Evaluating the use of tracer methods to infer changes in ventilation variability

Repeat occupations now allow for decadal changes to be inferred from a variety of tracer methods including apparent oxygen utilization (AOU) and mean ages from 1d transit time distributions (TTD) constrained by CFC-11, CFC-12, and SF₆. However, these sometimes yield conflicting results and it is unclear how much of the inferred changes are due to actual variability in ventilation or to biases in method. Here the skill of AOU and 1d TTD methods to determine circulation variability is assessed using ideal age, oxygen, CFC-12, and SF₆ from an offline transport model. We find that in most of the upper thermocline of the subtropical Indo-Pacific, AOU strongly correlates with ideal age variability (p<0.1 and R²>0.5) in thermocline waters over a wide range of ages (up to 200 years). Mean age from 1d TTDs (using either single or dual tracers) only has skill in tracking variability of waters which are less than about 60 years old and for water masses above each basin’s intermediate waters. By comparing the tracer-constrained TTDs with the model-calculated TTD, we find that these biases likely arise from the inability of the 1d TTD to capture complex transport pathways. In particular, most intermediate water TTDs are a mixture between a young, rapidly ventilated pathway which is well-constrained by CFC-12 and SF₆ but also significantly older pathways arising from slow diffusion from sub-thermocline waters. Based on these results, changes in ventilation can be diagnosed with high confidence when AOU and tracer-inferred TTDs agree in sign, but in regions where they disagree, AOU is often the more reliable indicator.