Bracher, Astrid; Soppa, Mariana A; Banks, Andrew C; Xi, Hongyan; Chaikalis, Spyros; Pitta, Elli; Röttgers, Rüdiger; Zeri, Christina (2025): Absorption coefficients by coloured dissolved organic matter at 443 nm from a global in-situ collection of open ocean, coastal and inland surface waters matched to Ocean Land Colour Instrument [dataset]. PANGAEA, https://doi.org/10.1594/PANGAEA.983195

This in situ data set of absorption coefficients by coloured dissolved organic matter at band centred at 442.5 nm (aCDOM(443)), note: for brevity the band is named 443) consists of different data sets gathered together from measurements collected in open, coastal, and inland surface waters spread around the globe and covering the time from first data delivery by Ocean Land Colour Imager (OLCI) on Sentinel-3A (S3A) in May 2016 until November 2022 which were matched to OLCI on S3A and S3B and used in the paper by Bracher et al. (2025). We only used the absorption coefficient data derived from measurements on discrete water samples to ensure a similar method procedure followed and a similar uncertainty. It includes publicly available data and newly collected, measured and analysed data sets from the Phytooptics group at the Alfred Wegener Institute Helmholtz Centre for Polar and Marine Research (AWI, PI: Astrid Bracher) and Hellenic Centre for Marine Research (HCMR, PI: Andrew C. Banks). This collection was matched that in situ data points had to fall within the 3x3 OLCI FR pixel box and a time window of + 12 hours which followed established community protocols (IOCCG 2018, IOCCG 2019) and particularly EUMETSAT's OLCI matchup protocol (EUMETSAT 2022). Firstly, a pre-processing for quality control and a conversion of the considered in situ data to a common format following Valente et al. (2022) was performed. We flagged and disregarded the following data from the final quality-controlled data set which had (1) unrealistic or missing date or geographic coordinate fields, (2) poor quality (e.g., original flags) or method of observation that did not meet the criteria for the dataset (e.g., not defined in the community protocols (IOCCG 2018, 2019a), and (3) spuriously high or low data. For the last item, the following limits were imposed: [0.0001–10] m−1 for aCDOM(443). OLCI pixels were discarded when flagged with the recommended flags in (EUMETSAT 2022), and the remaining matchups were only considered valid if more than 50% of satellite pixels were available at remote sensing reflectance centred at band 560 nm (Rrs(560), e.g., 5 out of 9 for the 3x3 criterion) per an in situ data point, and a coefficient of variation <0.2. Dedicated matchup software developed by EUMETSAT was used to ensure that the validation process followed the established guidelines, ThoMaS (the Tool to generate Matchups of OC products with S3 OLCI https://gitlab.eumetsat.int/eumetlab/oceans/ocean-science-studies/ThoMaS). In situ data from AODN-2 and Lehmann22 (see description of specific datasets below) were already provided at the nominal OLCI band 443 nm. All other aCDOM(λ) data were provided in hyperspectral resolution (1nm, 2nm or around 3.3 nm resolution). Following Zibordi et al. (2023), these hyperspectral absorption coefficients were transformed to the nominal OLCI bands by averaging over the specific bandwidth. The OLCI matchup data, based on their associated RRS data at the first eight OLCI bands, were assigned to the specific optical water classes (OWCs) according to the Mélin & Vantrepotte (2015) classification. This contains 17 OWCs which range from very turbid to (OWC 1) oligotrophic to very clear waters (OWC 17). The OWC is also delivered for each matchup point (if the assignment fails the field contains "NaN". We provide also for OLCI the standard deviation of the OLCI matchup data to a in situ data point within the 3x3 pixels. For the in situ data we provide the estimate of the uncertainty for each matchup point further described in Bracher et al. (2025).

The different data compilations are described as follows:

- Bracher22: This collection contains aCDOM(λ) published in Pangaea from AWI (Bracher et al. 2021a, 2021b, 2021c, 2021d, 2021e, 2021f, 2024a, 2024b) matching the S3A and S3B mission time and considered for coupled model evaluation in Alvarez et al. (2022). This data set encompasses data from six Atlantic expeditions (2016-2019: PS99.1, PS99.2, PS103, PS107, PS113, PS121) covering polar, temperate, tropical and shelf seas. which were missing in the original "Bracher" data set used in Alvarez et al. (2022).

