DÉPLOIEMENT PILOTE DES MINISTATIONS SAIGA POUR LA SURVEILLANCE DE LA QUALITÉ DE L’AIR À KINSHASA: CALIBRATION STATISTIQUE ET ÉVALUATION LOCALE INITIALE DES PM₂.₅

This study presents a pilot deployment of SAIGA low-cost sensor systems for air-quality surveillance in Kinshasa, Democratic Republic of the Congo, focusing on statistical calibration and the initial local evaluation of PM₂.₅. Three units were first compared in Belgium in 2025 with Brussels Environment data provided through IRCELINE. For PM₂.₅, NO₂ and O₃, the 41R012-Uccle station was located about 115 m from the SAIGA units; this represents a near-site comparison rather than strict metrological co-location. Simple and multiple linear regressions, including temperature and relative humidity, were fitted on a training period and evaluated on an independent temporal test set. For PM₂.₅, R² values ranged from 0.66 to 0.74 and RMSE from 1.58 to 1.67 µg m⁻³. In Kinshasa, evaluation against MODULAIR was limited to January 2026 because of a subsequent data-logging malfunction. Direct transfer of the Belgian model showed a bias explored through a local offset of 29.70 µg m⁻³, yielding R² ≈ 0.67 and RMSE ≈ 14.09 µg m⁻³. Because the offset was estimated on the same dataset, it requires independent confirmation. Gaseous-pollutant results remained more limited.