Characterizing spatial and temporal variation in ¹⁸O and ²H content of New Zealand river water for better understanding of hydrologic processes

Abstract

Time series of hydrogen and oxygen stable isotope ratios (δ 2 H and δ 18 O) in rivers can be used to quantify groundwater contributions to streamflow, and timescales of catchment storage. However, these isotope hydrology techniques rely on distinct spatial or temporal patterns of δ 2 H and δ 18 O within the hydrologic cycle. In New Zealand, lack of understanding of spatial and temporal patterns of δ 2 H and δ 18 O of river water hinders development of regional and national-scale hydrological models. We measured δ 2 H and δ 18 O monthly, together with river flow rates at 58 locations across New Zealand over a two-year period. Results show: (a) general patterns of decreasing δ 2 H and δ 18 O with increasing latitude were altered by New Zealand's major mountain ranges; δ 2 H and δ 18 O were distinctly lower in rivers fed from higher elevation catchments, and in eastern rain-shadow areas of both islands; (b) river water δ 2 H and δ 18 O values were partly controlled by local catchment characteristics (catchment slope, PET, catchment elevation, and upstream lake area) that influence evaporation processes; (c) regional differences in evaporation caused the slope of the river water line (i.e., the relationship between δ 2 H and δ 18 O in river water) for the (warmer) North Island to be lower than that of the (cooler, mountaindominated) South Island; (d) δ 2 H seasonal offsets (i.e., the difference between seasonal peak and mean values) for individual sites ranged from 0.50‰ to 5.07‰. Peak values of δ 18 O and δ 2 H were in late summer, but values peaked 1 month later at the South Island sites, likely due to greater snow-melt contributions to streamflow. Strong spatial differences in river water δ 2 H and δ 18 O caused by orographic rainfall effects and evaporation may inform studies of water mixing across landscapes. Generally distinct seasonal isotope cycles, despite the large catchment sizes of rivers studied, are encouraging for transit time analysis applications.

A method for predicting hydrogen and oxygen isotope distributions across a region's river network using reach-scale environmental attributes

“…Modelled river water isotope values were compared to annual average values from 58 sites from the New Zealand River water network (NRWQN; (Davies-Colley et al, 2011)) selected to represent catchments nationally (Yang et al, 2020). The residual calculations used in this study compared predicted flow-weighted annual average values to averages of measurements.…”

“…Measurements were flow-weighted monthly values at each site from April 2017-March 2020. Data from April 2017-March 2019 is published in Yang et al (2020). For the current analysis, an additional year of data (April 2019-March 2020) was collected from all NRWQN sites following the methods of (Yang et al, 2020).…”

“…Data from April 2017-March 2019 is published in Yang et al (2020). For the current analysis, an additional year of data (April 2019-March 2020) was collected from all NRWQN sites following the methods of (Yang et al, 2020). Briefly, samples were taken monthly from each of the 58 sites, and river flow rates were recorded at the time when samples were collected.…”

“…An initial list of over 30 potential independent variables for the regression model were selected from the REC and FENZ geodatabases based on our understanding of the hydrological processes which potentially influence stream isotope components, including geological factors (e.g. slope, elevation, aspect, drainage density and lake coverage), climate factors (e.g., precipitation and evaporation), and land cover (Yang et al, 2020;Yang et al, 2019). 160 Table S1 includes descriptions of these variables; we use abbreviations hereafter as described in the table.…”

“…This validation dataset contains largely synoptic sampling data; due to a paucity of annual average values in the validation dataset we included unpublished δ 2 H and δ 18 O data from 9 river sites for which regional government staff have conducted monthly sampling over date ranges between 2017 and 2020. Water samples from 180 these sites were stored and analysed following the methods of Yang et al (2020) Baisden et al (2016), showing improved monthly fits, but a poorer longer-term fit for δ 18 O. This is likely due to updates to the climate network data used for VCSN model input, and updates to 200 the interpolation scheme as described above.…”