CASCADE includes OC data from the entire Arctic Ocean with special focus on the seven Arctic continental shelf seas (Fig. 1; Table 1). Accordingly, a distinction is made among the central Arctic Ocean and the following marginal seas: Beaufort Sea, Chukchi Sea, East Siberian Sea, Laptev Sea, Kara Sea, Barents Sea (including White Sea) and the Canadian Arctic Archipelago (we exclude data from Baffin Bay, Foxe Basin and Hudson Bay, as they are outside the circum-Arctic scope of the database). For defining the limits of these Arctic shelf seas, Jakobsson (2002) is followed, which distinguishes the Arctic Ocean constituent seas using hypsometric criteria. Therein, shelf is defined as the seaward extension of the continental margin until the increase in steepness at the shelf break (Jakobsson, 2002). Data for the central Arctic Ocean were treated as one individual unit that covers all area beyond the shelf break and includes the continental slope, rise, deep basins and mid-ocean ridges.

The coordinate system used for CASCADE is WGS1984, and coordinates are kept in machine-readable decimal degrees (latitude in ∘ N, longitudes in the -180 to 180∘ format) to harmonize the data across all GIS applications. The collection of data from oceanographic stations is the main part of CASCADE and is organized in a table format that contains columns for the station number (“STATION”) and geographical coordinates (“LAT”; “LON”). The spatial references also include information about the sediment depth interval that reported data represent (“UPPERDEPTH'; “LOWERDEPTH”), where the upper depth is equal to 0 cm in the case of surface sediments. In addition, the table contains a column for water depth (“WATERDEPTH”) as reported by the data source. In cases where the water depth was not reported, the water depth was estimated using the latest version (v4) of the bathymetric map of IBCAO (Jakobsson et al., 2020) corresponding to the position of the oceanographic station and reported in a separate column (“IBCAODEPTH”). Furthermore, the name of the expedition and/or ship (“EXPEDITION”) and the year when the sample was taken (“YEAR”) are reported. For samples where the sampling year was unknown, users may use the year of publication instead.

The first stage of the CASCADE development focused on maximizing spatial coverage for surface sediments of the seven circum-Arctic shelf sea systems and the central Arctic Ocean. Here, surface sediments are defined as those collected from the water–sediment interface to a depth of maximum 5 cm. Data for surface sediments are provided in a table (“CASCADEsurfsed”) as .txt and .xlsx files and in a ready-to-use GIS shapefile format. This database also includes deeper sediments from sediment cores, which represent longer timescales and add a third dimension to the geographical referencing. Types of sediment cores are distinguished in CASCADE such that different biogeochemical processes, acting on three depositional timescales, may be addressed. The three timescales are

centennial scale cores (core scale 1) in upper sediments of the Arctic Ocean, e.g., multi-corer, Gemini corer, box corer, van Veen grab sampler, other short gravity corers up to 1 m length;

CASCADE contains information about the concentration and isotopic and molecular composition of OC in marine Arctic sediments. In addition to (i) OC concentrations (column “OC”), the database includes (ii) concentrations of TN (“TN”) and (iii) the gravimetric ratio of OC / TN (“OC / TN”), which may provide additional information about the organic matter source (e.g., Goñi et al., 2005; van Dongen et al., 2008). Furthermore, CASCADE contains data of (iv) δ13C-OC (“d13C”) as a parameter to distinguish between marine and terrestrial sources (e.g., Fry and Sherr, 1989) and (v) Δ14C-OC (“D14C”) to assess the presence of aged organic matter released from permafrost deposits (e.g., Gustafsson et al., 2011; Vonk et al., 2012) or from petrogenic sources such as sedimentary rocks (e.g., Yunker et al., 2005; Goñi et al., 2013) in marine sediments. More details about the CASCADE parameters and their units are provided in Table 2.

Data of terrigenous biomarkers may facilitate further investigations of terrigenous OC input (Table 2). The first version of CASCADE compiles total concentrations of n-alkanes with high molecular weight (HMW) and C21–C31 carbon atoms (∑C21–C31; column “HMWALK”), as well as the often separately reported more specific n-alkanes ∑C27 + C29 + C31 (“HMWALK_SPEC”). CASCADE also contains the sum of the HMW n-alkanoic acids ∑C20–C30 (“HMWACID”). Both compound classes stem mostly from terrigenous compartments as they derive from epicuticular leaf waxes of land plants with a typical pattern of dominating odd-numbered homologues for HMW n-alkanes and even-numbered homologues for HMW n-alkanoic acids (Eglinton and Hamilton, 1967). Furthermore, the database holds concentrations of lignin phenols (∑syringyl, vanillyl, cinnamyl; “LIGNIN”), which are products from the break-up of the lignin biopolymer, a compound only produced by vascular plants (Hedges and Mann, 1979). These three compound classes are frequently used as tracers of the sources and fate of terrestrial organic matter sequestered in Arctic Ocean sediments (Fahl and Stein, 1997; Goñi et al., 2000; Tesi et al., 2014; Bröder et al., 2016). It is recognized that there are more parameters that could be included, and CASCADE can add further extensions in future versions.

