This dataset contains methane and nitrous oxide dissolved gas concentration (https://doi.pangaea.de/10.1594/PANGAEA.987944), sea-air fluxes (this dataset), and ancillary hydrographic data from research cruises in western Long Island Sound occurring in August 2023, October 2023, and May 2024. Water samples for methane and nitrous oxide analysis were collected from Niskin bottles mounted on a CTD rosette. Water was collected into glass serum bottles and at least three volumes of water were allowed to flow through the sample before it was preserved and sealed with a butyl rubber stopper and aluminum crimp seal. Gas concentrations were determined using a headspace equilibration method and gas chromatograph with electron capture detector (for nitrous oxide) and flame ionization detector (for methane). Equilibrium dissolved concentrations were calculated from the measured temperature and salinity following the solubility equations of Wiesenburg and Guinasso (1979) for methane and Weiss and Price (1980) for nitrous oxide. Dry atmospheric concentrations were set to 2020 ppbv for methane and 338 ppbv for nitrous oxide and adjusted to 100% humidity. These concentrations are taken from preliminary surface flask results from Mashpee, Massachusetts, USA (station MSH) from the NOAA Carbon Cycle Greenhouse Gases group website (https://gml.noaa.gov/ccgg/) and represent the average over the three cruises (Dlugokencky et al., 2020b, a; Hall et al., 2007). Oxygen concentration was determined using an oxygen sensor mounted on the Niskin rosette, calibrated with discrete samples analyzed by Winkler titration (Langdon et al., 2010), which were also collected from the CTD rosette. The mixed layer depth was defined based on a potential density difference criterion of 0.125 kg/m³ relative to the shallowest measurement (either 2 or 3 m), using CTD profiles binned to 1 m. The 15-day weighted gas transfer velocities and fluxes are calculated using wind speeds measured over 15 days at 15-minute intervals prior to sampling using the data from NOAA buoy 44022 with weighting based on the fraction of the mixed layer ventilated during each time step following the method of Teeter et al. (2018) as described in the main manuscript of Manning et al. (2025) and software of Manning and Nicholson (2022).