We clarified the spatio-temporal variations of the remarkably degraded water chemistry of the Miyajimanuma lake (a Ramsar site in Japan) and the contributing factors. The spatio-temporal data were obtained using an autonomously operated boat equipped with a water quality sensor and an automatic water sampler installed in the lake. This monitoring was monthly performed according to the agricultural activities in the surrounding area. All water chemical data were evaluated using principal component analysis (PCA), which showed heterogeneity of the water chemistry due to the inflow of effluent water from paddy fields via drainage ditches. The PCA scores derived from chemical fertilizers and river water increased at the fringe of the inlet ditches during the puddling (May) and irrigation periods (July), respectively, corresponding to the rice paddy cultivation. In contrast, the score indicating dissolved organic carbon, which was derived from groundwater of peatland, increased along the shoreline during the non-irrigation period (August). The spatio-temporal monitoring revealed point and non-point sources. This information combined with a sampling technique based on an autonomously operated boat could be a powerful tool for future environmental monitoring.
