Methodology:

Seawater samples were collected between 28 December 2010 and 28 February 2020, in Ryder Bay. The majority of samples were collected at 15 m depth. At the start of the Rothera Time Series in 1998, 15 m was chosen as the primary seawater discrete sampling depth, as it is the mean depth of the chlorophyll maximum (Venables et al., 2023). At a lower frequency, seawater samples were collected at additional depths, e.g., at 40 m, 5 m, 2 m, or 0 m. Seawater samples for DIC/TA analysis were collected in 250 mL or 500 mL borosilicate glass bottles from the side of a small boat or through a hole in the sea ice, following Standard Operation Procedure (SOP) 1 in Dickson et al. (2007). Samples were fixed with mercuric chloride upon return to Rothera Research Station to arrest any biogeochemical processes that could alter the DIC and/or TA content of the sample after collection. Bottles were closed with grease to create a gas-tight seal. Samples were analysed either at Rothera station, or at the University of East Anglia (UEA), United Kingdom, using VINDTA version 3C (Versatile INstrument for the determination of Dissolved inorganic carbon and Total Alkalinity), following SOP 2 and 3 in Dickson et al. (2007). Raw data were subsequently computationally processed using Python.

The Rothera Time Series spatially covers a single location (i.e. Ryder Bay) and temporally covers time between 2010-12-28 and 2020-02-28 at a variable time frequency, ranging between 1 to 2 weeks. The majority of samples (n=421) are collected at 15 m depth. Other samples are collected at 40 m (n=90), 2 m (n=19), 5 m (n=16), 0 m (n=12), or alternative depths (n=33).

Data collection:

Samples were analysed on land for DIC and TA using VINDTA version 3C (Versatile INstrument for the determination of Dissolved inorganic carbon and Total Alkalinity), following SOPs 2 and 3 in Dickson et al. (2007). The VINDTA collects a sample aliquot for DIC analysis with coulometry, and an aliquot for TA analysis, using potentiomatric titration. All raw analytical data were computationally processed following SOPs in Dickson et al. (2007). Calkulate Python package (version 23.2.2; Humphreys et al., 2022a) was used to determine the total alkalinity values in units of micro-mole per kilogram from the raw data of the potentiometric titration method, and using Certified Reference Material (CRM) from A. Dickson's laboratory at Scripps Institution of Oceanography to calibrate the titrant molinity. DIC content was calculated by determining the coulometer counts per micro-mole of sample aliquot, using CRMs as standards. This values were then converted to units of micro-mol per kilogram of seawater, using the volume analysed and the density of the sample. The analytical uncertainties for DIC and TA are less than 2.7 micro-mol kg-1, except for data between 2016-2017, which had higher uncertainties of less than 5.4 micro-mol kg-1.

PyCO2SYS Python package (version 1.8; Humphreys et al., 2022b) was used to determine the fugacity of CO2 (fCO2) based on DIC, TA, salinity, seawater temperature, and nutrient data, as well as the following parameterisations: dissociation constants of bisulfate (Dickson et al., 1990), hydrogen fluoride (Dickson & Riley, 1979), and carbonic acid (Goyet & Poisson, 1989), and the boron-salinity relationship (Uppstrom, 1974). fCO2 was only determined for the datapoints at 15 m depth. When salinity or temperature data was unavailable at 15 m depth, the missing CTD data was be replaced by monthly mean values in order to estimate fCO2. Using PyCO2SYS, the mean propagated uncertainty in fCO2 due to the uncertainties in DIC and TA content was estimated to be 8 micro-atm.

For information on nutrient sample analysis or CTD data, see Venables et al. (2023).

Data quality:

Analytical precision of DIC and TA content is determined from CRMs runs on the VINDTA. On each analysis day, sample runs are bracketed by a repeat run on a CRM bottle at the start and at the end. Results of these runs were used to quality check the instrument, and to determine the analytical uncertainty of the measurements by calculating the 1-sigma of all CRM run results per CRM batch. The certified values of the CRMs were used to calibrate the coulometer counts and potentiometric titration and determine DIC and TA content, respectively.

Results for CRMs were excluded when the analytical instrument clearly did not perform well (usually observed in the lab), or when their reproducibility was poor. Results for seawater samples were excluded if the instrument performed poorly, or when the sample had been compromised due to contamination during sampling and/or storage (e.g., if the sample was not gas-tight).

Analytical uncertainties for DIC and TA are less than 2.7 micromol kg-1, with the exception of data between 2016-2017, which had higher uncertainties of less than 5.4 micromol kg-1.

Quality flags in the dataset are indicated by the ''_FLAG''-extension in the column name and follow the WOCE quality codes.