Purpose and Scope

The purpose of this study was to monitor the quantity and quality of suspended sediments transported in the Big River downstream from the OLB as reclamation activities in the drainage basin progressed. Suspended sediment and trace element data were collected during water years 2019–21 (October 1, 2018, through September 30, 2021; phase two) downstream from the OLB at the Big River below Bonne Terre, Missouri (USGS site number 07017610; hereinafter referred to as the study site; table 1). This report describes the methods and techniques used in the collection and analysis of water quality samples during water years 2019–21 and presents the results and interpretations of that data. Comparisons of these data to results from phase one (water years 2012–13; Barr, 2016) also are presented to determine whether trace element concentrations have changed significantly in the Big River drainage basin in the OLB after major mine waste area remediation efforts.

Description of Study Area

The Big River drainage basin drains 955 square miles (mi²) of southeast Missouri (fig. 1) and is a tributary to the Meramec River. Land use within the Big River drainage basin is primarily forest and pasture, with some agriculture and urban land uses (Missouri Department of Conservation, 1997). The Big River headwaters are northeast of the city of Ironton, Iron County (not shown), where the river is as narrow as 50 feet (ft). The rugged topography and steep drainage divides create a streambed composed of mainly cobble, gravel, and sand. As the river meanders north towards the confluence with the Meramec River in Jefferson County, it widens to nearly 100 ft in some locations. The downstream topography transitions to floodplains, and the streambed changes to sand and silt (Barr, 2016).

The study site was chosen for consistency with phase one (Barr, 2016). The site is approximately 3 mi north of the city of Bonne Terre and is at the bridge crossing at County Highway E (fig. 1, Highway E not shown; table 1). The orifice line for the pressure transducer and continuous water-quality monitor is housed underneath the bridge and on the left riverbank (facing downstream). The site drains 409 mi², and the river channel is 2–3 ft deep and as much as 60 ft wide during base-flow conditions. The riverbed is composed of fine sands, gravels, cobbles, and some boulders (Barr, 2016).

Continuous Stage and Water-Quality Data Collection

During the entire study period (water years 2012–21 [October 1, 2011, through September 30, 2021]), the sampling site was equipped with a streamgaging station to measure streamflow and continuous water quality. Measurements of stage, temperature, and turbidity were recorded every 15 minutes although some periods are missing because of intermittent funding, equipment failures, low-flow conditions, biofouling, and (or) data corrections that exceeded calibration criteria. Continuous data collected during this study are available through the USGS National Water Information System (NWIS; USGS, 2023) database. Streamflow measurements were obtained every 4–6 weeks and during selected stormflow events using an acoustic Doppler current profiler. This information was used to maintain the stage-discharge relation following standard USGS methods and techniques (Rantz and others, 1982; Oberg and others, 2005; Turnipseed and Sauer, 2010; Mueller and others, 2013). Water temperature and turbidity were measured in phase one using a Yellow Springs Instruments (YSI) 6-series sonde and then upgraded in 2015 to an EXO2 sonde, which includes a central wiper. The temperature probe measures degrees Celsius (°C) and is calibrated using a three-point calibration with a National Institute of Standards and Technology thermometer. The turbidity probe measures formazin nephelometric units (FNU) and is calibrated using a three-point calibration and YSI’s calibration standards (Wagner and others, 2006). The YSI EXO2 was serviced approximately every 8–12 weeks, which included cleaning the sonde and calibration checks on the turbidity sensor. The temperature sensor calibration was checked approximately every 12 months.

The streamgage was equipped with a data collection platform that stored data from the pressure transducer that measured stage and the sonde that measured water temperature and turbidity. Data were transmitted hourly via satellite telemetry. The sonde was deployed using a polyvinyl chloride (PVC) pipe with several 1-inch (in.) holes drilled in it. Water from the river was able to flow through the pipe and around the sensors reducing issues associated with measuring stagnant water in the pipe.

Discrete Suspended Sediment Sample Collection

Traditional cross-sectional sampling methods were used to collect suspended sediment samples during base-flow and high flow conditions. During base-flow conditions, suspended sediment samples were collected approximately 40 ft upstream from the bridge by wading the stream and using the equal-width increment (EWI) method (USGS, 2006). The EWI method involves sampling at a minimum of 10 equally spaced locations (verticals) across the full width of the stream channel. A US DH-81, a depth-integrated and isokinetic sampler (a sampler in which stream water approaching and entering the sampling nozzle have the same velocity) was used. The US DH–81 consists of a 1-liter (L) sampling bottle with a 1/4-in. Teflon or polypropylene nozzle mounted on the end of a sampling rod. Samples were collected at each vertical with a consistent transit rate (vertical speed). The samples at each vertical were composited into one or more sampling bottles. If the minimum mean stream velocity was less than 1.5 feet per second (ft/s), the sample was collected using grab sample methodology because of the lack of isokinetic conditions (USGS, 2006). The equipment used are described in Wilde and others (2014). Streamflow measurements were collected in the same location as the sediment sample.

