“…To evaluate the performance of our three numerical approaches, we compile field measurements of O 2 /Ar and O 2 from 14 cruises (Figure 1, Table S1 in Supporting Information S1): two cruises in the Southern Ocean, SO‐GasEx‐2008 (https://www.bco-dmo.org/dataset/4059; Hamme et al., 2012), and LTER‐2012 (https://zenodo.org/record/5786352; Eveleth et al., 2017), three cruises in the North Atlantic (Bermuda‐2017, Bermuda‐2016, and Bermuda‐2015, https://zenodo.org/record/5786352; Tang et al., 2019; Wang et al., 2018), six cruises in North Pacific, KM1906‐2016 (https://zenodo.org/record/4009653; Juranek et al., 2020), Oceanus‐2017, Line P‐2019‐001, Line P‐2018‐040 and La Perouse‐2017‐009 (https://doi.pangaea.de/10.1594/PANGAEA.933345; Izett et al., 2021; R. Izett & Tortell, 2020), California‐2016 (https://oceaninformatics.ucsd.edu/datazoo/catalogs/ccelter/datasets/290; Wang et al., 2020), and three Arctic cruises, OS191901‐2019 (https://arcticdata.io/catalog/view/doi%3A10.18739/A2HH6C69V), CCGS Amunden‐2019 (https://doi.pangaea.de/10.1594/PANGAEA.933345; Izett et al., 2021), and CCIN13242‐2018 (https://zenodo.org/record/6124142; Izett et al., 2021). The data near the ice‐covered regions from the LTER‐2012 cruise are not included because of difficulties in parameterizing the effects of ice dynamics on gas saturation (Cassar et al., 2021; Eveleth et al., 2014).…”