Groundwater levels are under severe pressure globally due to over-extraction, pollution, and climate change necessitating continuous monitoring for sustainable aquifer management. This study introduces a novel ensemble machine learning (En) model that integrates shallow and deep machine learning (ML) models, optimized through the coronavirus herd immunity optimizer (CHIO), for accurate groundwater level forecasting. This En model was applied to the Ergene River Basin, Türkiye, a region facing severe groundwater depletion and contamination due to intensive agricultural and industrial activities. Groundwater level data spanning 1966 to 2023 on a weekly basis from four wells were used, split into 70% for training and 30% for testing under short- and long-term scenarios. Using the partial autocorrelation function and gamma test the best lag numbers were determined for input data, reflecting aquifer heterogeneity. Score analysis, supported by statistical metrics such as the coefficient of determination (R²) and root mean square error (RMSE), was employed alongside visual aids to assess the developed En model performance. Results demonstrated that deep ML models outperformed shallow ML models achieving R² ~ 0.99 and RMSE ~ 0.5 m. The developed En model outperformed all individual ML models, with score values exceeding 200, and its predictions closely aligned with measured water levels during both testing phases. The findings underscored the developed En model’s contribution to achieving sustainable development goals (SDGs) by enhancing water-use efficiency and addressing environmental, economic, and social sustainability challenges. The proposed approach offers a reliable and adaptable solution for groundwater level forecasting, applicable to other aquifers worldwide. Graphical Abstract

The Euphrates-Tigris basin in Turkey suffers from sedimentation and erosion problems. The purpose of this study was to estimate streamflow, sediment yield, and identification of soil erosion-prone areas for the Murat River Basin which is the headwater of Euphrates River, with Soil and Water Assessment Tool (SWAT). The model was calibrated and validated with observed streamflow and sediment yield data obtained from two gauging stations. The statistical model performance was evaluated by using NSE, R2, and PBIAS. NSE values were calculated as 0.57, 0.50 and 0.77, 0.75 for the stations E21A074 and E21A077, respectively, in the monthly streamflow calibration and validation periods. Monthly sediment yield generated the NSE values for stations E21A074 and E21A077 were 0.54, 0.69 and 0.62, 0.35 in the calibration and validation periods, respectively. Spatial analysis indicates that 3.9% of the basin is under very severe (>20 t/ha/y), and 21.3% is under severe (>10 t/ha/y) soil erosion conditions. Most of the soil erosion occurs at the barren land use and in the sub-watersheds with a slope greater than 25%. Overall, the SWAT model satisfactorily simulated the streamflow and sediment yield, and identified the erosion-prone areas at the sub-watershed scale at the Murat River Basin for water resource management.

The Yeşilırmak River Basin in northern Türkiye is crucial for the region’s water supply, agriculture, hydroelectric power generation, and clean drinking water. The primary goal of this study is to determine which modeling approach is most appropriate for various locations within the basin and how well meteorological data can predict river flow rates. Hydrological and meteorological forecasting both depend on the prediction of river flow rates. An artificial neural network (ANN), Univariate and Multivariate Long Short-Term Memory (LSTM) models have been utilized for streamflow forecasting. This research aims to determine the best model for several provinces in the basin area and give decision-makers a tool for reliable river flow rate estimates by combining LSTM and ANN models. According to research findings, the supervised multivariate LSTM model performed better than the unsupervised model in accuracy and precision. The sliding window methodology is suitable for estimating river flow based on meteorological datasets because it offers a primary method for reinterpreting time-series data in a supervised learning style. Compared to LSTM models, the ANN model that has been statistically optimized through experiments (DoE) design performs better in forecasting the river flow rate in the Yeşilırmak River basin (R² = 0.98, RMSE = 0.18). The study’s findings provided prospective cognitive models for the strategic management of water resources by forecasting future data from flow monitoring stations.