During stormflow events, measurements and samples were collected from the downstream side of the bridge deck on Highway E using the same EWI method. Samples were collected across the hydrologic event, targeting the rising limb, peak, and falling limb when possible. Streamflow was measured nearest to the peak as possible. Depending on stream conditions, either a US DH–95 or US DH–2 sampler was used. USGS personnel operated the sampler, which was attached to a reel and cable mechanism mounted on a vehicle. The sampler was raised and lowered at a constant speed according to the transit rate, which is calculated from the maximum stream velocity and depth. It should be noted that an unsampled zone exists when using these samplers (4 in. for the DH–81 and DH–2 and 4.8 in. for the DH–95; Davis, 2005; Wilde and others, 2014). The unsampled zone includes the area closest to the streambed that cannot be reached by the nozzle; this area can transport a higher concentration of coarse sediment than the rest of the water column. This part of unsampled suspended sediment may or may not account for a substantial portion of the total suspended sediment and is dependent upon stream velocity, depth, and turbulence (Edwards and Glysson, 1999).

Passive Suspended Sediment Sample Collection

Suspended sediment samples were collected passively during the duration of high-flow events using Walling tube samplers (Phillips and others, 2000). Passive sampling is ideal in situations when sampling teams cannot be onsite to collect suspended sediment. In some instances, smaller hydrologic events can happen quickly and without enough time to prepare for physical sampling. A Walling tube sampler can be set at a fixed height, and when inundated by the stream during an event, sediments are trapped inside as the velocity slows when water moves through the tube. Placement of the samplers at or near a streamgage allows for sampling a part of a hydrologic event. Samples collected by passive samplers are considered a general representation of a hydrologic event.

The sampling tubes were constructed from 1 m in length schedule 40 PVC well casings with an inner diameter of 98 millimeters (mm). A conical head is installed on the receiving end of the tube (facing upstream) that threads into the main tube body. A hole was drilled into the end of the conical head and threaded to accept a 1/4-in. USGS D–96 or D–74 sampler nozzle. A 4-in. PVC cleanout with a threaded plug was glued to the other end of the tube, and a 1/4-in. diameter hole was drilled in the center of the plug. Two samplers were secured to metal posts on the left stream bank (nozzles facing upstream) several feet above base-flow stage such that the lower sampler became inundated at a stage of about 4.82 ft (streamflow of approximately 445 cubic feet per second [ft³/s], per rating curve [relation between river stage flow] current for the deployment period) and the upper sampler was inundated at a stage of approximately 5.52 ft (streamflow of approximately 715 ft³/s).

If an event happened that would have inundated the samplers, a service visit was done to collect any suspended sediment trapped in the tube during the event(s). During servicing, the nozzle was removed and both openings on the sampler tubes were temporarily plugged before removing the samplers from the mounting posts and rocking them slowly to resuspend sediment within the tube. The tube was then held vertically, and the sediment slurry allowed to drain out the conical end into a clean 1- or 2-gallon plastic container. The plug on the rear of the tube was removed, and the inside of the tube was rinsed with 200–300 milliliters (mL) of native stream water using a small plastic 1-L sprayer. Sediment from the two samplers was composited before submission to the Central Midwest Water Science Center (CMWSC) Sediment Laboratory in Rolla, Mo., for further processing.

Suspended Sediment Laboratory Methods

All suspended sediment samples collected during this study were sent to the USGS CMWSC Sediment Laboratory in Rolla, Mo., for particle-size analyses, and all results are available via NWIS (USGS, 2023). Routine discrete samples were analyzed only for SSC using a filtration method (Guy, 1969). Discrete and passive stormflow event samples were analyzed for SSC and particle-size distribution of sands (2–0.063 mm) and fines (less than [<] 0.063 mm). All suspended sediment samples received at the laboratory were allowed to settle before excess water was decanted. Discrete stormflow event samples were analyzed for particle-size distribution using the wet-sieve method (Guy, 1969). Nylon sieves were used to reduce trace element contamination that could be introduced from the traditional brass sieves. Each portion was placed into separate glass dishes to dry at 80 °C and then for an hour at 103 °C (Guy, 1969). Passive sample sediments were analyzed for particle-size distribution by the dry-sieve method (Guy, 1969) also using nylon sieves after being air-dried and disaggregated.

