Background & Summary

Coastal systems act as dynamic and vital reactors modulating the flux of water and elements between land and sea1,2. The biogeochemical functioning of the coastal ocean is supported by a balance of inputs from rivers, submarine groundwater discharge (SGD), sediments, and the atmosphere as well as outputs from water mixing with other water bodies and burial in sediments1,3,4. While these inputs and outputs maintain natural biogeochemical functioning, increased land-based and atmospheric chemical loads deliver excess nutrients, carbon, metals, and greenhouse gases to the coast1,5,6. SGD, for instance, frequently carries higher concentrations of nutrients and dissolved inorganic carbon (DIC) to the coast compared to rivers3, potentially driving eutrophication and coastal acidification7,8. Understanding the sources and transport pathways for water and dissolved chemical species and carbon species is critical for addressing pervasive water quality problems, eutrophication, and hypoxia in aquatic systems9.

The Baltic Sea is a large semi-enclosed marginal sea characterized by strong physiochemical gradients, particularly in salinity, dissolved oxygen, and associated element loads10. A restricted water exchange with the North Sea and substantial freshwater inputs result in a practical salinity gradient ranging from ~2 in the northernmost waters of the Bothnian Bay to 34 in the Skagerrak Sea10,11, creating estuarine conditions over ~2000 km. The limited water exchange also leads to long water residence times (~20–30 years in the surface layer)12,13. Water exchange in the deeper parts of the Baltic Sea is not only restricted laterally, but especially vertically due to a strong permanent halocline ~40–80 m below the surface that maintains hypoxic/anoxic conditions in large bottom areas14. The specific hydrography, deep basins supporting the development of euxinic conditions, and substantial anthropogenic impact has resulted in prolonged water quality issues such as eutrophication and hypoxia leading to significant changes in biogeochemical cycling14,15,16.

The Baltic Sea is a well-studied water body due to these long-term water quality problems10,14,15. Existing monitoring programs usually focus on the main sources (e.g., riverine nutrients) and implications (e.g., oxygen levels, fish status) of eutrophication10. However, monitoring programs rely on repeat observations at the same stations to the detriment of large-scale snapshots of the entire region. In addition, groundwater tracer data (e.g., radium isotopes, 222Rn)17,18,19,20 have only been reported in a few local-scale studies in the Baltic region21,22,23,24,25,26. SGD has been identified as a critical, yet largely unquantified contributor to Baltic Sea water and biogeochemical budgets27. Here, we report results from extensive hydrogeochemical and isotopic characterizations collected during two Baltic-wide research expeditions. Our observations characterize groundwater and river waters and associated biogeochemical parameters in the Baltic Sea water column. These types of data are not only sparse in the Baltic Sea, but also in low oxygen environments in general. Therefore, these data contribute to Baltic Sea research and pollution modeling efforts as well as the growing community interested in SGD research.

This dataset is currently used by the authors for specific evaluations that will result in other scientific publications. The first of these publications quantifies vertical mixing and benthic silicate fluxes using vertical profiles of radium and silicate28, and estimates horizontal nutrient fluxes using 224Ra observations in coastal transects29. The full dataset is of further potential to those interested in topics such as coastal hydrology (e.g., SGD), marine biogeochemistry, and benthic-pelagic coupling. The data may also be of interest to those working with coastal seas like the Baltic Sea, as well as for comparative studies with other regions (e.g., North Sea, Black Sea, Mediterranean Sea) or on the global scale.

Methods

Sampling approach

We collected hydrogeochemical data during two extensive Baltic Sea expeditions targeting both marine surface and deep waters, as well as land-based endmembers, including river and beach groundwater (Fig. 1).

Fig. 1
figure 1

Schematic of site and sample types. Beach transect sites included beach groundwater 0.5 m below the sediment surface (dark red X’s) and nearshore surface water (dark red circles) up to 500 m from the shoreline sampled in 2024. Cruise shore-perpendicular transects aligned with beach transects (red circles) were sampled 1–20 km from the shoreline in 2023. For cruise transects, deep water (red crosses) was sampled at the transect endpoints. Conductivity, temperature, depth (CTD) sites included deep water vertical column samples (white crosses) and surface water samples (white circles). Additional sample types include longitudinal sites (white circles, surface water only sites during the cruise) collected between the CTD sites and river surface water (land sampling campaign, see Fig. 2B).

