Weathering Dynamics in Watersheds: Connecting River Chemistry to Subsurface Mineral Distributions
Open Access
- Author:
- Shaughnessy, Andrew
- Graduate Program:
- Geosciences
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- September 29, 2022
- Committee Members:
- Donald Fisher, Program Head/Chair
Susan Brantley, Chair & Dissertation Advisor
Matthew Fantle, Major Field Member
James Kasting, Major Field Member
Shujie Wang, Outside Unit & Field Member - Keywords:
- Weathering
Watersheds Science
Hydrology
Geochemistry
Pyrite
Data Science - Abstract:
- Mineral weathering significantly contributes to global biogeochemical cycles and is hypothesized to control Earth’s long-term climate through release and removal of CO2 to and from the atmosphere. Although it is known that underlying geology influences water quality and weathering, few studies consider the subsurface beneath a watershed with respect to weathering dynamics. This is because in most watersheds, subsurface mineral distributions remain largely uncharacterized. By exploring river chemistry in comparison to the subsurface architecture of a watershed, I sought insights into weathering mechanisms for different minerals. For a given mineral weathering reaction, it is often assumed that one step limits the rate, dictating the weathering regime in that watershed. Rate limiting steps such as water throughput (the regime of runoff limitation), delivery of weathering reagents other than H2O (advective transport limitation), supply of fresh minerals (erosive transport limitation), and interfacial reaction kinetics (kinetic limitation) have been considered. But how the dominant weathering regimes change across spatial and temporal scales or for different compositional systems is generally unknown. In this dissertation I explore how weathering fluxes in rivers vary with discharge, watershed size, and climate to evaluate weathering regimes for a few compositional systems. I link this variability to subsurface mineral distributions through borehole observations, multivariate statistics, physically based models, and data driven techniques. The first mineral whose weathering I assess is pyrite – a ubiquitous iron sulfide mineral that releases much of the sulfate in rivers that flows to the ocean. When pyrite oxidizes, it releases sulfuric acid, dissolving minerals in proximity and often releasing CO2. Additionally, some of the oxidant (O2) can be incorporated into the sulfate byproduct from pyrite oxidation. Oxygen isotopes in sulfate deposits could, therefore, help to reconstruct Earth’s atmospheric oxygen history; however, we need a better understanding of modern pyrite oxidation mechanisms to accurately interpret historic or long-term geologic oxidation dynamics. In my first study I developed a machine learning-based mixing model to discern the signal of pyrite oxidation from other sources of sulfate in rivers. This model leverages the fact that different minerals dissolve along deep and shallow flowpaths, leading to unique chemical signatures that can be detected in streams. I apply this model to three extensively studied watersheds to investigate modern and long-term weathering-CO2 dynamics. I found that watersheds intermittently switch between acting as CO2 sources and CO2 sinks depending on whether the dominant flowpath sampled by the stream is deep or shallow, respectively. Comparing soil chemistry to river chemistry, I found that long-term CO2 dynamics (recorded in soils) are roughly consistent with short-term CO2 dynamics (recorded in streams). Acid rain at the three sites competes with CO2 as a weathering agent and has reduced the short-term capacity of the watersheds to sequester CO2. Only recently have the watersheds begun recovering from the impacts of acid rain. From my first study, I observed that pyrite-derived sulfate concentrations changed with time, and that concentrations were highest in the dry season and lowest in the wet season. To explore these temporal dynamics, I coupled the machine learning-based mixing model to a physically based-oxidation model that describes how pyrite-derived sulfate concentrations vary with discharge. Utilizing concentration and discharge data from 15 watersheds in the northeastern United States and ~300 watersheds in the western United States, I fit the oxidation model using a non-linear least squares regression. From these modeling results, I found that, in the northeastern United States, concentrations vary little with discharge when coal mining is absent within the drainage area, and that concentrations decrease with discharge when coal is mined in the basin. I interpreted this trend as a shift in the rate limiting step of oxidation from advection of dissolved oxygen to interfacial reaction kinetics as a result of mining-related exhumation of pyrite. Coal mining and spatial scale co-vary in the Susquehanna River Basin of the northeastern United States (i.e., only small watersheds contain no coal); therefore, it is difficult to disentangle their effects on concentration-discharge (C-q) dynamics. In the western United States, I was able to deconvolve some of these relationships by focusing on watersheds without agriculture and with low inputs of acid rain. I found that C-q behaviors are consistent with the pyrite reaction front becoming shallower as drainage areas of watersheds increase. This could indicate that pyrite-derived sulfate fluxes in small watersheds tend to be limited by transport of reagent (short timescales) or supply of fresh mineral (long timescales). In the first two studies, I focused on pyrite weathering, which shows relatively fast reaction kinetics. As a comparison, I sought to explore the other end of the weathering spectrum, namely slow-dissolving silicate minerals. For my third study, I aggregated data for water quality, discharge, climate, and catchment characteristics for 148 shale-underlain watersheds across the United States to explore the rate limiting step of silicate weathering. Shales cover a significant proportion of global land area, but most weathering studies focus on granitic and basaltic systems. I found that when potential evapotranspiration (PET) is much larger than mean annual precipitation (MAP), weathering is limited by the throughput of water in the system (i.e., runoff). For humid watersheds, I found that weathering rates show a dependence on erosion rate, temperature, and precipitation rate, which is consistent with a mixture of weathering regimes and rate limiting steps. When watersheds are small and characterized by steep slopes on average, the silicate weathering rate is only dependent on temperature and precipitation. When watersheds are large and are characterized on average by gentle slopes, the weathering rate is predominantly dependent upon the erosion rate. I quantify the weathering dependencies for different groups of watersheds via regression analysis. For US shale-underlain watersheds, 48%, 37%, and 7% of their land area is limited by runoff, erosion, and reaction kinetics, respectively. The remaining 9% represent a mixed weathering regime. Weighting these land fractions by the average fluxes observed for each weathering regime, I found that 17%, 56%, 11%, and 16% of silicate weathering fluxes from shale watersheds in the US are limited by runoff, erosion, reaction kinetics, or a mixture, respectively. Overall, the results presented in this dissertation shed new light on the impact of subsurface mineral distributions on weathering dynamics in watersheds. I present a new interpretive framework for evaluating the CO2 sequestration efficiency of weathering using both stream and soil chemistry. This has implications for evaluating CO2-weathering dynamics over geologically relevant timescales. I also present a new method for evaluating C-q relationships with respect to depths of groundwater flowpaths and reactive minerals. This has implications for predicting legacy contamination and the transport of redox-sensitive species. Lastly, I evaluate and quantify the silicate weathering dependencies in shale watersheds. I classify weathering regimes in these watersheds based on climatic and geomorphologic characteristics of a watershed. Results in this dissertation can improve future hydrologic and geochemical models through integrating subsurface mineral distributions across scales.