Dried parts of sands and fines from discrete and passive samples were retained after analysis to be used for trace element analyses. The dried fractions of sediments from the passive samplers were split into two equal parts using a stainless-steel universal sample splitter. One portion was submitted for trace element analysis to the USGS Mineral Research Laboratory (MRL) in Denver, Colorado, and the other portion was retained and archived. Any discrete or passive stormflow event sample that had a mass of sands or fines larger than 0.5 gram was sent for trace element analysis on the individual fraction. If any dried size fraction did not meet the minimum mass requirement for trace element analysis, the two fractions were recombined and submitted as a bulk sample (sediment <2 mm in size). Samples were shipped in sealed plastic bags. Submitted samples were analyzed for 49 major and trace elements by inductively coupled plasma mass-spectrometry (Taggart, 2002). The results for the trace elements barium, cadmium, lead, and zinc are presented in this report. Results for the additional major and trace elements are available from the NWIS database (USGS, 2023) or Markland and others (2024).

Quality Assurance and Quality Control

Quality assurance (QA) and quality control protocols were followed during this study to ensure data met USGS quality standards. A QA plan, which is summarized here, was created to document techniques and methods. The study site was previously selected by Barr (2016) to reduce fouling, presence of low-flow conditions, and damage from flooding or vandalism. Data transmissions to NWIS were reviewed daily for erroneous data. All site visit and sampling information, including a description of tasks completed and additional notes, was documented and archived. Laboratory analysis request forms were used for the sediment laboratory and MRL for tracking and documentation purposes.

Streamflow and stage data from the streamgage were used to monitor and predict the need for stormflow event sampling and servicing of the continuous water-quality equipment. Calibration and maintenance of equipment was completed following USGS methods described by Wagner and others (2006). During each visit, equipment was cleaned and checked for damage and calibration drift. Corrections for calibration drift, fouling corrections, or both were applied to data as needed. If corrections exceeded USGS criteria, the erroneous data were removed from the record (Wagner and others, 2006).

Precautions were taken to minimize contamination of samples from sampling equipment, vehicles, and surroundings. The “clean hands/dirty hands” sampling technique was followed during sediment sampling (USGS, 2006). Sample containers were kept sealed and protected in a large plastic bag to prevent contamination by airborne debris introduced by wind, vehicle traffic, sampling equipment, bridge railings, and precipitation.

To identify and quantify any bias and variability associated with the sampling and processing techniques, quality control samples were collected in the form of field replicates, laboratory replicates, equipment blanks, and analysis of standard reference samples of known chemistry. Field replicate samples for the passive samples were collected by treating each Walling tube as a separate sample for selected periods, with one sample assigned as the regular sample and the other as the field replicate. Field blanks were not collected; however, pure silica and silica powder were used to prepare a source blank “sediment,” which was submitted to the USGS CMWSC Sediment Laboratory in Rolla, Mo., and to the MRL. In the sediment laboratory, the silica blanks were dried, sieved, and split in an identical manner to the environmental samples. A sample of the pure silica powder also was submitted as a source blank. Splits from the silica blanks were submitted to the MRL for chemical analysis. All USGS sediment laboratories undergo biannual QA tests to maintain consistency for all laboratories. Results from the QA tests are available from the USGS Branch of Quality Systems at https://bqs.usgs.gov/SLQA/. The MRL splits 10 percent of samples into split replicates and submits samples of known composition with each batch of samples submitted to the contract laboratory. Additional information on the methods used at the MRL can be found at https://minerals.cr.usgs.gov/. The reporting limits for laboratories used during the study are provided in table 2.

Each batch of samples analyzed at the MRL contract laboratory included the analysis of a standard reference sample with preferred/expected concentrations for barium, cadmium, lead, and zinc. The goal of the MRL program is for values in the reference samples to be reported within 15 percent of the preferred/expected value (recovery between 85 and 115 percent). The mean recoveries for the four trace elements ranged from 93.0 to 106 percent (table 3). It was noted that the analytical batch associated with most of the analyzed passive samples had reported maximum recoveries of barium and lead in the reference samples that were 117 and 129 percent of the expected/preferred values, respectively (table 3).

The results of the replicate passive samples (n=9) had barium, cadmium, lead, and zinc relative percent differences of 5.2, 7.4, 6.0, and 4.9 percent, respectively. The results from the silica source blanks (n=12) that were processed in an identical manner to the environmental samples had average barium, cadmium, lead, and zinc concentrations of 12.6, <0.02, 6.1, and 5.3 mg/kg, respectively. The range of concentrations in the blanks were 8–23 mg/kg for barium, <0.02–0.03 mg/kg for cadmium, 1.4–34.5 mg/kg for lead, and 2.6–9.2 mg/kg for zinc. The lead concentration of 34.5 mg/kg was anomalous, as the lead concentration in the other 11 silica blanks sample results were less than 8 mg/kg. The results of the pure silica source material that was not processed but was submitted directly to the MRL for chemical analysis (n=3) had mean barium, cadmium, lead, and zinc concentrations of 10.7, 0.1, 12.6, and 10.9 mg/kg, respectively (Markland and others, 2024).

