Increasing the Low Oxygen Accuracy of Commercial Optodes

Abstract Using standard calibration schemes, commercial oxygen optode sensors typically yield oxygen concentrations in the range of 2–4 μ mol kg −1 under anoxic conditions. They are thus unable to detect the roughly 0.1 μ mol kg −1 levels of oceanic functional anoxia. Here, an empirically modified Stern–Volmer equation is used to characterize and calibrate 26 optodes deployed on 16 autonomous floats in the eastern tropical North Pacific oxygen-deficient zone (ODZ) using a combination of manufacturers’, laboratory, and in situ data. Laboratory calibrations lasting several months and conducted over 2 years show that optodes kept under anoxic conditions drift at rates of order ±0.2 μ mol kg −1 yr −1 , with much higher drifts in the first month. High accuracy, ship-based switchable trace oxygen (STOX) profiles show that the anoxic cores of the eastern tropical North Pacific (ETNP) ODZ have oxygen concentrations less than 0.05 μ mol kg −1 . An algorithm is developed to detect these cores in the float profiles and used to estimate and remove the optode drift. A section across the ETNP using these data shows the southward thinning of the functionally anoxic core from 600 to 100 m thick, demonstrating an order of magnitude improvement in the low oxygen performance of the optodes. Similar recalibration could be implemented on the existing database of Argo oxygen floats. A more detailed understanding and modeling of the drift rates facilitated by a combination of laboratory studies and routine STOX profiles could further improve the low-oxygen optode calibration. Significance Statement Subsurface regions of the ocean with very low oxygen concentrations are important in setting the overall ocean nutrient balance. They appear to be expanding and thus need to be monitored as the oceans change. This could be done using existing arrays of robotic floats, but the oxygen sensors used on these floats are not sufficiently sensitive. Here, we describe improvements in calibration for these sensors which can increase their sensitivity by about a factor of 10. This will allow robotic measurement of the extent and intensity of these low oxygen regions as they evolve in the future.