Measurement of oxygen concentrations and oxygen consumption rates using an optical oxygen sensor, and its application in hypoxia-related research in highly eutrophic coastal regions

Abstract To confirm the usefulness of optical oxygen sensors in hypoxia-related research in coastal regions, a contactless optical sensor (COS) (FireStingO2, Pyroscience) was used to measure dissolved oxygen concentrations, and the values obtained were compared with those obtained using the traditional Winkler method. During spring and summer seasons, bottom water samples from the Mikawa Bay, where hypoxia frequently occurs, were analyzed. The correlation between the oxygen concentrations obtained using the COS and Winkler methods had a slope of 1.01, an intercept of −6.17 μmol kg−1, and a coefficient of determination of 0.97. Further, when the oxygen consumption rates in the bottom water (OCRbw) were measured using these two methods, the slope, intercept, and coefficient of determination of the correlation between these values were 1.05, −0.08 μmol kg−1 d−1, and 0.63, respectively. These findings indicate that the COS method is suitable for obtaining accurate oxygen concentrations as well as OCRbw values for coastal waters, but it is still difficult to adapt it to small OCRbw values in the open ocean. The highest OCRbw (38.3 μmol kg−1 d−1) was observed at the inner part of the Bay in the end of August. Furthermore, the COS method also allowed the measurement of oxygen concentrations in the water overlaying the sediment core (OCRs + w), which in the Mikawa Bay were 3–5-fold higher than the corresponding OCRbw values. Observations made using the COS method in the Upper Gulf of Thailand (UGoT), which is also characterized by a highly hypoxic bottom layer, showed high OCRbw (>30 μmol kg−1 d−1) and OCRs + w (>40 μmol kg−1 d−1) near the estuarine of large rivers not in the rainy season but also in the dry season. The COS method has potential for use in the collection of OCR data that can significantly contribute to the accuracy of hypoxia prediction models.