Data Analysis and Reporting

Streamflow, water temperature, and turbidity were recorded every 15 minutes during the study period, and these measurements were used to calculate daily mean, monthly mean, and annual mean values. Additionally, SSC and SSL data were used to compute daily, monthly, and annual mean values. Statistics for all daily data collected by USGS during the study period are available from the NWIS database (USGS, 2023). Monthly precipitation information used in this report was gathered from the National Oceanic and Atmospheric Administration (NOAA) and are cumulative monthly precipitation measurements (NOAA, 2022).

This study used turbidity, streamflow, and discrete SSC data from stormflow event and routine base-flow samples to develop regression models to estimate continuous (15-minute) SSCs and SSLs. Because the magnitude of turbidity in riverine systems is typically proportional to the SSC (Rasmussen and others, 2009), continuous turbidity measurements can be used to develop a regression model between turbidity and discrete SSC samples collected during base-flow conditions and stormflow events. Collection of suspended sediment samples across a range of hydrologic conditions increases the robustness of the regression models and confidence in the results (Rasmussen and others, 2009). Outliers were identified in the dataset after reviewing data for erroneous entries and errors in sampling or laboratory methods. Duration curves of turbidity data indicate that turbidity remained within the range of sensor accuracy (1,000 FNU) during the entirety of phase two (fig. 2).

During phase two (water years 2019–21), 40 discrete suspended sediment samples were collected, although only 35 were used in the development of the turbidity-derived regression model because of missing turbidity data from the sensor at the time of discrete sampling. Graphing nontransformed SSC against turbidity values identified a strong relation between the two properties. Inspection of the data and residual plots identified no outliers. The 35 data pairs were used to calibrate the SSC-turbidity model and are plotted on a log-log scale (fig. 3; table 4), with the final regression model as follows:

where The model information is as follows:

This regression model using phase two data was compared to the regression model from phase one. An analysis of covariance was used to compare the two regression lines and whether they were significantly different with respect to either slope or intercept (defined as p-value <0.05). The slopes of the two regression models were not significantly different (p-value equal to 0.815), but the intercepts were different from each other (p-value <0.001). Because of the significant difference between the two regression models, the phase two model replaced the phase one model for SSC and SSL calculations for phase two of the study period. The shift in the regression between the two phases could be because of a change in the nature or grain-size of the sediment.

For periods when instantaneous turbidity was not available, a gap-fill framework was used by computing SSC using a secondary regression model based on instantaneous streamflow. Graphing the relation between SSC and streamflow identified a less statistically significant relation than the turbidity relation (fig. 4). Streamflow-derived regression models tend to have greater uncertainty than turbidity-derived regression models (Rasmussen and others, 2009). Suspended sediment concentrations computed using the streamflow-derived model (approximately 42 percent of the record) were flagged as estimated values. The suspended sediment model derived from streamflow is as follows:

where The model information is as follows:

Streamflow and the regression-computed SSCs were used to compute daily mean SSLs. These SSC and SSL results are available via NWIS (USGS, 2023). To understand the effect of sampled events on sediment transport, event-based SSLs were computed using the runoff volumes and regression-computed SSCs. The total runoff volume is the sum of the 15-minute interval streamflow during the stormflow event divided by the duration of the event. The beginning of a stormflow event was generally determined as the time 1 hour before the streamflow doubled, and the end of a stormflow event was generally defined as when the streamflow decreased consistently by approximately 20 ft³/s during each 15-minute measurement for a continuous 2-hour period. Event-based SSL was computed using the following equation for each 15-minute interval, followed by a summation of all intervals:

where Sediment yield is the amount of land surface erosion during a period of time per unit area of the hydrologic basin. Sediment yield normalizes SSL by including basin size and stormflow event duration. Event-based suspended sediment yield was computed using the following equation: where

Event-based loads and yields of barium, cadmium, lead, and zinc were computed from concentrations of the trace elements in the suspended sediment samples. Both sand and fine sediment size fractions were used to compute a bulk mass accumulation (sediment less than 2 mm in size) using the following equation:

where

In instances when only enough mass was available for trace element analysis on the bulk sample (rather than the separate size fractions), the bulk sample was analyzed, and the bulk concentrations were used. The arithmetic mean of the trace element concentrations for each stormflow event were computed from the suspended sediment samples collected during the event. The mean stormflow event concentration of each trace element was then expressed as a concentration in milligrams per liter using the SSC derived from the regression model for each 15-minute value. The following equation was used:

where Using these concentrations, each of the event-based loads and yields of barium, cadmium, lead, and zinc were computed as described for event-based SSL and suspended sediment yields in equations 3 and 4.