Full size image

Sampling of the Baltic Sea water column was conducted onboard R/V Skagerak from September 10 through October 3, 2023. We collected spatial surface water survey data, vertical water column depth profiles via conductivity, temperature, depth (CTD) instrument casts, and discrete water samples along a 5,000 km cruise track (Fig. 2A). Overall, 542 discrete samples from 302 stations were collected, including 302 samples from surface water (~4 m depth) and 240 samples from the deeper water column from CTD casts. The sampling scheme for discrete samples included three main types of stations – CTD, longitudinal, and transect. Among these main station types, sample types included surface and deep water (Table 1). Longitudinal stations were spaced on transects every ~20 km, where only surface water samples (n = 134) were collected. CTD stations were spaced approximately every 4 stations and consisted of deeper water (n = 217) and surface water (n = 50) samples. Cruise transects (n = 14 transects) consisted of shore-perpendicular transects where we collected both surface water (n = 118) and deep water (n = 23) samples. The approach for sampling transects included a surface and deep water sample pair at the closest and furthest from shore stations (~10 to 20 km apart), with typically 7 surface water stations in between these two points at ~1–5 km intervals (higher density closer to shore). Date, time, duration of sampling, and station/sample type were noted for all stations and samples.

Fig. 2
figure 2

Map of sample locations. (A) Cruise track (blue line) and 302 discrete sampling stations included in the cruise data. Cruise station numbers (BS_###) progress from 001–303 (only transect stations are labelled in the figure for readability). CTD stations are indicated by white crosses. Longitudinal stations are indicated by light blue circles. Stations belonging to cross-slope transects are indicated by red circles with the range in station numbers in red text. (B) Land-based sampling, with sample numbers BL_### ranging from 001–170. Beach transect sites (including nearshore surface water and groundwater samples) are shown by dark red X’s, with corresponding sample numbers labeled in dark red text. Blue triangles denote river sampling stations Fig. 2B).

Full size image
Table 1 Number of discrete samples collected by station type and from surface water, deep water, and groundwater.
Full size table

A second, land-based sampling campaign targeting terrestrial endmembers took place in May and August 2024 (Fig. 2B). During this campaign, a total of 170 discrete samples were collected from rivers (n = 46), beach groundwater (n = 77), and beach surface water (n = 47) to characterize the multiple terrestrial endmembers discharging into the Baltic Sea. Beach samples were organized along shore-perpendicular transects and targeted the same areas as the cruise transect sites (Figs. 1, 2B). Surface water samples (rivers and beach surface water) were collected from the upper 0.5 m of the water column. Groundwater was sampled by digging holes on the beach until the water table was reached. We then used a peristaltic pump to fully drain the standing water in the hole three times, allowing the hole to fully recharge before finally taking the sample. Samples were then collected using a peristaltic pump. While we carefully sampled recently-recharged beach groundwater, groundwater sampling may be subject to minor degassing that could lead to underestimated concentrations of CH4, CO2, and DIC30.

Cruise sampling campaign

Surface water survey sampling

We conducted a continuous underway spatial surface water (~4 m depth) survey for hydrographic parameters. An onboard Ferrybox system31 (Ferrybox I; -4H-JENA engineering GmbH, Germany) was used to measure water temperature (°C), specific conductivity (S/m), and practical salinity with a SBE 45 sensor at 1-minute intervals. Additional variables, including datetime (Coordinated Universal Time; UTC), latitude (decimal degrees N), longitude (decimal degrees E), course (degrees), ship speed (nautical miles per hour), oxygen concentration (mg/L), oxygen saturation (%), chlorophyll-a (µg/L), phycocyanin (ppb), turbidity (Nephelometric Turbidity Unit; NTU), and pressure (mbar) were logged simultaneously on the Ferrybox system. Salinity refers to practical salinity following the standard established by the Joint Panel on Oceanographic Tables and Standards32, unless otherwise specified. pH (total scale) in surface water was measured using a CONTROS HydroFIA pH flow-through pH analyzer (measurement interval = every 10 minutes). During the measurements, the dye m-Cresol Purple is added to the sample, and the absorption spectra are measured using VIS absorption spectrometry at 25 °C, allowing for calculation of pH33. The HydroFIA pH has an analytical accuracy of ± 0.01 pH units and precision of 0.005 pH units34,35.

