Author(s): V. Gomez-Escalonilla; F. Fussi; B. Lopez Luna; I. Martinez Garcia; J.J. Rahobisoa; P. Martinez-Santos
Linked Author(s):
Keywords: Groundwater; Machine learning; Quantity; Quality
Abstract: The present study proposes an innovative approach towards addressing the challenge of improving access to drinking water in southern Madagascar. The proposed methodology integrates information on groundwater quantity and quality into a single database, a novel approach in the existing literature. A database from southern Madagascar containing information on more than 2000 boreholes was analyzed to assess groundwater potential based on the potential yield of the boreholes but also considering quality in terms of electrical conductivity. Consequently, a range of topographical, climatic, geological and soil-related variables were incorporated into the study, with the aim of assessing their impact on the quality and potential for exploitation of groundwater resources. Machine learning algorithms were applied to search for correlations between these conditioning factors and the target variable, composed of both quality and quantity components. As a result, it was possible to identify areas that are both productive and suitable for consumption. Findings demonstrate the combined influence of different explanatory variables on groundwater availability and quality, as well as the potential of this methodological approach for field data-based planning, which could ultimately serve to improve sustainable access to water in arid and vulnerable regions.
Year: 2026