Preprocessing approaches in machine-learning-based groundwater potential mapping: an application to the Koulikoro and Bamako regions, Mali

Abstract

Abstract. Groundwater is crucial for domestic supplies in the Sahel, where the strategic importance of aquifers will increase in the coming years due to climate change. Groundwater potential mapping is a valuable tool to underpin water management in the region and, hence, to improve drinking water access. This paper presents a machine learning method to map groundwater potential. This is illustrated through its application in two administrative regions of Mali. A set of explanatory variables for the presence of groundwater is developed first. Scaling methods (standardization, normalization, maximum absolute value and max–min scaling) are used to avoid the pitfalls associated with reclassification. Noisy, collinear and counterproductive variables are identified and excluded from the input dataset. A total of 20 machine learning classifiers are then trained and tested on a large borehole database (n=3345) in order to find meaningful correlations between the presence or absence of groundwater and the explanatory variables. Maximum absolute value and standardization proved the most efficient scaling techniques, while tree-based algorithms (accuracy >0.85) consistently outperformed other classifiers. The borehole flow rate data were then used to calibrate the results beyond standard machine learning metrics, thereby adding robustness to the predictions. The southern part of the study area presents the better groundwater prospect, which is consistent with the geological and climatic setting. Outcomes lead to three major conclusions: (1) picking the best performers out of a large number of machine learning classifiers is recommended as a good methodological practice, (2) standard machine learning metrics should be complemented with additional hydrogeological indicators whenever possible and (3) variable scaling contributes to minimize expert bias.

Oversampling methods for machine learning model data training to improve model capabilities to predict the presence of Escherichia coli MG1655 in spinach wash water

“…This significant increase ( p < 0.05) in specificity performance is due to the compression of the data when outliers are removed and standardized, that is, the data are easier for the model to discriminate what data points to use as cutoffs in the tree nodes. Furthermore, the RF model does not take into consideration the magnitude of the data when assigning weights to that data (Gomez‐Escalonilla et al., 2022). For example, the range of pH between 7.05 and 9.06 is not considered important, compared to the water conductivity, which can have values between 7.9 and 196.2 µS or to the water temperature with values ranging from 15.8 to 26.0°C.…”

“…This result shows that with SVM and RF models, non‐preprocessed data perform better when using the physiochemical features to predict E. coli MG1655 in spinach wash water. The difference in performance between the preprocessed and non‐preprocessed data that trains SVM and RF models can be attributed to the compression of training data and the removal of outliers, which makes certain predictive features less important in the models (Gomez‐Escalonilla et al., 2022; Sriraam & Raghu, 2017).…”

Spatial Prediction of Groundwater Withdrawal Potential Using Shallow, Hybrid, and Deep Learning Algorithms in the Toudgha Oasis, Southeast Morocco

“…Recently, numerous novel methods and algorithms related to artificial intelligence (AI) based on machine learning (ML) and deep learning (DL) have been developed, assessed, and approved in the field of GWP mapping determination; this has been conducted with respect to inventories of water withdrawal points and geological, hydrogeological, hydrological, topographic and climatic factors [11,17,[19][20][21]. On this matter, the following models were commonly used and applied in the sub-cited studies: random forest (RF), support vector machine (SVM), linear regression (LR), decision tree (DT), naive Bayes (NB), convolutional neural network (CNN), long short-term memory (LSTM) and artificial neural network (ANN) [6,9,14,[22][23][24][25]. Furthermore, a variety of methods have also been proposed to improve the efficiency and precision of the prediction models, such as optimization algorithms and ensemble models [23,26,27].…”

Identification of Groundwater Potential Zones (GWPZ) Using Geospatial Techniques and AHP Method: a Case Study of the Boudinar Basin, Rif Belt (Morocco)

“…In Africa, where rapid population growth is occuring, water resources are among the lowest in the world and are expected to fall over 50% by 2050, according to the food and agriculture organization of the United Nations (FAO). Despite many studies having identified the potential groundwater in African countries: Ethiopia [4], Mali [5], Egypt [6], Morocco [7,8], Nigeria [9], the mapping and estimation of groundwater recharge are still required to avoid water scarcity.…”