Abstract: Study region this study focuses on Madagascar. This island is characterized by a great diversity of climate, due to trade winds and the varying topography. This country is also undergoing extreme rainfall events such as droughts and cyclones. Study focus the rain gauge network of Madagascar is limited (about 30 stations). Consequently, we consider relevant satellite-based precipitation datasets to fill gaps in ground-based datasets. We assessed the reliability of 21 satellite-based and reanalysis precipitation products (P-datasets) through a direct comparison with 24 rain gauge station measurements at the monthly time step, using four statistical indicators: Kling-Gupta Efficiency (KGE), Correlation Coefficient (CC), Root Mean Square Error (RMSE), and Bias. Based on this first analysis, we produced a merged dataset based on a weighted average of the 21 products. New hydrological insights for the region based on the KGE and the CC scores, WFDEI (WATCH Forcing Data methodology applied to ERA-Interim), CMORPH-BLD (Climate Prediction Center MORPHing satellite-gauge merged) and MSWEP (Multi-Source Weighted Ensemble Precipitation) are the most accurate for estimating rainfall at the national scale. Additionally, the results reveal a high discrepancy between bio-climatic regions. The merged dataset reveals higher performance than the other products in all situations. These results demonstrate the usefulness of a merging approach in an area with a deficit of rainfall data and a climatic and topographic diversity.

Abstract: In the present study, the ground water quality parameters were monitored during pre- and post-monsoon seasons across Narmada district, Gujarat, India. Monitoring was done in 89 drinking water samples collected by grid sampling method from the study area. Uranium and fluoride were analyzed along with associated parameters such as pH, dissolved oxygen, Cl−, NO3−, F−, SO42−, total alkalinity, total dissolved solids and hardness. In 4% samples the fluoride content was found to be above WHO permissible limits of 1.5 mg/L (2.36 mg/L in Undaimandava, 1.55 mg/L in Shira, 3.04 mg/L in Fatehpur and 1.83 mg/L in Dholivav) during pre-monsoon season (PRM) and 4.74 mg/L, 2.41 mg/L, 2.34 mg/L and 3.99 mg/L respectively in Bantawadi, Shira, Undai Mandava and Fatepur villages during post-monsoon (POM). The uranium level was within WHO limits in both POM and PRM seasons. The quality of the water was evaluated by Principal Component and Pearson Correlation statistical analysis techniques. The PRM and POM correlation study indicated a strong correlation of TDS with EC, Chloride, total alkalinity and bicarbonate and U while moderately strong correlation of TDS with fluoride were observed indicating that chloride, total alkalinity, bicarbonate, U and fluoride contributed to TDS and EC. Principal component analysis was applied for 14 variables, from which 3 factors were extracted during PRM and POM seasons. The extracted components, contributed 84.391% and 83.315%, to variation during PRM and POM seasons respectively. The study indicated that the analyzed water samples in Narmada district were safe for drinking purpose. However, Tilakwada tehsil groundwater was observed to be unsustainable for drinking, without further water treatment, but was appropriate for agricultural purposes. The study will help the residents of the district to understand the present water quality status and will also help in future management to protect the ground water of Narmada district.

Abstract: Predicting groundwater levels is pivotal in curbing overexploitation and ensuring effective water resource governance. However, groundwater level prediction is intricate, driven by dynamic nonlinear factors. To comprehend the dynamic interaction among these drivers, leveraging machine learning models can provide valuable insights. The drastic increase in computational capabilities has catalysed a substantial surge in the utilisation of machine learning-based solutions for effective groundwater management. The performance of these models highly depends on the selection of hyperparameters. The optimisation of hyperparameters is a complex process that often requires application-specific expertise for a skillful prediction. To mitigate the challenge posed by hyperparameter tuning’s problem-specific nature, we present an innovative approach by introducing the automated machine learning (AutoML-GWL) framework. This framework is specifically designed for precise groundwater level mapping. It seamlessly integrates the selection of best machine learning model and adeptly fine-tunes its hyperparameters by using Bayesian optimisation. We used long time series (1997-2018) data of precipitation, temperature, evaporation, soil type, relative humidity, and lag of groundwater level as input features to train the AutoML-GWL model while considering the influence of Land Use Land Cover (LULC) as a contextual factor. Among these input features, the lag of groundwater level emerged as the most relevant input feature. Once the model is trained, it performs well over the unseen data with a strong correlation of coefficient (R = 0.90), low root mean square error (RMSE = 1.22), and minimal bias = 0.23. Further, we compared the performance of the proposed AutoML-GWL with sixteen benchmark algorithms comprising baseline and novel algorithms. The AutoML-GWL outperforms all the benchmark algorithms. Furthermore, the proposed algorithm ranked first in Friedman’s statistical test, confirming its reliability. Moreover, we conducted a spatial distribution and uncertainty analysis for the proposed algorithm. The outcomes of this analysis affirmed that the AutoML-GWL can effectively manage data with spatial variations and demonstrates remarkable stability when faced with small uncertainties in the input parameters. This study holds significant promise in revolutionising groundwater management practices by establishing an automated framework for simulating groundwater levels for sustainable water resource management.