Measurement of CO2 and CH4 dissolved from urban rivers and canals in London in September 2021 using a global hydrological-environmental database
The carbon dioxide that is breathed into the atmosphere by the rivers and streams is the second-largest carbon dioxide transfer from an environmental system. Riverine CO2 is thought to come from atmospheric carbon that has been fixed by plants. Such carbon is cycled swiftly into soils as litter and root detritus, broken down by microorganisms into smaller organic fragments and CO2, and ultimately carried by water through the ‘veins’ of the landscape into rivers3–5. A paper by Dean et al.6 in Nature shows that much of the CO2 escaping from the rivers today comes from carbon dating back thousands of years. This source was thought previously to have been buried in Earth’s deep carbon reservoirs.
Eight dissolved CO2 and three dissolved CH4 samples were collected for 14C analysis from a range of urban rivers and canals in London in September 2021. CO2 samples were collected from the Regent’s Canal and the River Thames, while CH4 samples were collected from Bow Creek. The CO2 was collected using the method of equilibrating 3 liters of water with 1 l of CO2-free headspace and then injected into a molecular sieve for transport to the NEIF Radiocarbon Laboratory. The CH4 samples were collected using a coiled membrane method in which water was being pumped through a gas-permeable surface into a headspace containing ambient air. After recovering CH4 from the ground, the vessel left to collect the headspace in foil gas bags and transport it to the NEIF Radiocarbon Laboratory. CH4 samples were corrected for the ambient air in the headspace following refs. 21,52.
Owing to inconsistencies in how catchment characteristics were reported in the literature from which we assembled the radiocarbon measurements (if they were reported at all), we used a global database of hydrological-environmental characteristics (HydroATLAS56) to extract key catchment information for each sampling location in a consistent manner (for example, catchment size, lithology, biome).
Tropical grasslands and shrublands, which included HydroATLAS biomes ‘7. Tropical & Subtropical Grasslands, Savannas & Shrublands’ and ‘9. Flooded Grasslands & Savannas’ (which occur mostly in tropical regions62).
A New Radiocarbon Age and Uncertainty from a More Than One Repetition-Sample Sample in Fraction Modern Atmosphere
in which F14Catm is the normalized F14C value of the sample (Fmsample in fraction modern) divided by the F14C value of the atmosphere in the year of sampling (Fmatmosphere in fraction modern)35.
The locations in the database were visited more than once. The combination of experimental approaches include repetition sampling, exploration of temporal variations, and method development. When a sample location was repeated four times in a calendar year, we took average F14C observations from that location and created a new radiocarbon age and uncertainty. This removal left n = 1,020 observations (Extended Data Table 1).
Source: Old carbon routed from land to the atmosphere by global river systems
Measurements of river water DIC concentration in Taiwanese rivers and the Mekong River using 1-l bottles and a Niskin-type sampler
River water DIC samples from Taiwanese rivers and the Mekong River in Cambodia were collected using the methods outlined in refs. 18,58. The Taiwanese rivers had 1-l bottles submerged in the middle of the channel with a weighted Teflon sampler. In the Mekong River, near-surface samples were collected using a Niskin-type sampler. Preweighed 1-l foil bags were used to filtrate river water and avoid any atmospheric air mixing. The foil bag was filled with approximately 200–500 ml of filtered river water (depending on expected DIC concentration) and then gently squeezed before closing to ensure that no air was trapped. The filled bag was reweighed and stored at 4 °C during fieldwork, before shipping to the UK, in which the sample was frozen within about a week of collection58.
Sedimentary, which included the HydroATLAS classes ‘1. Unconsolidated Sediments (SU)’, ‘3. The Siliciclastic Sedimentary Rocks (RSS) were recorded in the 5th century. The SM and 6 are mixed rocks. Carbonate Sedimentary Rocks (SC)’.
One data point was excluded from the analysis because they had no data from the HydroATLAS classes. It was not possible to return ‘Ice and Glaciers (IGLAS)’ from the Hydro ATLAS classes due to a lack of data.
Source: Old carbon routed from land to the atmosphere by global river systems
Modelling River Carbon Emissions from HydroATLAS Data Using Nonparametric Random Forests with Conover-Iman Tests and Wilcoxon Tests
R version 4. 1.1 contains statistical analyses. We used nonparametric Kruskal–Wallis tests with the kruskal.test function in R, supplemented by post hoc analyses consisting of Conover–Iman tests using the conover.test function and unpaired two-sample Wilcoxon tests using the wilcox.test function. We undertook linear regression analyses using the lm function. The data for each analysis is provided in the Figures in the main text, extended data and supplementary information.
The potential drivers of the age of river carbon emissions were explored. A machine learning model that integrates many regression trees is called a random forest. This approach has proved to be successful for unraveling the interplay among variables64,65,26 in numerous environmental studies due to its capacity to capture nonlinear relationships and mitigate risk of data overfitting. In this study, we use random forest models to investigate the relationships between key catchment characteristics extracted from HydroATLAS and F14Catm values in the database. We aimed to identify which variables have the strongest control on F14Catm of river carbon emissions.
To select the input variables for the model, we first removed variables that correlated significantly with other potential input variables based on a Spearman correlation greater than 0.6 to avoid the results being influenced by correlated input variables. The year of sample collection is shown in Supplementary Table 5.
We divide the model into four parts: the model runs by the catchment size, the reach characteristics for rivers with 10 km2 or more, and the total area of the river. The model was only applied to the DIC data, which had no more than 100 observations, because of limits on the number of points.
Source: Old carbon routed from land to the atmosphere by global river systems
The association between F14Catm predictor variables and the random forest model: partial dependence analysis and a simplified computation of the total of river, ocean, and CO2 emissions
The association between predictor variables and F14Catm was assessed using the pdp R package67. The partial dependence plots show how F14Catm changes when a given input variable (Supplementary Table 5) varies but all other variables are held constant in the random forest model. The mean values from the ten runs were plotted by the shaded area and we used a partial dependence analysis to mirror the random forest models.
Using the mean F14C value for DIC, CO2 and CH4 across all rivers in our database of F14Criver = 0.919 (Extended Data Table 1) and subtracting petrogenic C inputs (0.150–0.218 Pg C year−1) from the sum of lateral DIC export to the ocean and vertical river CO2 emissions (2.5 ± 0.4 Pg C year−1), we can simplify equation (7) to:
The total of the river, DIC, and ocean is referred to as startarraylrmTotal.
$$\begin{array}{l}{F}^{14}{{\rm{C}}}{{\rm{river}}}\,\times \,2.5={F}^{14}{{\rm{C}}}{{\rm{decadal}}+{\rm{millennial}}}\,\times \,(2.28\,{\rm{to}}\,2.35)\ \,\,\,\,\,\,\,+\,{F}^{14}{{\rm{C}}}_{{\rm{petro}}}\times (0.15\,{\rm{to}}\,0.218)\end{array}$$
If the source is assumed to have no radiocarbon, we can calculate the non-petrogenic F14C value. The F14Cdecadal+millennial is the number provided by this estimate.
We were not able to collect site-specific concentration and emission data from the literature to match F14C data. This means that we were not able to scale the F14C values in the database with local and regional emission fluxes (Supplementary Information section 4).