CTD sampling

Vertical depth profiles and discrete deep water samples were collected from 69 stations (including 19 transect stations) via the onboard CTD (Sea-Bird SBE 911, SBE 43, WET labs ECO AFL/FL) and rosette water sampler (Sea-Bird SBE 32). At each CTD station we collected six discrete deep water samples, in increments starting ~2–4 m from the seafloor (labelled “BOT”), and then 10, 20, 30, 40, and 50 m from the “BOT” sample as well as a surface water sample. In some cases, fewer samples were collected due to a shallower water column. Measured parameters include datetime (UTC), pressure (dbar), water depth (m), water temperature (°C, Sea-Bird SBE 3), conductivity (m/S, Sea-Bird SBE 4), salinity, dissolved oxygen concentration (mL/L, Sea-Bird SBE 43), dissolved oxygen saturation (%, Sea-Bird SBE 43), fluorescence (mg/m3, WET Labs ECO-AFL/FL), light transmission (%), and turbidity (NTU, WET Labs ECO).

Meteorology

Continuous monitoring of meteorological parameters was obtained with an onboard weather station (Observator OMC) and logged every 2 minutes during the cruise. An average was taken in cases where two measurements were logged at the same time stamp at slightly different latitudes or longitudes ( ± 0.000003°). Parameters collected include datetime (UTC), air temperature (°C, OIC-406 probe, ± 0.1 °C accuracy), relative humidity (%, OIC-406 probe, ± 0.8% accuracy), air pressure (mbar, OMC-506 sensor, ± 0.3 mbar accuracy), relative wind speed (m/s, OMC-160, ± 2% accuracy), and wind direction (degrees, OMC-160, ± 3° accuracy).

222Rn

A 222Rn spatial surface water survey was conducted throughout the cruise. 222Rn activities (Bq/m3) in surface water were measured using a radon-in-air detector (Durridge RAD7) with RAD-AQUA accessory36. Water was run continuously from the onboard pump system to the air-water exchanger and then to the radon-in-air detector with a measurement interval of 30 minutes. The minimum detection limit of the instrument for 222Rn in air was 4.0 Bq/m3. Overall, the mean analytical uncertainty during the survey (2σ; given as an output from the RAD7 instrument) was 63% for data points above the minimum detection limit. RAD7 uncertainties are based on counting statistics and integration time, thus higher uncertainties were due to relatively low 222Rn activities for most of the survey and short integration time (30 min). Considering only data points with 222Rn in water concentrations >30 Bq/m3, the mean analytical uncertainty becomes 34%.

Land sampling campaign water parameters

During the land sampling campaign, standard water quality parameters including water temperature (°C; ± 0.2 °C accuracy), salinity ( ± 1% accuracy), pH (NBS scale; ± 0.2 units accuracy), and oxygen saturation (%, ± 0.1% accuracy) were measured using a calibrated YSI ProDSS Digital Water Quality Meter. Latitude and longitude were determined using a handheld GPS. For beach transect sites, distances between sampling points were measured using a measuring tape to ensure accuracy and to correct for GPS resolution.

Experimental analyses for discrete samples

Short-lived radium isotopes

Radium samples were collected into 5 L (groundwater), 60 L (cruise deep water and land-based surface waters), and 300 L (cruise surface water) containers. The collected water was passed through cartridges containing 20 g of MnO2-coated fibers that extract radium. 224Ra (t1/2 = 3.6 days) and 223Ra (t1/2 = 11.4 days) activities were then measured with a Radium Delayed Coincidence Counter (RaDeCC)19 within 48 hours of sample collection. Samples were re-run 4 weeks after collection to correct 224Ra activities for ingrowth of its parent isotope, 228Th. Quality control of the raw radium data included screening for counting anomalies such as spurious counts and leaks. Further quality control on the raw radium counting data assessed for cross-channel interferences due to logistical constraints that did not allow for a second 1 week count to correct for 227Ac, which can impact 223Ra37.

