River thorium concentrations can record bedrock fracture processes including some triggered by distant seismic events

Fractures at the base of the critical zone facilitate the entry of geogenic elements into rivers and control subsurface hydrology and biogeochemistry. While borehole and active geophysical methods increasingly characterize fracture distributions, they largely capture static structures. Passive seismic monitoring detects motions on faults, and surficial seismic studies have linked rock fracturing to environmental stresses, but few methods can detect changes to near-surface bedrock fractures. The authors hypothesize that transient changes in river thorium (Th) concentrations, captured by long-term, high-frequency water chemistry monitoring, can provide a chemical signature of subsurface fault or fracture events that affect fault-zone water and solute transport in mountainous shale-dominated watersheds. The study aims to test whether Th excursions in rivers reflect bedrock fracture processes, assess potential hydrologic or atmospheric drivers, and evaluate possible links to regional seismicity including dynamic earthquake triggering.

Prior studies have used borehole records and active geophysics to map fracture distributions in watersheds but primarily reveal static structures. Passive seismic methods have shown that earthquakes can alter flow pathways, permeability, and groundwater chemistry, and surficial seismic studies have tied rock fracturing to stress and weathering, illuminating environmental controls on rock failure. However, methods detecting near-surface bedrock fracture changes are scarce. Previous observations documented hydrogeochemical changes before and after major earthquakes and seismically enhanced solute fluxes, typically for larger-magnitude events. Work at critical zone observatories has linked concentration–discharge behavior to hydrologic processes and mineral weathering (e.g., pyrite vs. carbonate weathering) and documented particulate-associated transport with high temporal variance for certain elements. Collectively, these studies motivate searching for geochemical signatures, potentially including Th, that record dynamic fracture processes in mountainous watersheds where bedrock fractures strongly influence storage and transport.

Field setting and monitoring: The study focuses on neighboring mountainous catchments in the greater East River watershed, Colorado, including East River (Pump House, PH) and Coal Creek (Coal-11). Watersheds are underlain by late Cretaceous sedimentary rocks (Mancos Shale) with varying metamorphism and sulfide mineralization. High-frequency river chemistry: An extensive suite of solutes (major/trace elements) was monitored frequently (up to daily) at PH and Coal-11, and less frequently at additional sites. Time series included Th, As, Be, Zr, major cations/anions, DOC, turbidity, discharge, and meteorological data. Th excursions were identified as sudden increases followed by decay and fitted with biexponential functions (fast and slow time constants), with parameters summarized in Table S1. Event Coincidence Analysis (ECA): To test associations between Th excursions and potential drivers (precipitation, wind, seismic ground motion), the authors performed parametric ECA varying thresholds for precursor events and time windows (1–2.5 days). For Th events, a criterion of ≥0.05 µM (example ≥0.1 µM in supplementary illustration) increase between successive days was used; precipitation thresholds ranged 0.5–30 mm/day; seismic thresholds varied in ground velocity. Analytical p-values (valid for sparse events) were computed using established formulae and validated by Monte Carlo simulations. Groundwater sampling: Depth-resolved groundwater chemistry was measured in May–June 2017 along a hillslope transect (PLM wells) and in floodplain piezometers (near PH). Transient Th (and As) excursions at depth (>3 m) were compared with river Th to assess source and mixing/dilution effects. Laboratory leaching study: Unweathered and partially weathered Mancos shale (four locations) were powdered and leached for 48 h in simulated river water (10 mM NaCl, 1 mM Na2SO4, 1 mM NaNO3) under aerobic and anaerobic conditions. Solutions were prepared and handled to control redox (degassing, N2 sparging). Post-leaching suspensions were centrifuged/filtered and dissolved concentrations were measured. Analytical methods: ICP-MS (Elan DRC II) quantified Th, Zr, Be, As (MDLs: 0.009, 0.05, 0.02, 0.006 ppb; RSDs <10% for Th/Be, <5% for others); Ca and K by DRC ICP-MS (NH3 reaction gas; MDLs 2.2 and 0.4 ppb). DOC by TOC analyzer (NPOC, RSD <3%). Anions (NO3−, SO42−, Cl−) by ion chromatography (RSD <5%). Stream/porewater samples were field-filtered (0.45 µm PTFE), cation subsamples acidified (2% HNO3), shipped cold, stored at 3°C. Seismic and meteorological data: Seismic time series from regional stations (SMCO ~50 km, SDCO, MVCO, ISCO) were obtained from IRIS, converted to velocity using station scale factors; catalogued events identified via SeismicCanvas. Precipitation/wind from local weather stations (KCOMTCRE2; SNOTEL 380). GPS elevation data from UNAVCO were detrended for comparison. Concentration–discharge and detrended correlation analyses were performed in IgorPro. Data analysis: Time-series detrending for solutes with strong seasonality (e.g., Ca, SO42−) via linear spline; assessment of co-excursions (Th with As, Be), correlation checks with hydrologic and atmospheric variables, and comparison across sites for spatial coherence of events. Biexponential fits characterized Th decay constants following excursions.