Streamflow Conditions

Precipitation can greatly affect streamflow conditions. Monthly precipitation data were obtained from the DeSoto, Mo., climate station, which is about 13 mi north of the study site (NOAA, 2022; fig. 5). During phase two, total monthly precipitation ranged from 1.0 in. (September 2020) to 7.23 in. (May 2019). The normal annual precipitation amount for the climate station (years 1991–2020) is 44.73 in. All water years during phase two exceeded the normal annual precipitation. Water year 2019 had the most precipitation (56.29 in.) followed by water year 2020 (48.79 in.). The driest year of phase two was water year 2021 (46.04 in.; NOAA, 2022).

Phase two consisted of wetter conditions and higher mean streamflows than the first year of phase one (table 5; Barr, 2016). Annual mean streamflows during phase one were 189 ft³/s in water year 2012 and 695 ft³/s in water year 2013. Annual mean streamflows during phase two were 640 ft³/s in water year 2019, 450 ft³/s in water year 2020, and 461 ft³/s in water year 2021 (table 5). In comparison, the mean annual streamflow measured while the streamgage has been operational (water years 2012–21) was 508 ft³/s (USGS, 2023). During phase one, daily mean streamflow was typically highest in the winter and spring months and lowest in the fall (fig. 6). The maximum daily mean streamflow for phase two was 12,000 ft³/s, which happened on January 11, 2020 (fig. 6, table 5). The minimum daily mean streamflow was 40.3 ft³/s on October 10, 2019. The maximum peak flow for each water year was 10,200 ft³/s (water year 2019), 16,000 ft³/s (water year 2020), and 13,100 ft³/s (water year 2021; table 5, fig. 6). The historical maximum peak streamflow was 41,100 ft³/s recorded on April 30, 2017 (table 5).

A comparison of streamflow to the previous report by Barr (2016) highlights the drier conditions during water year 2012 (annual mean streamflow of 189 ft³/s). Streamflow conditions in water year 2013 (annual mean streamflow of 695 ft³/s) during phase one were comparable to the conditions in water year 2019 during phase two (annual mean streamflow of 640 ft³/s). Overall, streamflow during phase two was more consistent than in phase one (water years 2012–13).

Continuous Water-Quality Parameters

Water temperature data were collected at the study site because of its importance in monitoring sonde operations, such as indications of biological activity, streamflow conditions, and human influences (Wilde, 2006). Additionally, sediment transport efficiency is affected by water temperature because of its relation with viscosity (Charlton, 2008). The maximum water temperature recorded at the study site was 31.3 °C on July 9, 2020, and August 12, 2021 (table 6). The minimum water temperature recorded was −1.0 °C on several days (table 6). Water temperatures generally followed the expected diurnal and seasonal patterns (fig. 7). A few periods of the temperature record were removed during QA review.

Suspended sediment transport in the Big River may be better understood with the use of continuous turbidity data as turbidity can be used as a surrogate for SSC. The daily mean turbidity during phase two ranged from 0.4 FNU on November 17, 2018, and February 12–13, 2021, to 263 FNU on January 11, 2020 (table 6; fig. 8). The turbidity record is fragmentary in nature (missing approximately 44 percent of daily mean turbidity values) because of instrument biofouling during warmer months and siltation of the deployment pipe during stormflow event flows, as well as calibration drift and instrument malfunction (fig. 8).

Whereas increases in turbidity correlated with increases in streamgage height and streamflow, the rate at which each increased was not always consistent. Peaks in turbidity typically precede streamflow peaks for the same stormflow event (Mather and Johnson, 2014). The lag time between the peaks increased with higher flow events. As an example, the maximum turbidity for the study period (965 FNU on November 22, 2020; table 6) did not correlate with the maximum streamflow (16,000 ft³/s on January 12, 2020; table 5). External factors such as timing of previous stormflow events, runoff amounts, and location of rainfall can affect sediment availability and observed turbidity during stormflow events.

Turbidity observed during a stormflow event is the result of several factors, including sediment washed into the stream via runoff, increased suspended sediment transport, and erosion of streambanks because of higher stream velocities. When multiple stormflow events happen within a short time span, subsequent events can have reduced amounts of suspended sediment and turbidity as sediment becomes less available because it was transported from the land surface, stream bed, and streambank during prior events. The location and intensity of rainfall and where it falls within the drainage basin also may affect turbidity.