Each data source added to CASCADE is fully cited (in the formatting style of Earth Systems Science Data; ESSD) to maintain a high level of transparency. When applicable, citations also include a digital object identifier (DOI) that is linked to the reference in the primary literature next to each parameter column. Accordingly, the CASCADE data sheet distinguishes between a common reference for OC, TN and OC / TN data (“CN_CITATION”) as they are often combined in one measurement and separate references for OC isotopes (“d13C_CITATION”; “D14C_CITATION”) and concentrations of biomarkers (“BM_CITATION”). This facilitates registration of multiple measurements based on the same or split sediment sample material for individual oceanographic stations. A full list of references is separately provided on the CASCADE website and in the Supplement of this paper.

A part of CASCADE builds on previous separate and partly inaccessible databases of OC parameters that key partners of the CASCADE consortium and others have collected over the years. This includes data from the informal Swedish–Russian collaboration network called the International Siberian Shelf Study (ISSS; Semiletov and Gustafsson, 2009) and the “Carbon” database of the Shirshov Institute of Oceanology. This basis for CASCADE was strengthened by an extensive survey of the peer-reviewed literature and data mining in the grey literature of scientific cruise reports. To facilitate quality assurance criteria by the end users, the database also records metadata (e.g., sampling technique in the field, sample storage) and quality data when available. The quality assurance information for data in CASCADE is as follows.

Data need to be (geo-)referenceable and located in the target region (i.e., the Arctic Ocean).

Recalculations of literature data (e.g., for unit conversions) were in some cases necessary to harmonize the data to the standard units as defined in Table 2.

In CASCADE the concentration of OC is reported in percent (%) of the dry weight; values previously published as milligrams of OC per gram of dry weight were divided by a factor of 10.

CASCADE uses Δ14C with age correction (Eq. 1) to report the activity of radiocarbon according to convention (Stuiver and Polach, 1977; Stenström et al., 2011). For radiocarbon values that were reported as conventional 14C ages we used Eq. (2) to calculate the age-corrected Δ14C. 1Δ14C=Fm⋅eλC(1950-YC)-1⋅1000‰2Δ14C=e-λL⋅T14C-years⋅eλC(1950-YC)-1⋅1000‰ Here Fm is the fraction modern, λC the decay constant of the Cambridge half-life of 14C (T1/2-C=5730; λC=1/8267), YC the year of sample collection, λL the decay constant of the Libby half-life of 14C (T1/2-L=5568; λC=1/8033) and T14C-years the conventional 14C age (Stuiver and Polach, 1977).

All biomarker concentrations of HMW n-alkanes and n-alkanoic acids are reported as micrograms per gram of OC while lignin phenols are reported as milligrams per gram of OC. Biomarker concentrations that in the original publication were reported as normalized to dry sediment weight were for CASCADE normalized to the OC concentration of the sample.

CASCADE provides interpolated files (GEOtiff, ASCII; coordinate system WGS 1984 Arctic Polar Stereographic) for OC content, δ13C-OC and for Δ14C-OC in surface sediments across the Arctic Ocean. OC data were mapped in ArcGIS 10.6 and interpolated to a resolution of 5×5 km per grid cell using the empirical Bayesian kriging function (EBK; Gribov and Krivoruchko, 2020) in the commercially available ArcGIS 10.8 software package (ESRI). Kriging builds on the assumption that two points located in proximity are more similar than two points further apart and creates a gridded surface of predicted values using an empirical semivariogram model. As an advancement to kriging, EBK repeatedly simulates semivariogram models in subsets of up to 200 data points and thus not only improves the prediction but also optimizes interpolation across areas with strongly varying data availability in the Arctic Ocean (e.g., shelf seas vs. central basins).

