Search

Groundwater quality is typically measured through water sampling and lab analysis. The field-based measurements are costly and time-consuming when applied over a large domain. In this study, we developed a machine learning-based framework to map groundwater quality in an unconfined aquifer in the north of Iran. Groundwater samples were provided from 248 monitoring wells across the region. The groundwater quality index (GWQI) in each well was measured and classified into four classes: very poor, poor, good, and excellent, according to their cut-off values. Factors affecting groundwater quality, including distance to industrial centers, distance to residential areas, population density, aquifer transmissivity, precipitation, evaporation, geology, and elevation, were identified and prepared in the GIS environment. Six machine learning classifiers, including extreme gradient boosting (XGB), random forest (RF), support vector machine (SVM), artificial neural networks (ANN), k-nearest neighbor (KNN), and Gaussian classifier model (GCM), were used to establish relationships between GWQI and its controlling factors. The algorithms were evaluated using the receiver operating characteristic curve (ROC) and statistical efficiencies (overall accuracy, precision, recall, and F-1 score). Accuracy assessment showed that ML algorithms provided high accuracy in predicting groundwater quality. However, RF was selected as the optimum model given its higher accuracy (overall accuracy, precision, and recall = 0.92; ROC = 0.95). The trained RF model was used to map GWQI classes across the entire region. Results showed that the poor GWQI class is dominant in the study area (covering 66% of the study area), followed by good (19% of the area), very poor (14% of the area), and excellent (< 1% of the area) classes. An area of very poor GWQI was observed in the north. Feature analysis indicated that the distance to industrial locations is the main factor affecting groundwater quality in the region. The study provides a cost-effective methodology in groundwater quality modeling that can be duplicated in other regions with similar hydrological and geological settings., (© 2023. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.)

Groundwater drawdown is typically measured using pumping tests and field experiments; however, the traditional methods are time-consuming and costly when applied to extensive areas. In this research, a methodology is introduced based on artificial neural network (ANN)s and field measurements in an alluvial aquifer in the north of Iran. First, the annual drawdown as the output of the ANN models in 250 piezometric wells was measured, and the data were divided into three categories of training data, cross-validation data, and test data. Then, the effective factors in groundwater drawdown including groundwater depth, annual precipitation, annual evaporation, the transmissivity of the aquifer formation, elevation, distance from the sea, distance from water sources (recharge), population density, and groundwater extraction in the influence radius of each well (1000 m) were identified and used as the inputs of the ANN models. Several ANN methods were evaluated, and the predictions were compared with the observations. Results show that the modular neural network (MNN) showed the highest performance in modeling groundwater drawdown ​​(Training R-sqr = 0.96, test R-sqr = 0.81). The optimum network was fitted to available input data to map the annual drawdown ​​across the entire aquifer. The accuracy assessment of the final map yielded favorable results (R-sqr = 0.8). The adopted methodology can be applied for the prediction of groundwater drawdown in the study site and similar settings elsewhere., (© 2022. The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.)

Understanding annual extraction volumes and fluctuations in groundwater depth is essential in water resource management. This study applied various machine learning methods to predict the spatial variability of groundwater depletion (GWD) in an alluvial aquifer located along the southern coasts of the Caspian Sea (Mazandaran Plain). Initially, mean GWD values were measured from 250 piezometric wells across the plain. Subsequently, the factors controlling GWD were identified and provided. These factors encompass natural elements, including precipitation, evaporation, topography, groundwater depth, aquifer transmissivity, and proximity to water bodies. They also include anthropogenic factors such as extraction rates and proximity to industrial centres. Three machine learning models – Random Forest (RF), Deep Learning (DL), and Extreme Gradient Boosting (EGB) – were employed to model GWD using consistent training and test data. In this context, GWD was used as the output, and its controlling factors were incorporated as input variables. Correlation coefficients between GWD and its controlling factors revealed that the transmissivity of aquifer formations (R = 0.81), groundwater withdrawal within well radii (R = 0.69), and groundwater depth (R = 0.6) were the most significant factors. All three models demonstrated high efficacy during the training and testing stages. However, during testing, EGB offered the highest performance in modelling GWD due to extraction (R2 = 0.8, NSE = 0.82). The verified model was then used to predict annual drawdown or extraction across the plain. The predicted results were visually represented as an annual GWD or extraction map within a GIS framework. A comparison between observed GWD values in piezometric wells and the predicted values confirmed the robust performance of the proposed methodology (R2 = 0.8). This methodology can be employed effectively to predict spatial changes in GWD or annual extraction in other plains. [ABSTRACT FROM AUTHOR]

Dendrohydrogeology can be used to simulate historical groundwater depth, water table drawdown, groundwater recharge and piezometric lines. We simulated paleohydrogeologic conditions via tree-rings and vessel chronologies using an artificial neural network (ANN) in the alluvial aquifer of the Caspian southern coast of Iran during the past century. Tree-ring width, vessel features, secondary piezometric well data, and precipitation from different sites within the study area were evaluated. After cross-dating, standardization and time series analysis, the relationships between tree-rings and vessel chronologies with groundwater depth were defined and simulated. Additionally, paleohydrogeologic records during the past century were simulated. The results generally demonstrate that tree-ring width is a better index than vessel features. However, we obtained the most exact groundwater depth modeling results by using the combination of tree-rings and earlywood vessel diameter from periods of low precipitation and groundwater fluctuations and significant temperature fluctuations. We also found that dendrohydrogeology has more applicability in groundwater modeling in areas where groundwater depth fluctuates 10–20 m below ground surface (based on root depth and water access). Moreover, using the simulated groundwater depths, piezometric lines in 1927 and 2000 (the years with maximum natural recharge and maximum drawdown respectively) were extracted using an interpolation technique and Geographic Information System (GIS). Finally, we suggest applying dendrohydrogeology for paleohydrogeologic modeling in alluvial aquifers. [ABSTRACT FROM AUTHOR]