Daily Suspended Sediment Concentrations and Loads

Daily SSCs and SSLs were computed using the regression models and followed similar patterns during the phase two study period, with maximum and minimum daily mean SSCs and SSLs frequently happening on the same dates (figs. 10 and 11). The daily mean SSL ranged from a minimum of 0.4 ton on September 24, 2020, to a maximum of 24,900 tons on March 13, 2021. During water year 2019, the maximum daily mean SSC of 816 mg/L and maximum daily mean SSL of 19,400 tons were measured on December 16, 2018 (table 8), which happened during the same stormflow event as the maximum daily mean streamflow for that water year (table 5). During water year 2020, the maximum daily mean SSC of 716 mg/L and maximum daily mean SSL of 22,600 tons were measured on January 12, 2020 (table 8), which was also during the same stormflow event as the maximum daily mean streamflow for water year 2020 (table 5). The maximum daily mean SSC and SSL (1,010 mg/L and 24,900 tons, respectively) during water year 2021 happened on separate dates (maximum concentration on November 23, 2020, and maximum load on March 13, 2021; table 8). The maximum daily mean streamflow for water year 2021 coincided with the same stormflow event as the maximum daily mean SSL in March 2021 (tables 5 and 8). The duration and magnitude of the stormflow event and availability of suspended sediment in the system likely played a role in the differing occurrences of the maximum daily mean concentrations and loads in water year 2021.

The estimated annual SSL at the study site varied during phase two and was correlated with the annual mean streamflow with the largest mass of 113,000 tons transported during water year 2019 (annual mean streamflow of 640 ft³/s; tables 5 and 8). Water years 2020 and 2021 had annual SSLs of 83,400 and 96,500 tons, respectively (table 8). The lowest annual SSL for phase two (83,400 tons) happened during water year 2020, which coincided with the lowest annual mean streamflow (450 ft³/s; tables 5 and 8). Additionally, more turbidity record was missing during water year 2020 when compared to water years 2019 and 2021, which resulted in more of the SSC and SSL record for water year 2020 being estimated using the secondary streamflow-derived regression model. The secondary regression model underestimates the SSC more often than the primary regression model based on turbidity. Approximately 58 percent of the calculated (15-minute) SSCs for the study period were determined using the turbidity-based regression model; the remaining 42 percent of the record was calculated using the secondary streamflow-derived regression model.

Event-Based Suspended Sediment Concentrations, Loads, and Yields

Eight stormflow events were sampled during the phase two study period. Mean stormflow event SSCs and total stormflow event SSLs calculated using the regression model are reported for each of the individual events. Mean estimated stormflow event SSCs ranged from 50 to 717 mg/L, and total SSLs ranged from 45.3 to 32,500 tons (table 9). Events lasted an average of about 47 hours. Event runoff volumes ranged from 26,600,000 ft³ (event 5 on August 31–September 2, 2020) to 1,770,000,000 ft³ (event 4 on January 10–13, 2020). Event-based yields normalize each stormflow event based on the drainage area and the event duration. Sediment yield ranged from 5.03 pounds per square mile per hour (lb/mi²/hr) to 3,460 lb/mi²/hr (table 9). The stormflow events with the largest runoff volumes generally had the largest sediment yields.

Sediment availability and rainfall intensity play critical roles in sediment transport in the Big River. Similar-size stormflow events (similar peak or total streamflow volume) are not always comparable. Events 2, 3, and 7 had comparable event runoff volumes, and they also had similar average SSCs (398, 471, and 491 mg/L, respectively; table 9); however, the two largest stormflow events—event 4 (6th highest peak streamflow for period of record) and event 8 (8th highest peak streamflow for period of record)—had differing mean SSCs (table 9). The mean estimated SSC during event 4 was 475 mg/L; the mean for event 8 was 717 mg/L. The timing and magnitude of prior stormflow events and the characteristics of each rainfall event can affect the transport of suspended sediment and associated trace elements.

The total SSL for the eight sampled stormflow events during phase two was computed as about 90,400 tons and accounted for 30.9 percent of the total SSL calculated using the regression models for phase two. To estimate a representative base-flow SSL for the phase two study period, the median base-flow SSC (10 mg/L; table 7) and the median of the daily mean streamflows associated with each base-flow sample during phase two were used (130 ft³/s; table 7). The estimated base-flow SSL was 3,840 tons, which accounted for 1.31 percent of the total SSL. This highlights that most suspended sediments in the Big River are transported during stormflow events as was concluded in phase one (Barr, 2016).