Samples collected from low oxygen deep water (n = 94) were processed by aligning 2–4 cartridges containing MnO2-coated fibers to quantify a potentially lower radium extraction efficiency due to the presence of sulfides and lack of oxygen. These samples were then analyzed on the RaDeCC as described above. Radium activities from the low oxygen deep waters were calculated by quantifying the radium extraction efficiency28. All radium activities were corrected for radioactive decay to the sampling time. RaDeCC detector efficiency was quantified at both the beginning and end of each sampling campaign. Final uncertainties (median =  ± 6.1%) represent 2σ and account for detector efficiency, counting statistics, and radioactive decay38. Samples with fewer counts than the background or <0.02 cpm in the 220 channel were below the minimum detection limit.

Dissolved inorganic nutrients

Samples for dissolved inorganic nutrient concentrations (µmol/L) were first collected into 60 mL syringes, which were then passed through a 0.45 µm cellulose acetate filter into 50 and 15 mL polypropylene Falcon tubes. Dissolved inorganic nitrogen (DIN; sum of NO3−, NO2−, and NH4+) and dissolved inorganic phosphorus (DIP; PO43−) samples were kept frozen until analysis. Dissolved silicate (DSi; SiO44−) was kept at 4 °C until analysis. All dissolved nutrients were analyzed with an AA500 AutoAnalyzer (SEAL Analytical) at the Institute of Oceanology, Polish Academy of Sciences. Repeat measurements of certified reference materials (CRM RM-BU; National Metrology Institute of Japan) were used to confirm analytic quality Accuracy, expressed as recoveries, was >99% for all parameters, while precision, indicated by the relative standard deviation, was <2%. The detection limit was 0.006, 0.003, 0.045, 0.012, and 0.027 µmol/L for NOx (NO3− + NO2−), NO2−, NH4+, PO43−, and SiO44−, respectively. Final NO3− concentrations were calculated by subtracting the NO2− concentration from NOx.

Dissolved Inorganic Carbon and Total Alkalinity

Samples for dissolved inorganic carbon (DIC) were collected into 60 mL syringes and immediately filtered through 0.7 µm GF/F filters into 12 mL borosilicate glass exetainers without headspace and stored at 4 °C until analysis. Samples were not poisoned due to prohibited use of mercury(II) chloride (HgCl2) at Swedish institutions. Instead, DIC samples were analyzed within 24 hours (onboard the research vessel for the cruise sampling) and 72 hours (land sampling) of collection to mitigate sample alteration. DIC concentrations in filtered, unpoisoned samples can change by ±0.5 to 2% within 4 days of sample collection according to in-house lab storage experiments39. DIC samples were analyzed using an AS‐C5 (multi‐port version, Apollo SciTech, LLC) with a non-dispersive infrared CO2 detector (LI‐850, LI‐COR, USA) using 3% H3PO4 acid with 7% NaCl, and lab grade N2 carrier gas. Samples were run in replicates of 3–5 times, until three replicates consistently met a suitable precision (<±2 and ±1 µmol/kg for the samples measured during the cruise and land campaigns, respectively). Final DIC concentrations (µmol/L) were calibrated using certified reference materials40 (CRM Batch nos. 189 and 203) using the replicate mean and converted from µmol/L to µmol/kg using in-situ salinity and the temperature measured during the analysis.

The method for analyzing total alkalinity (TA) differed slightly between the cruise and land sampling campaigns. During the cruise sampling, we collected TA samples in 250 mL borosilicate glass bottles, which were then kept for ~20 min in a 20 °C water bath before analysis ( <24 hours of sample collection). TA was then measured via a semi-closed potentiometric titration technique using Gran-point evaluation with 0.05 M HCl41. The system measures alkalinity in µmol/L. During the land sampling, water for TA analysis was filtered through 0.7 µm GF/F filters into 50 mL polypropylene Falcon tubes. Samples were stored at 4 °C and analyzed with semi-closed potentiometric titration using Gran-point evaluation with a high precision titrator (Metrohm 888 Titrando with Tiamo light)42. Concentrations <2500 µmol/kg were titrated with 0.01 mol/L HCl, concentrations >2500 µmol/kg with 0.025 mol/L HCl. For both the cruise and land campaigns, TA samples were calibrated with certified reference material40 (CRM Batch nos. 189 and 203 for the cruise and CRM Batch no. 216 for the land). For all samples and CRM analyses, TA in µmol/kg was calculated using in-situ salinity and the temperature measured during the titration. The mean precision was <±3 µmol/kg and ±6.7 µmol/kg for the cruise and land sampling, respectively.

