Mamun Md Abdullah-Al, Islam Abu Reza Md Towfiqul, Aktar Mst Nazneen, Uddin Md Nashir, Islam Md Saiful, Pal Subodh Chandra, Islam Aznarul, Bari A B M Mainul, Idris Abubakr M, Senapathi Venkatramanan
Department of Data Science, Tampere University, Finland.
Department of Disaster Management, Begum Rokeya University, Rangpur 5400, Bangladesh; Department of Development Studies, Daffodil International University, Dhaka 1216, Bangladesh.
Sci Total Environ. 2024 Nov 25;953:176024. doi: 10.1016/j.scitotenv.2024.176024. Epub 2024 Sep 4.
The groundwater (GW) resource plays a central role in securing water supply in the coastal region of Bangladesh and therefore the future sustainability of this valuable resource is crucial for the area. However, there is limited research on the driving factors and prediction of phosphate concentration in groundwater. In this work, geostatistical modeling, self-organizing maps (SOM) and data-driven algorithms were combined to determine the driving factors and predict GW phosphate content in coastal multi-aquifers in southern Bangladesh. The SOM analysis identified three distinct spatial patterns: KNapH, CaMgNO₃, and HCO₃SO₄POF. Four data-driven algorithms, including CatBoost, Gradient Boosting Machine (GBM), Long Short-Term Memory (LSTM), and Support Vector Regression (SVR) were used to predict phosphate concentration in GW using 380 samples and 15 prediction parameters. Forecasting accuracy was evaluated using RMSE, R, RAE, CC, and MAE. Phosphate dissolution and saltwater intrusion, along with phosphorus fertilizers, increase PO content in GW. Using input parameters selected by multicollinearity and SOM, the CatBoost model showed exceptional performance in both training (RMSE = 0.002, MAE = 0.001, R = 0.999, RAE = 0.057, CC = 1.00) and testing (RMSE = 0.001, MAE = 0.002, R = 0.989, RAE = 0.057, CC = 0.998). Na, K, and Mg significantly influenced prediction accuracy. The uncertainty study revealed a low standard error for the CatBoost model, indicating robustness and consistency. Semi-variogram models confirmed that the most influential attributes showed weak dependence, suggesting that agricultural runoff increases the heterogeneity of PO distribution in GW. These findings are crucial for developing conservation and strategic plans for sustainable utilization of coastal GW resources.
地下水资源在保障孟加拉国沿海地区的供水方面发挥着核心作用，因此，这一宝贵资源的未来可持续性对该地区至关重要。然而，关于地下水中磷酸盐浓度的驱动因素和预测的研究有限。在这项工作中，结合了地质统计建模、自组织映射（SOM）和数据驱动算法，以确定驱动因素并预测孟加拉国南部沿海多含水层中的地下水磷酸盐含量。SOM分析确定了三种不同的空间模式：KNapH、CaMgNO₃和HCO₃SO₄POF。使用包括CatBoost、梯度提升机（GBM）、长短期记忆（LSTM）和支持向量回归（SVR）在内的四种数据驱动算法，利用380个样本和15个预测参数来预测地下水中的磷酸盐浓度。使用均方根误差（RMSE）、相关系数（R）、相对绝对误差（RAE）、一致性相关系数（CC）和平均绝对误差（MAE）来评估预测准确性。磷酸盐溶解、海水入侵以及磷肥会增加地下水中的磷含量。使用通过多重共线性和SOM选择的输入参数，CatBoost模型在训练（RMSE = 0.002，MAE = 0.001，R = 0.999，RAE = 0.057，CC = 1.00）和测试（RMSE = 0.001，MAE = 0.002，R = 0.989，RAE = 0.057，CC = 0.998）中均表现出卓越的性能。钠、钾和镁对预测准确性有显著影响。不确定性研究表明，CatBoost模型的标准误差较低，表明其具有稳健性和一致性。半变异函数模型证实，最具影响力的属性显示出较弱的依赖性，这表明农业径流增加了地下水中磷分布的异质性。这些发现对于制定沿海地下水资源可持续利用的保护和战略计划至关重要。