Selected Trace Element Concentrations in Suspended Sediments

Stormflow event samples were analyzed separately for trace element concentrations among the sand and fine fractions except for samples collected during event 6 and a single sample from both events 1 and 7. All stormflow event samples analyzed for trace element concentrations were made up of mostly fines (median of 74 percent fines) with a range of 63 to 98 percent fines (table 10). Although samples collected during event 6, as well as a single sample from both events 1 and 7, were analyzed for trace elements in bulk, the samples were analyzed for particle-size distribution before chemical analysis, which indicated these samples also were predominately fines (88–90 percent fines for event 6; 70–75 percent fines for bulk-analyzed samples from events 1 and 7; table 10). In general, the fine fractions had higher concentrations of barium, cadmium, lead, and zinc than the sand fractions except for some samples collected during event 8. Barium concentrations in the fine fractions were about twice the concentrations of the sand fractions for all stormflow event samples. Differences in concentrations among sand and fine fractions for cadmium, lead, and zinc were less, with cadmium concentrations being most similar in the two size fractions (fig. 12; table 11).

Trace element concentrations of cadmium, lead, and zinc in suspended sediment from stormflow event samples generally were consistent throughout the duration of each individual event, except events 1 and 8, which indicated a general increase in concentrations within the fine fractions during the event (table 11; fig. 12). These two stormflow events were larger events, with the highest mean event sample SSCs during phase two (522 mg/L for event 1 and 1,050 mg/L for event 8; table 9). This general pattern could indicate new contributions of mine waste material with high concentrations of trace elements that were mobilized from the landscape upstream from the Big River below Bonne Terre, Mo., as the result of the large storm event. Concentrations of barium were generally consistent throughout the duration of each individual event.

Two samples were collected during event 6, which happened on October 29–30, 2020. The samples from event 6 were bulk analysis samples because of the lack of sediment mass between the two size fractions. Both bulk samples had the highest concentrations of cadmium (23.8 and 24.1 mg/kg), lead (3,060 and 2,820 mg/kg), and zinc (1,660 and 1,630 mg/kg) from all event samples during the phase two study (fig. 12, table 11). Suspended sediment samples collected during event 6 were 88–90 percent fines with SSCs ranging from 67 to 215 mg/L, which were low compared to other sampled stormflow events (table 10). Event 6 was the second smallest event (by event runoff volume) sampled during phase two (table 9); however, the heaviest rainfall was localized (average of 2.1 in.) to the immediate area of the major mine waste piles within the OLB (Federal, Desloge, Elvins, and National mine waste piles) within and just west of Park Hills. During the same period, the mean rainfall across the entire 409-mi² drainage basin upstream from the streamgage below Bonne Terre was less than 1.5 in. Digital raster data of 24-hour rainfall estimates using NOAA Multi-Radar Multi-Sensor System data were obtained from the Iowa Environmental Mesonet (Iowa State University, 2023). One hypothesis is that the localized heavier rainfall and resulting runoff during event 6 eroded and suspended mine-waste contaminated sediments primarily in the area of the major mine waste piles within the OLB. These highly contaminated sediments were not diluted by runoff containing uncontaminated sediments from upstream in nonmined areas of the drainage basin. These signature high trace element concentrations from mine-waste contaminated sediments are likely present during all stormflow events, but they are diluted by runoff from other areas of the drainage basin during the larger stormflow events typically sampled. A second hypothesis is that this smaller event transported primarily fine bed sediments deposited from a previous event. Because the small event only transported the fine, more-contaminated bed material, the suspended sediment samples consisted of mostly these materials and were not diluted by coarser, lower concentration materials.

Cadmium, lead, and zinc are considered toxic to aquatic life at certain concentrations in bed sediments (MacDonald and others, 2000). Because the suspended sediments collected during this study were mobilized from bed sediments and many will return to the riverbed, the sediment quality guidelines for bed sediments were used for comparison to suspended sediment trace element concentrations. The toxic effect concentration (TEC) is defined as the concentration below which adverse effects are not expected to occur. The probable effect concentration (PEC) is the concentration at which adverse effects are expected to occur when concentrations are at or above the PEC. There are thresholds within these sediment quality guidelines, including the highest threshold in the PEC group, the toxic effect threshold (TET), which is the concentration above which sediment is considered heavily contaminated and likely to cause adverse effects on sediment-dwelling organisms (MacDonald and others, 2000). The bulk concentration of barium in suspended sediments ranged from 416 to 531 mg/kg with a median of 485 mg/kg (table 11; fig. 12). Barium does not have an established TEC, PEC, or TET. Bulk cadmium concentrations had a median of 7.90 mg/kg and ranged from 4.60 to 24.1 mg/kg, all of which all exceeded the TEC for cadmium in sediment (table 11, 12). Bulk lead concentrations exceeded the TEC, PEC, and TET in all samples, with a range of 747 to 3,060 mg/kg and a median of 1,070 mg/kg. Bulk zinc concentrations ranged from 271 to 1,660 mg/kg with a median of 500 mg/kg. Zinc concentrations in bulk suspended sediments exceeded the TEC, frequently exceeded the PEC, and occasionally exceeded the TET (table 11, 12).