Methane concentrations

Samples for dissolved CH4 concentrations were sampled directly from the Niskin bottle into 60 mL syringes via gas tight tubing. Samples were then transferred into 22 mL borosilicate glass vials and spiked immediately with 0.2 mL 7 M ZnCl2 for preservation43. Vials were capped without headspace and stored at 4 °C until analysis. The sample was measured 4-5 months after sample collection using an N2 gas headspace technique44 with a flame ionization detector gas chromatography (Thermo Scientific Trace 1300). Two certified gas standards were used for calibration (Air Liquide Gas AB): one low concentration (1.9 ± 0.04 ppm), and one high concentration (50.1 ± 1.0 ppm). For samples with headspace concentrations <1.9 ppm, only the low concentration standard was used, with the calibration curve was forced through origin. The detection limits were ~0.2 ppm, with an analytical precision (expressed as relative standard deviation of multiple 1.9 ppm CH4 standard area counts; n = 50) of 2.8% and accuracy (represented by the relative error) of 2.2%. Dissolved CH4 concentrations (nM) were calculated from the measured headspace concentration (ppm) based on the solubility coefficient45. Replicate sample analyses resulted in a median precision of 8% for the peak area and 5% for the headspace concentration based on the relative standard deviation.

δ13CDIC, δ13CCH4, and δ13CCO2

Samples for δ13CDIC were collected into syringes and then passed through a 0.7 µm GF/F filter into borosilicate glass 12 mL exetainers. Cruise samples were kept frozen until analysis within 1.5 years of sampling. Samples from the land campaign were frozen until arrival at Leibniz Institute for Baltic Sea Research, where they were spiked with HgCl2. These samples were then kept under dark conditions at 4 °C until analysis within 10 months of sampling. Stable isotope analyses by means of continuous-flow isotope-ratio mass spectrometry (CF-irms) were carried out using a Thermo Finnigan MAT253 gas mass spectrometer attached to a Thermo Electron Gas Bench II via a Thermo Electron Conflo IV split interface46. Sample solutions were allowed to react with H3PO4 for at least 18 h at 23 °C in the Gasbench II before introduction into the gas mass spectrometer. International calibration materials (NBS-19, IAEA Li carbonate standard), and Solnhofener Plattenkalk were used to calibrate the measured isotope signals against the V-PDB scale. According to replicate measurements of standards, the reproducibility was better than ±0.1‰46.

Samples for δ13CCH4 and δ13CCO2 were collected directly into 60 mL syringes with gas tight tubing and transferred in 250 mL and 112 mL borosilicate glass bottles fixed with 0.4 mL 7 M ZnCl247,48. Bottles were sealed with butyl rubber stoppers and stored without headspace in 4 °C. δ13CCH4 and δ13CCO2 were measured within 1 year of sampling using a Small Sample Isotope Module (SSIM Picarro) coupled to a cavity ring-down spectrometer (Picarro G2201-i) using synthetic air (HiQ Synthetic Air 5.0; Linde Gas) as carrier gas and headspace. Samples outside the instrument operation ranges of 1.5 to 50 ppm CH4 were diluted. The operation range for CO2 was 380 to 2000 ppm. Stable isotope standard gas (Airgas Specialty Gases) was used to calibrate stable isotope values (−69 to −45‰, and −29 to −8.6‰ vs. V-PDB for δ13CCH4 and δ13CCO2, respectively). Precision (expressed as standard deviation) ranged from ±0.14 to ±1.08‰ for δ13CCH4 and from ±0.60 to ±1.04‰ for δ13CCO2. Accuracy (expressed as absolute error) ranged from 0.96 to 7.30‰ and 0.56 to 4.09‰ for δ13CCH4 and δ13CCO2, respectively. Quality control measures included removing samples with CH4 headspace concentrations <1.5 ppm. Five samples in the top and bottom 10th percentiles of the distribution were removed.

Dissolved organic carbon and total dissolved nitrogen

During the cruise campaign, we collected samples for dissolved organic carbon (DOC) and total dissolved nitrogen (TDN) analysis from the surface and bottom water at CTD stations. Water samples were taken from three different Niskin bottles from the CTD rosette into pre-rinsed 20 L carboys (HDPE). Samples were sequentially filtered through pre-rinsed 1.0 µm Causapure (CPR-001-09-DOX, polypropylene, Infiltec, Germany) and 0.1 µm Causa-PES (CPS-S10-10-DOV-A, polyethersulfone, Infiltec) filter cartridges into pre-rinsed 10 L carboys. An aliquot (30 mL) was collected in HDPE vials and immediately acidified with diluted hydrochloric acid (∼8 mol/L) to a pH of 2 for the analysis of DOC and TDN.