Surface sediments show by far the largest data availability. The dataset of OC concentrations in CASCADE includes 4244 different locations across the Arctic Ocean (Fig. 2), while the concentration of TN and the OC / TN ratio are known for 2317 locations (Table 1). For carbon isotopes, the number of individual δ13C-OC values is 1555, and for Δ14C-OC it is 268. CASCADE also holds concentrations of terrigenous biomarkers at 131–213 locations per compound group. Most of the biomarker data are for HMW n-alkanes, with either concentrations of HMW n-alkanes (∑C21–C31; 213 stations) or chain lengths more specific for higher plants (∑C27, C29, C31; 164). Fewer data are available for concentrations of HMW n-alkanoic acids (∑C20-C30; 131) and the concentrations of lignin phenols (145).

In addition to surface sediments, a total number of 326 sediment cores (79 centennial, 229 millennial and 18 glacial cycle scale cores) are included in the first version of CASCADE. Combined, these hold another 10 552 observations of OC concentrations, 4769 concentrations of TN and 2122 δ13C–OC ratios in core samples from across the Arctic Ocean.

The data coverage for surface sediments is highly variable among the shelf seas, yet improved by the extensive gap-filling analysis (Table 1). The largest number of OC concentrations is in the Barents Sea (1092; Table 1). Despite the large total number of available Arctic sediment OC concentrations, there are only 236 samples analyzed for δ13C-OC and 33 with Δ14C-OC in the Barents Sea, and of these most are located in the Norwegian (western) sector of the Barents Sea. For the eastern Siberian Arctic and the North American sector of the Arctic Ocean, observations of OC concentrations are lower, but the availability of δ13C-OC data is higher (Table 1, Fig. 2b, c). Accordingly, the Kara, Laptev, East Siberian and Chukchi seas each support more than 200 δ13C-OC observations. The number of Δ14C-OC observations is generally lower but reveals the highest coverage in near-coastal areas, with 28 values in the Kara Sea, 42 values in the Laptev Sea and 71 values in the East Siberian Sea. Data availability in the Chukchi Sea for Δ14C-OC is lower (n=12), stressing the need for future analysis. The lowest availability of data is in the Canadian Arctic Archipelago. Gap-filling analysis of OC here increased the number of OC concentrations from 21 to 54, with a similar number for carbon isotopes (51 of δ13C-OC; 22 of Δ14C-OC) distributed over its vast area of 1 171 000 km2. The largest individual regime area is covered by the interior basins of the central Arctic Ocean, which holds 529 observations of OC concentrations, 130 of δ13C-OC and 27 of Δ14C-OC values.

Based on the quality assurance data available, CASCADE provides detailed information about the techniques involved in analyzing OC concentrations, isotopes and biomarkers. The development of CASCADE included the collection of meta-information about sampling, storage and analysis, as described in Sect. 2.6. This information is included and detailed in CASCADE. The quality assurance information shows that 86 % of the reported OC concentrations were analyzed using EA, and only a minority were analyzed by Rock-Eval pyrolysis. For δ13C-OC, in 66 % of the cases IRMS coupled to EA was reported as the method of analysis. Regarding sample storage, information was given in about 59 % of all data sources that the samples were kept frozen between sampling and analysis, while for < 1 % of the cases it was documented that the samples were stored refrigerated; this means that for 40 % of the samples, there was no information provided about sample storage. For 78 % of the Δ14C-OC values, the laboratory 14C/AMS label was documented and thus also added to the CASCADE sheet.

Visualization of CASCADE data directly reveals several large-scale features of OC in Arctic Ocean sediments. These include clear differences in both OC concentration and source-diagnostic isotope composition among the shelf seas. For instance, interpolated OC concentrations (Fig. 2) indicate that high sedimentary OC content is found both in regions of high terrestrial input (e.g., Kara Sea, Laptev Sea, East Siberian Sea and Beaufort Sea) and in regions of high nutrient availability and marine primary productivity (Barents Sea and Chukchi Sea). The combination of δ13C and Δ14C isotope values delineates large-scale differences in OC sources. Values of δ13C-OC close to marine OC (-21 ‰; Fry and Sherr, 1989) and Δ14C reflecting contemporary carbon are consistent with high marine primary productivity in the Barents Sea and Chukchi Sea. The Kara Sea receives input from major West Siberian catchments (Ob and Yenisey rivers), with sediment OC that appears to reflect OC from contemporary terrestrial sources (∼ -27 ‰; Fry and Sherr, 1989). By contrast, the terrigenous OC fraction in the Laptev and East Siberian seas is much older with a presumably substantial contribution from remobilization of thawing permafrost or other old deposits via erosional or fluvial processes (Figs. 1, 2). These and other features can now be investigated through CASCADE in greater quantitative detail over large intra- and inter-system scales.