Dataset of 18O and 2H in streamflow across Canada: A national resource for tracing water sources, water balance and predictive modelling

Oxygen-18 and deuterium were measured in streamflow samples collected from 331 gauging stations across Canada during 2013 to 2019. This dataset includes 9206 isotopic analyses made on 4603 individual water samples, and an additional 1259 analysis repeats for quality assurance/quality control. We also include arithmetic and flow-weighted averages, and other basic statistics for stations where adequate data were available. Station data are provided including station code, name, province, latitude, longitude and drainage area. Flow data were extracted from the historical database of the Water Survey of Canada. Details on the preliminary application of these data are provided in “18O and 2H in streamflow across Canada” [1]. Overall, these data are expected to be useful when combined with precipitation datasets and analytical or numerical models for water resource management and planning, including tracing streamflow source, water balance, evapotranspiration partitioning, residence time analysis, and early detection of climate and land use changes in Canada.

model prediction can potentially advance representation of hydrological processes and pathways, forecasting of water availability including floods and droughts, and as an early warning of hydrologic changes occurring due to land use impacts and climate change.

Data Description
Analytical isotope data for 18 O and 2 H are provided for individual samples collected during the survey at 331 stations ( Fig. 1 ) operated by Water Survey of Canada during regular site visits by staff. Data include raw analytical results and routine repeats, mean sample results, station arithmetic means, analytical uncertainty, and lab and field comments. In addition, flow-weighted values for 18 O and 2 H for 161 stations, as well as drainage basin areas (km 2 ), mean flow (m 3 ·s −1 ), water yield (mm ·yr −1 ) and station elevation (m.a.s.l.) are also provided ( mmc1.xlsx ). Summaries of isotope results by station (georeferenced) are provided for each province/territory ( Tables 1-13 ).

Network design
331 stations were selected in cooperation with WSC to establish full national coverage and to incorporate a range of watershed scales for various stakeholder applications. The temporal       ( continued on next page )       frequency of sampling ranged from 1 to 12 times per year, with schedules constrained by the frequency of visits by technical staff. A core network of 161 stations had more than 12 samples collected for isotopic analysis during 2013-2019 and had available discharge data to enable meaningful flow weighting of the isotopic signatures. Overall, the network included watersheds ranging in size from 100 to greater than 10,0 0 0 km 2 and situated across 90 °longitude and 25 °l atitude.

Field methods and sampling
Water samples were collected from mid-channel locations at mid-depth within the river water column at each gauging station to ensure representativeness of discharge. If this was not possible due to ice conditions or other safety issues, staff typically collected samples from an adjacent bank, avoiding poorly mixed zones below tributaries and/or backwater areas. Alternately, water was collected during ice-on conditions from a borehole augered in the ice. Metadata such as station number, sampling date, time, backwater effects, and ice conditions were routinely recorded. Water samples were collected in 30 mL high density polyethylene (HDPE) bottles with tightly sealed lids to prevent evaporation, and stored at room temperature prior to and during shipment to the lab in Victoria. Freezing of samples was avoided. Sample bottles were labelled in the field with station number, date, time, ice conditions, and name of sampler, as well as any applicable comments. A list of samples was also generally provided to InnoTech Alberta by field staff.
Locations for sampling were selected in consultation with the Water Survey staff from the network of active monitoring stations, based on considerations such as coverage, basin size, and anticipated applications for regional stakeholders such as Indigenous groups, hydroelectric generating plants, oil sands and mining operators, recreation, and municipal, provincial and national governments. In general, water samples were collected either as mid-depth, mid-channel grab samples, or dip samples from shore, bridges or cableways. In the case of poor mixing conditions or difficult access, sites were reviewed and a specific plan was developed with water technicians familiar with each site. Locations were selected to maximize the potential for obtaining a representative sample based on knowledge of the mixing conditions of the river at each station, and their seasonal variability including differences during ice-on, melt periods, and ice-break-up conditions. Locations downstream from major tributaries were avoided where possible due to potential for incomplete lateral mixing. By coordinating water isotope sampling with WSC gauge locations where discharge is measured, we sought to develop a dataset where both discharge and isotopic composition was known to allow for flow weighting of the isotopic signatures. This is often the preferred method for obtaining long-term isotope values for continental runoff, although we evaluate the significance of this approach later by comparing flow-weighting and simple averaging.

Laboratory analysis
All isotope results were analyzed by isotope ratio mass spectrometry using a Thermo Scientific Delta V Advantage located at InnoTech, Victoria. Oxygen was prepared using a Gasbench II by equilibrating water and CO 2 and then introducing CO 2 into the dual inlet using an autosampler [21] . Hydrogen was analyzed by auto-injecting water into a chromium reactor heated to 875 °C in the HDevice to produce H 2 , which was streamed to the dual inlet for analysis [22] . In all cases, analyses were made within one year of sample collection as confirmed to be appropriate for HDPE bottles [23] . Results are reported in "δ" notation in permil ( ‰ ) relative to Vienna Standard Mean Ocean Water (V-SMOW) and normalized to the SMOW-SLAP scale where SLAP is Standard Light Arctic Precipitation [24] . Analytical uncertainty estimated based on 2-σ of repeats is ±0.14 for δ 18 O ( n = 382) and ±0.52 for δ 2 H ( n = 450).

Ethics Statement
We hereby assert that the manuscript is the authors' own original work, which has not been previously published elsewhere, nor is it currently being considered elsewhere for publication. The paper reflects the authors' own research and analysis in a truthful and complete manner, and properly credits the meaningful contributions of co-authors and co-researchers. We have sought to appropriately place the results in the context of prior and existing research and have properly disclosed all sources. All authors have contributed to the work and will take public responsibility for its content.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships which have or could be perceived to have influenced the work reported in this article.