During the land campaign, we collected water samples into syringes and then filtered through a 0.7 µm GF/F filters into pre-cleaned 60 mL amber borosilicate vials and preserved with 0.2 mL of 1 M H3PO4. All samples were stored at 4 °C until further analysis.

DOC and TDN were measured in triplicates by high-temperature catalytic oxidation using TOC-VCPH/CPN and TOC-LCPH/CPN Total Organic Carbon Analyzers (both Shimadzu Corp.) within 1-3 months after collection at the Institute for Chemistry and Biology of the Marine Environment in Germany. Each analytical replicate consisted of three to five injections, with peak area variability kept below 3%. Analytical accuracy and precision were validated using the deep seawater reference as provided by the Hansell Organic Biogeochemistry Lab (University of Miami, USA). Accuracy usually fell within the consensus range of the reference material but never deviated by more than 2 µmol/L. Precision was generally better than 5% except for one run with a precision of 6.2%. All reported values exceeded the limit of detection, which was determined as the mean of ultrapure water blank measurements plus 10 times the standard deviation.

Water isotopes

Samples δ18OH2O and δ2HH2O analysis were collected into 4 mL glass vials without headspace and kept cool and dark until measurement within 1.5 years and 7 months of sample collection for the cruise and land campaigns, respectively. Stable isotope analysis was carried out by means of Laser-cavity ring down spectroscopy (LCRDS) using a Picarro L2140-i49. For brackish water samples a metal liner was used to reduce the potential effect of the salt50. Reference materials (SLAP, VSMOW, USGS48, internal standards) were used to calibrate the measured isotope ratios towards the Vienna Standard Mean Ocean Water (VSMOW) scale. Results are reported in ‰ vs. VSMOW. Average analytical uncertainties for δ18OH2O were ± 0.045‰ and ± 0.25‰ for δ2HH2O.

Data processing

Initial quality control

We first performed conversions to ensure matching units for all data sources. All data time stamps were then converted to UTC. The survey and CTD data underwent additional processing and were matched to the discrete samples as described below.

Cruise surface water survey and discrete sample matching

Cruise surface water samples were first matched to the latitude and longitude logged on the Ferrybox based on sampling time and then confirmed with GPS notes taken by the ship captain during sampling. After this, other variables of interest from the Ferrybox (salinity, water temperature, dissolved oxygen saturation, and turbidity) were matched to discrete samples by taking the average value for the variable from the time sampling began until it was completed.

Survey data from the HydroFIA pH were matched to the Ferrybox data by date and time and then processed using the Ferrybox salinity data. Final pH values from the cruise are on the total scale at 25 °C and in-situ salinity. The conversion to 25 °C was made in PyCO2SYS v1.8.351 on the total scale using the dissociation constants of carbonic acid52 and bisulfate53. A mean surface (<5 m) TA content (1538 µmol/kg) was used as the second carbonate system input variable. When converting to the total scale, the difference between using the mean surface TA value versus the closest discrete TA sample was 2.3 × 10−5 ± 5.0 × 10−5 pH units (n = 2568). This is at least two orders of magnitude lower than both the accuracy and precision of the method and does not introduce large systematic pH errors.

Post-processing and cleaning of 222Rn data included checking data quality for voltage spikes, that relative humidity remained below 10%, and time stamps converted to UTC. 222Rn survey data were matched to the Ferrybox data by downsampling the 1-minute interval data to 30-minute intervals using the mean and shifting by one time-step to account for equilibrium adjustment for the RAD7 instrument. 222Rn in exchanger air activities were then converted to 222Rn in water (Bq/m3 by calculating the Ostwald solubility coefficient54 using the matched water temperature and salinity data from the Ferrybox. Final 222Rn in water uncertainties were calculated using error propagation.