- River Th showed abrupt, reproducible excursions followed by biexponential decay. At PH (East River), 22 episodes occurred over 20 months (2016–2018). Decay time constants: fast 0.38–1.6 days and slow 1.8–8.2 days; largest excursions persisted up to ~3 weeks. - Similar Th behavior occurred at Coal Creek and other sites; some excursions were temporally correlated across hydrologically connected and unconnected locations, indicating a geographically dispersed driver. - In East River, Th excursions coincided with As and Be excursions; correlations with other particulate-associated elements (e.g., Zr, Mn, Ti, Al) were absent, and those elements did not show exponential decay. - ECA found no statistical support for precipitation or wind as drivers of Th excursions across a wide parameter range (precipitation thresholds 0.5–30 mm/day; time windows 1–2.5 days). Turbidity excursions rarely coincided with Th (3 potential matches), arguing against landslide-driven flushing as a common cause. - Groundwater profiling revealed transient Th (and As) excursions at depth (>3 m) without corresponding near-surface increases. Maximum groundwater Th reached 1.8 ppb (7.8 nM), exceeding concurrent river concentrations (e.g., 0.07 ppb typical; 0.95 ppb maximum over 20 months), consistent with a bedrock source and dilution upon mixing to the river. - Laboratory leaching demonstrated Th release from Mancos shale under simulated conditions, with higher concentrations from deeper cores and under anaerobic relative to aerobic conditions; As was also released, Be not detected. - Many Th excursions lacked detectable seismic signatures at the nearest station (~50 km), implying aseismic fracture/fault processes can mobilize Th. Nonetheless, parametric ECA indicated a weak association between seismic ground motion and Th excursions (p-values ~0.12–0.25), lower than for precipitation, suggesting that a subset of events may be dynamically triggered by distant earthquakes (dynamic earthquake triggering), albeit without formal statistical significance. - Hydrologically driven seasonal weathering signals (e.g., SO42−, Ca) showed no abrupt changes coincident with Th excursions, and detrended data were uncorrelated with Th, further supporting a geomechanical origin.

The findings support the hypothesis that abrupt bedrock fracture processes inject detectable trace solutes into groundwater and streams. Processes such as fracture creation or slip and reorganization of subsurface hydraulic pathways can expose fresh mineral surfaces and/or open conduits, enhancing release and transport of Th and co-mobilized elements (e.g., As). The spatial coherence of Th excursions across neighboring watersheds suggests the involvement of regional fault zones that span streams and rivers. These fracture processes likely reflect combined topographic (gravitational) and tectonic stresses, potentially modulated by cyclical elevation changes, though detrended GPS elevation did not correlate with Th. Th mobility mechanisms remain uncertain. Although Th4+ is typically insoluble and sorptive, complexation by oxyanions such as sulfate—abundant due to sulfide weathering—could enhance solubility. Alternatively, part of the Th signal may be associated with fine particulate or colloidal transport, given the high variance observed; however, lack of association with classic particulate tracers (Zr, Mn, Ti, Al) and the characteristic biexponential decay distinguish Th behavior. Deeper groundwater peaks and higher concentrations relative to river water, together with lab leaching results (greater Th release under anaerobic conditions and from deeper cores), indicate a subsurface bedrock source and dilution to the river. While many Th excursions lack detectable regional seismic signatures, weak statistical associations with small ground motions from distant earthquakes suggest that dynamic triggering could initiate some fracture/hydraulic changes sufficient to mobilize Th, extending geochemical evidence for dynamic triggering into a new domain. The signal’s detectability appears favored in mountainous catchments where bedrock fractures dominate storage and flow; comparable long-term, high-frequency Th datasets are rare, limiting cross-site comparisons. Overall, Th time series provide a new geochemical window into fracture dynamics that often evade geophysical detection.

This study demonstrates that high-frequency river thorium measurements can record subsurface bedrock fracture processes in mountainous shale-dominated watersheds. Th excursions exhibit abrupt onsets and biexponential decays, are not explained by precipitation, wind, or seasonal hydrologic weathering trends, and are consistent with a subsurface bedrock source supported by groundwater profiling and laboratory leaching. Many events likely reflect aseismic fracture processes, and a subset may be linked to dynamic earthquake triggering by distant seismic waves, as suggested by weak but systematically lower ECA p-values for seismicity than for precipitation. Main contributions: (i) identification of Th as a chemical tracer of bedrock fracture dynamics; (ii) establishment of characteristic temporal signatures (dual time constants) of Th excursions; (iii) integration of river chemistry, groundwater profiling, laboratory leaching, and event coincidence analysis to constrain sources and drivers; (iv) preliminary geochemical evidence for dynamic earthquake triggering. Future directions include coordinated high-frequency water chemistry and dense local seismic monitoring to quantify thresholds for dynamic triggering, resolve the dissolved versus particulate nature and speciation of Th during excursions, assess the role of ligands such as sulfate, expand monitoring across diverse mountainous terrains and fault settings, and develop predictive frameworks linking fracture mechanics to geochemical signatures.

- Statistical association between seismicity and Th excursions did not reach conventional significance (lowest p ≈ 0.12), so causal links to dynamic triggering remain tentative. - Many Th excursions lacked detectable seismic signals at regional stations (~50 km), and local microseismicity may have gone undetected, limiting attribution. - Thorium speciation and transport mode (dissolved vs. colloidal/particulate) were not resolved; mechanistic controls on mobilization remain uncertain. - Monitoring network limitations: absence of ambient seismic landslide monitoring; one known landslide coincided with instrument failure, constraining evaluation of landslide contributions; turbidity data provided limited overlap with Th events. - Background metal variability in the mining-affected Coal Creek complicated cross-element correlations (e.g., As, Be), potentially obscuring patterns. - Generalizability is constrained by the need for long-term, high-frequency sampling; comparable datasets in similar geologic settings are scarce.