ABSTRACT
The management of drinking water potability within crystalline basement regions represents a critical public health challenge in Madagascar. This study, conducted in the Isandra District (Haute Matsiatra Region) between 2022 and 2025, evaluates the quality of water resources through a hybrid framework integrating statistical characterization and artificial intelligence. Physicochemical and microbiological assessments of various sampling points reveal alarming disparities: paroxysmal turbidity events (77.2 NTU), ferric loads reaching 4.40 mg/L in Andrainjato, and atypical alkalinity in Mahasoabe (pH 10.20), significantly exceeding international regulatory thresholds. The originality of this research lies in the implementation of a Mamdani-type Fuzzy Inference System (FIS) to calculate the Fuzzy Water Quality Index (FWQI). While conventional statistical methods are constrained by binary compliance assessments, fuzzy logic facilitates a holistic risk interpretation by effectively incorporating field uncertainties and non-linear data behavior. The results demonstrate a critical degradation of sanitary safety, with FWQI scores plummeting to 12.3/100 for the most exposed sites. Furthermore, the study identifies a robust correlation (\rho = 0.84) between turbidity and metallic concentrations, confirming the substantial impact of surface runoff within the fractured basement environment. Policy recommendations emphasize the urgent need for immediate source protection zones and the adoption of the FWQI as a strategic decision-support tool for regional water authorities. In conclusion, the hybridization of artificial intelligence and classical environmental engineering provides a resilient governance framework to ensure sustainable access to safe drinking water in Malagasy rural settings.
Keywords : Isandra, crystalline basement, fuzzy logic, FWQI, sanitary safety.