- Röttgers23: From the Röttgers et al. (2023) large IOP data compilation measured during campaigns in the German Bight and adjacent regions from 2008-2021, we selected the data from S3 mission lifetime. This data set encompasses aCDOM(λ) from two RV Heincke North Sea campaigns (HE488 and HE517 in late spring 2017 and late summer 2018, respectively, and from the LP2021 German Bight campaign in summer 2021.

- SEABASS: From the absorption coefficient data submissions to SeaBASS (https://seabass.gsfc.nasa.gov/, download 28 September 2023) we used all hyperspectral aCDOM(λ) data overlapping S3A and S3B mission which are not contained within the Valente et al. 2022 and Alvarez et al. 2022 compilation. This published data comprises mainly US waters (campaigns: ACIDD, BBOP, Bowdoin-Buoy, CARBON_ESTUARIES, CBWQ, Cyanate, ECOA2, gomecc-3, IDS_LIS, Lake_Superior, PACE_ABSCLOSURE, PLUMES_AND_BLOOMS, PRUDHOE_FRESHETS, SEAHAWK, Sea2Space, SFMBON) and several US oversea expeditions (Arctic-RSWQ in Kaktovik, Norton Sound and Yukon_in 2019, ArcticCC in Northern Bering Sea in 2022, CPA at coral reefs in West Pacific in 2017).

- AODN: Here we included new IOP data submissions to the Australian Open Access to Ocean Data portal (AODN, https://portal.aodn.org.au/, download 19 July 2023), not provided in Valente et al. (2022) or Lehmann et al. (2023) and matching the S3A and S3B OLCI lifetime. AODN-1 contains hyperspectral aCDOM(λ) from the CSIRO (Commonwealth Scientific and Industrial Research) Hydrochemistry Facility Integrated Marine Observing System (IMOS, https://research.csiro.au/hydrochemistry/projects/integrated-marine-observing-system-imos/). They contain hyperspectral aCDOM data and are from several expeditions in Australian waters (at Torres Strait in 2016, at the mouth of the Fitzroy River in 2017, and at the Coral Sea and Queensland Shelf in 2016 (IN2016) and 2020 (IN2020) and from the Lucinda Jetty Coastal Observatory (https://researchdata.edu.au/imos-srs-satellite-observatory-ljco/476837). The AODN-2 data set contains just aCDOM(443) data from the Australian Great Barrier Reef Marine Monitoring Program (https://www2.gbrmpa.gov.au/our-work/programs-and-projects/marine-monitoring-program) and the Australian Inshore Water Quality Monitoring Program (https://www2.gbrmpa.gov.au/our-work/programs-and-projects/marine-monitoring-program/inshore-water-quality).

- Banks-new: These are hyperspectral aCDOM(λ) data from the under-sampled oligotrophic Eastern Mediterranean. These were collected by HCMR (PI: A. Banks) and the Joint Research Centre (JRC) on a joint optics cruise (HCMR-JRC OPTICS) in April to May 2022 and the MARRE HCMR cruise in the same area in September 2020. This data set is not included in Zibordi et al. (2023), but follows the same measurement procedure (Pitta et al. 2017, Mannino et al. 2019).

- Bracher-new: AWI (PI: A. Bracher) has conducted recently (January 2020 until November 2022) four more large expeditions spread over the temperate and polar Atlantic Ocean (MSM93, PS126, PS131 and PS133-1) and four weekly campaigns at Germany's largest inland water, Lake Constance (BS-1, BS-2, BS-3, BS-4) where about 1000 valid measurements for hyperspectral aCDOM(λ) have been collected. The measurement protocol is the same as for the "Bracher" data set in Álvarez et al. (2022).

- Lehmann22: This collection contains from the global coastal and inland water compilation from Lehmann et al. (2022), focusing on hyperspectral in situ Rrs. We considered the accompanying in situ aCDOM(443) data which fall within the S3A and S3B OLCI lifetime and are not contained in any of the other data sets (Valente22, Bracher22, Castagna22, SEABASS-new, Bracher-new).