We compiled surface (0m-12m depth) marine N2O concentrations and partial pressures measurements from a variety of sources. The core of the data is sourced from the MEMENTO database (Kock\u00a0& Bange, 2015). We complement MEMENTO with additional published N2O measurements from the literature, and unpublished N2O measurements from 16 additional cruises (see Supplemental File \"SuppCruiseTable_dec11.xlsx\"), including 11 cruises from the Global Ocean Ship-Based Hydrographic Investigations Program (GO-SHIP). We do not perform any further quality control of the N2O data from published sources besides that performed by the individual contributors and the MEMENTO database administrators (Kock\u00a0& Bange, 2015). A description of the quality control performed on new unpublished N2O data is reported as footnotes to the annotations labeled as qc1**, qc2**, or qc3** (see \"SuppCruiseTable_dec11.xlsx\"). We convert each marine N2O measurement to XwN2O (the N2O mixing ratio in seawater, in units of ppb) using, when needed, the N2O solubility coefficient (Weiss & Price, 1980). The coefficient is calculated using co-measured temperature, and salinity, as well as sea level pressure from the ERA5 reanalysis (Copernicus Climate Change Service, 2017), at the time (month and year), and location of the measurement. If the measurement time is not available in the ERA5 reanalysis prediction, we instead use the climatological atmospheric pressure at sea level, calculated from the monthly predictions for the years from 1979 through 2018. We then calculated N2O disequilibrium as DN2O = XwN2O \u2212 XaN2O, where XaN2O is the atmospheric N2O mixing ratio estimated by linear interpolation of NOAA\u2019s flask measurement dataset (Hall\u00a0et al., 2007) at the time and latitude of each marine N2O measurement.;
\nTo convert sparse observations to a global climatology, we trained 100 ensembles of regressions trees (Random Forests) to predict DN2O based on its relationship to well-sampled physical and biogeochemical predictors. We note that, while the prediction of N2O disequilibrium is done in mixing ratio units (ppb), the results are reported in the more commonly used pressure units (natm): pN2O = XN2O . P, where P is the climatological atmospheric pressure at sea level in atm, predicted by ERA5, included as part of the relevant data file for easy conversion. (see Data File: dn2o-mapped-Yang2020.nc).\u00a0
\nWe calculate the N2O air-sea flux using two wind-speed dependent parameterizations: an updated version of a commonly-used quadratic formulation (Wanninkhof, 1992; Wanninkhof, 2014) and a recent formulation that explicitly accounts for the effect of bubble-mediated fluxes (Liang et al., 2013). We apply each parameterization to two high-resolution wind products (Copernicus Climate Change Service, 2017;\u00a0Wentz et al., 2015), yielding four permutations of the piston velocity. In total, we obtain an ensemble of 400 global N2O air-sea flux estimates, from which we calculate a mean and uncertainty range (see Data File: n2oFlux-Yang2020.nc).
\nSampling and analytical prodcedures:
\nThe data is compiled from multiple sources, published and unpublished.\u00a0 Refer to the associated Supplemental File \"SuppCruiseTable_dec11.xlsx\"\u00a0for a detailed description of sampling and analytical methods associated with new data and references associated with published data. \"qc\" refers to \"qualtiy control and methods\"; see related references and descriptions in the Supplemental File.
GOSHIP (qc1): N2O was measured using shipboard gas chromatography-electron capture detection (GC-ECD) using analytical techniques modified from those described in Bullister and Wisegarver (2008). N2O was purged from 200 mL seawater samples using N2 carrier gas and trapped onto a trap that included MS5A held at -60\u00b0C. The trap was subsequently heated to 175\u00b0C to release N2O, which was further separated and purified via two precolumns before being quantified using electron capture detection. (The carrier gas for the N2O analyses was a 95%Ar/5% CH4 mix) The analytical system was calibrated frequently using internal standards of known N2O compositions or standards from Working Group no. 143 of the Scientific Committee on Oceanic Research (SCOR) (Wilson et al. 2018). Concentrations of N2O in seawater samples and gas standards are reported relative to the SIO98 calibration scale.