Cruise CTD profiles and discrete sample matching

Variables of interest from the raw downcast CTD profile data were extracted and processed from.cnv files using the Python package pycnv. Extracted CTD profiles for each station were cleaned, including removing rows that were flagged by the sensor for quality control reasons, and cutting the data at the maximum depth. Additional derived CTD variables including conservative temperature (CT; °C), absolute salinity (SA; g/kg), and water density (ρ; kg/m3) were calculated from CTD data using the Thermodynamic Equation of Seawater 2010 (TEOS-10) equation of state55 and the Gibbs Seawater (GSW) Oceanographic Toolbox in Python56. Discrete samples were then matched to CTD profile variables by water column depth.

Data Record

Data products and the associated metadata can be found in an open-source Zenodo repository57. This dataset spans from September 10 to October 1, 2023, and from May 2 to August 28, 2024, for the cruise and land-based campaigns, respectively. Both campaigns cover latitudes 53.35 to 65.88 °N and longitudes 10.59 to 24.99 °E. In all, 35 variables were measured during the cruise and 22 variables during the land-based campaign. Uncertainties for all parameters in the dataset are either available on a per-sample basis (i.e., as an additional column in the dataset) or can be estimated on a per-method basis described in the text, allowing users to estimate values for their own work based on well-quantified uncertainty reporting. The data are organized into seven CSV files following Findable, Accessible, Interoperable, and Reproducible (FAIR) principles58:

  1. 1.

    Ferrybox surface water survey (1-minute resolution) (1_Cruise_Ferrybox_1min.csv)

  2. 2.

    Meteorological survey (2-minute resolution) (2_Cruise_meteo_2min.csv)

  3. 3.

    HydroFIA pH survey with matched and down-sampled Ferrybox data (10-minute resolution) (3_Cruise_HydroFIA_pH_10min.csv)

  4. 4.

    222Rn surface water survey with matched and down-sampled Ferrybox and meteorological data (30-minute resolution) (4_Cruise_RAD7_30min.csv)

  5. 5.

    Processed CTD profiles (5_Cruise_CTD.csv)

  6. 6.

    Cruise campaign discrete sample data (6_Land_discrete.csv)

  7. 7.

    Land campaign discrete sample data (7_Land_discrete.csv)

Survey, CTD, and land-based discrete data variables, column names, and file names are summarized in Table 2. Discrete data can be found separate files for the cruise and land-based campaigns. For the land campaign, discrete measurements of 4 additional hydrographic parameters (salinity, water temperature, dissolved oxygen %, and pH) were also taken for each sample but are not shown in the Table 2.

Table 2 Summary of survey and vertical profile data collected on the cruise.
Full size table

Technical Validation

CTD data underwent a two-stage cleaning procedure. First, we removed data that were flagged by the instrumentation (given a value of −9.99 × 10−29 in any column, n = 106,911), which left 258,082 data points remaining. This was most frequently found at the start of the record for an individual CTD profile due to instrumentation startup and equilibration. We then cleaned the profiles to ensure the CTD was in the water and included downcast data only by cutting the record at the maximum depth recorded. This resulted in further removal of 6,195 rows, with 251,887 rows remaining in the final CTD record from all profiles.

Cleaning of survey (surface water basic water quality parameters, meteorology, and 222Rn) data was minimal compared to the CTD data. However, several periods of missing data were identified (Table 3). In the cases of the Ferrybox, HydroFIA, and meteorological data, these were caused by either instrumentation error or deliberate pauses in recording due to harboring to avoid storms or allow for crew changes. For the 222Rn survey, the first day of data were removed due to high (>10%) relative humidity in the RAD7, which can cause inaccurate readings.

Table 3 Data gaps in survey data from the cruise.
Full size table

All experimental data were analyzed using well-established methods and checked using method specific quality control measures. Discrete sample matching quality control included confirming the station name, location, and sampling date and time. As an additional quality control check, we calculated the difference in depth between the discrete sample with the matched CTD profile scan to ensure this value never exceeded 5 cm. A second data validation was carried out to ensure the matched cruise surface survey and CTD data were consistent with individual discrete samples. Variables from different sources (i.e., CTD, Ferrybox, and experimental data) in discrete cruise samples were plotted against each other identify outliers and unexpected trends. Additionally, surface and deep water data were plotted by station for CTD stations for comparison. For the land campaign, which only consisted of discrete samples, the final quality control consisted of checking for anomalously high or low values and were evaluated for any unreasonable trends between variables.