Eddy Covariance measurements of air-sea CO2 flux in Summer 2025 at Helgoland, and associated water-side measurements.

Between 1. June and 13. September 2025, eddy covariance (EC) and associated water-side measurements were conducted on the northern shore of Helgoland, a small island in the southern North Sea.Observations of air-sea CO₂ exchange were made with an EC system built around a closed-path infrared gas analyzer (Li-Cor 7200R/S) and sonic anemometer (Gill Windmaster), both of which responding at 10 Hz and mounted ca 4.5 m above mean water level. All wind speed measurements were scaled to a reference height of 10 m (U₁₀) assuming an exponential profile. Additional meteorological measurements from the station include net radiation (Kipp & Zonen NR Lite2), photosynthetically active radiation (Li-190R), and temperature and relative humidity.Water-side measurements were provided by an in-situ CO₂ sensor (4H-Jena, HydroC-CO2) and CTD (Sea and Sun Technology) installed at the MARGATE site operated by AWI (Walcher et al., 2013) approximately 800 m NW of the EC station, at an average water depth of 10m. Due to a period of biological fouling in late September to early October, spurious salinity values were trimmed by removing values with salinity less than 28, or values where salinity changed by more than 0.5 in one hour. To reduce the risk of biofouling-induced data loss, the CO₂ sensor was exchanged 3 times over the course of the deployment, and between exchanges, the sensors were also inspected and cleaned by divers from AWI at approximately 2-week intervals. The CO₂ sensor completed a self-zeroing routine once per day, and the data during this period were discarded. Air-side CO₂ concentrations were provided by the ICOS site located ca 800m west of the EC station (ICOS ID: HEL; Lat: 54.1804; Lon: 7.8833; measurement height: 110 m).Water currents were measured with a bottom-mounted upward-facing TRDI Workhorse 1200 kHz. The ADCP collected one ensemble of 50 pings every minute with 1 m bins, with coordinates transformed with ADCPs sensor data, following on-site compass calibration. Raw binary data were filtered with flags for currents exceeding reasonable local maximum values (1.2 m/s), bins with low amplitude (50 counts) and correlation (64 counts), high error velocities and bins beyond the surface side lobe interference range during post processing. All remaining ADCP current data were considered good.