend analysis employing the modifiedMann-Kendall test and Sen’s slope estimator indicated statistically significant alterations in water quality parameters influenced by climate change and hydrological variability (<i>p</i> &lt; 0.01). The findings demonstrate a prolonged warming trend, exacerbating evaporation and hastening ion accumulation. Furthermore, the rise in EC and ionic concentrations along with diminishing water amounts underscore increasing aridification, diminished dilution capacity, and worsening hydrochemical imbalances. Finally, this proposed model outperforms conventional approaches and demonstrates its ability to effectively capture hydrological patterns, anticipate water quality deterioration, and support decision-making for sustainable water resource management.</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

Assessing Aridification’s Impact on Water Quality and Hydroclimatic Dynamics in Hammam Boughrara Reservoir Using a Hybrid Convolutional-LSTM-Attention Model

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • Mohammed Rais,
  • Radia Gherissi,
  • Abdelmadjid Boufekane

摘要

Precise water quality forecasting is crucial for sustainable resource management, especially in areas facing rising salinity and hydrological strain. This research utilizes the Convolutional-LSTM with Attention model to forecast essential water quality parameters(Cl−, Mg2+, SO42− HCO3−, and EC)and hydro-meteorological parameters (rainfall and water volume) in the Hammam Boughrara dam, Northwest Algeria. The model demonstrated remarkable predictive accuracy, attaining R² values surpassing 0.98 for all parameters. The RMSE values for Cl− (0.09 mg/L), Mg2+ (0.08 mg/L), SO42− (0.12 mg/L), EC (0.131 µS/cm), and HCO3− (0.123 mg/L) indicate its dependability in reflecting hydrochemical fluctuations. Furthermore, the estimates for rainfall and water volumedam demonstrated highperformance, exhibiting RMSE values of 0.087 mm and 0.108 hm³, respectively. For the period 2023–2043, forecasts indicate a 12–18% decline in annual rainfall, a reduction in reservoir storage from 160 to below 25 hm³ (~ 85%), EC levels approaching 2,500 µS cm⁻¹, Cl⁻ stabilizing near 600 mg L⁻¹, HCO₃⁻ rising to ~ 500 mg L⁻¹, and SO₄²⁻ decreasing below 50 mg L⁻¹, highlighting the model’s efficacy in predicting long-term hydrological trends. Bias correction was applied to temperature projections to enhance predictive accuracy. Trend analysis employing the modifiedMann-Kendall test and Sen’s slope estimator indicated statistically significant alterations in water quality parameters influenced by climate change and hydrological variability (p < 0.01). The findings demonstrate a prolonged warming trend, exacerbating evaporation and hastening ion accumulation. Furthermore, the rise in EC and ionic concentrations along with diminishing water amounts underscore increasing aridification, diminished dilution capacity, and worsening hydrochemical imbalances. Finally, this proposed model outperforms conventional approaches and demonstrates its ability to effectively capture hydrological patterns, anticipate water quality deterioration, and support decision-making for sustainable water resource management.