\nSPOT (qc3): Dissolved N2O concentrations were measured using a headspace equilibration method modified from Laperriere et al. (2019). A 30-mL ultra-high purity N2 headspace was introduced into 160 mL seawater samples using a 30-mL syringe with a second empty 30-mL syringe inserted into the septum to collect displaced sample water. Each headspace was overpressured with 10 mL of ultra high purity N2 to minimize atmospheric contamination. Samples were analyzed on an SRI 8610 Greenhouse Gas Monitoring Gas Chromatograph (GC) equipped with an electron capture detector (ECD), dual HayeSep D packed columns, and a 1-mL sample loop (SRI Instruments, Torrance, California, USA). Ultra-high purity N2 gas was used as the carrier with the sample loop kept at 60 \u00b0C and the column oven kept at 100 \u00b0C. Two certified standards, 0.1 ppm and 1 ppm N2O (Matheson Tri-Gas) were used for daily calibration using a linear calibration scheme.
\nOthers (qc2): N2O concentrations were measured with a GV IsoPrime Continuous Flow Isotope-Ratio Mass Spectrometer (CF-IRMS) as described in Bourbonnais et al. (2017). Briefly, seawater was pumped from sample bottles and completely extracted using a gas-extractor continuously sparged with He. N2O was then concentrated and purified in a purge-trap system. CO2 and H2O were removed with chemical and cryogenic traps. N2O was cryofocused with liquid N2 traps and passed through a gas chromatography (GC) column before IRMS analysis. N2O concentrations were calculated from relative peak heights between the samples and seawater standards of known N2O concentration equilibrated with the atmosphere at 5C and 20C. Equilibrium surface N2O concentrations were calculated based on the global mean atmospheric N2O dry mole fraction at the time of the cruise. The data were inter calibrated with samples also measured using purge-trap gas extraction systems coupled with either a GC-Electron Capture Detector (ECD) or a GC-quadrupole mass spectrometer (Fenwick et al., 2017) when available (e.g., P18 GO-SHIP, ArcticNet 2017 expedition) and yielded comparable N2O concentrations (generally less than 5% difference).
\nThe data is compiled from multiple sources, published and unpublished. Refer to the Supplemental File, \"SuppCruiseTable_dec11.xlsx\", for a detailed description of sampling and analytical methods associated with new data and references associated with published data.
Global reconstruction of surface oceanic N2O disequilibrium and its associated flux. The dataset includes 3 files:
\n(1) surfocean-n2o-compilation.csv: available as a .csv/.tsv\u00a0file via the \"Get Data\" button on the BCO-DMO landing page. A global compilation of observed surface ocean nitrous oxide pressure, concentration, mixing ratio measurements and their implied nitrous oxide disequilibrium used to train the supervised learning algorithm. Also attached as a .mat file (surfocean-n2o-compilation.mat) under \"Data Files\".
\n(2) dn2o-mapped-Yang2020.nc: available under \"Data Files\". Predicted N2O disequilibrium.
\n(3) n2oFlux-Yang2020.nc: available under \"Data Files\". The predicted ocean to atmosphere N2O flux.
This dataset contains a compilation of data from multiple sources. A list of all datasets and the associated information, including cruise name, is included in the associated Supplemental File, \"SuppCruiseTable_dec11.xlsx\".
Data Processing:
\nThe N2O disequilibrium at the location of each measurement is estimated following: _DXN2O = XN2O_ocean \u2212 XN2O_atm. Here, XN2O_ocean is the ocean-side N2O measurement converted to mixing ratios, and XN2O_ atm is the atmospheric N2O mixing ratio estimated by interpolating the National Oceanic and Atmospheric Administration atmospheric N2O flask dataset at the latitude and time of the oceanic measurements.
BCO-DMO Processing:
\nadded \"date\" column as YYYY-MM-DD.