Samples for dissolved trace element concentrations were not collected during phase two or during the phase one completed by Barr (2016) because of low dissolved lead concentrations reported in previous studies of the Big River. Smith and Schumacher (1991, 1993) collected samples for dissolved trace element concentrations at different sites on the Big River—upstream near Desloge (USGS site number 07017260) and downstream near Richwoods (USGS site number 07018100; fig. 1)—and reported concentrations of dissolved lead below the reporting limit of 10 micrograms per liter (µg/L). Concentrations of dissolved lead in 55 samples collected during water years 2000 through 2021 from the Big River near Richwoods site as part of the Missouri Ambient Water Quality Network had a median dissolved lead concentration of 3.3 µg/L and a maximum lead concentration of 15.7 µg/L (USGS, 2023).

Trace element concentrations in the fine fraction of suspended sediments transported by the Big River generally are similar to those from phase one (water years 2012–13), with small differences in barium and zinc concentrations and no significant changes in cadmium and lead concentrations (fig. 13). In the fine fraction of suspended sediment samples, Barr (2016) reported median concentrations of barium, cadmium, lead, and zinc of 590, 10.0, 1,000, and 860 mg/kg, respectively. During phase two, the median concentrations of barium, cadmium, lead, and zinc in the fine fractions were 568, 8.15, 1,180, and 535 mg/kg, respectively (table 11). Wilcoxon rank-sum tests were performed to determine whether the concentrations of these trace elements in the fine fractions were significantly different from those determined for phase one of the study (Barr, 2016). The barium and zinc concentrations were significantly smaller (p-values of 0.028 and <0.001, respectively) than those reported by Barr (2016), whereas the cadmium and lead concentrations were not significantly different between the study period phases (p-value greater than 0.05; p-values of 0.1304 and 0.3116, respectively). It is possible that remediation of major mine waste areas within the OLB might have contributed to the significantly smaller zinc concentrations; however, barium is not associated with mineralization within the OLB and given the small dataset, additional monitoring data would be beneficial.

The trace element concentrations in the suspended sediment also were expressed as whole-water concentrations (milligrams per liter, using the measured SSC concentration in water; table 10). The median barium and cadmium concentrations were 0.240 and 0.004 mg/L, respectively. Lead and zinc concentrations had median concentrations of 0.507 and 0.273 mg/L, respectively. The maximum lead concentration was 1.30 mg/L and happened during the first hydrologic rise during event 8. The maximum barium, cadmium, and lead concentrations also were observed in the same sample from event 8. This sample did not have the highest concentrations of trace elements in the sediment itself, but because of the high SSC (1,740 mg/L), the overall concentration of barium, cadmium, lead, and zinc were ultimately greatest in this sample. The concentration of trace elements in suspended sediments remained consistent during a sampled stormflow event (table 11); however, the calculated concentrations of trace elements in milligrams per liter generally decreased during a stormflow event as the SSC decreased as the event receded (table 10).

Loads and yields of barium, cadmium, lead, and zinc were calculated for the sampled stormflow events (table 13). Barium and cadmium event loads ranged from 0.199 to 17.1 tons and 0.009 to 0.239 ton, respectively. Lead event loads ranged from 1.14 to 34.0 tons, and zinc event loads ranged from 0.636 to 14.5 tons (table 13). The largest barium and zinc loads happened during event 4, and the largest cadmium and lead loads happened during event 8, with these two events having the largest suspended sediment loads of phase two (SSL of 32,500 tons during event 4 and 31,800 tons during event 8; table 9, 13).

The total event-based loads for each water year displayed in table 13 represent a minimum of the total trace element load because of the small number of stormflow events sampled or analyzed for trace elements. Event-based yields were calculated to make individual events more comparable by considering the drainage area (409 mi² for all samples) and individual event durations. Yields were greatest for barium, cadmium, lead, and zinc during event 8, and trace element yields were similar between events 1, 2, and 7 (table 13).

When compared to phase one of the study by Barr (2016), the event-based trace element loads during water year 2012 (a drought year; Barr, 2016) were markedly less than the water years included in the current study period. For water year 2013, the trace element loads for sampled stormflow events were greater than water years 2019–21. The maximum daily mean streamflow for water year 2013 (18,600 ft³/s; Barr, 2016) was 6,600 ft³/s greater than the maximum daily mean streamflow recorded during phase two (12,000 ft³/s; table 5).

Selected Trace Element Concentrations in Suspended Sediments from Passive Samplers

Passive samplers were deployed throughout the duration of phase two and retrieved after major stormflow events or at periodic intervals throughout the water year. Samplers were located such that they would become inundated and begin collecting sediment during stormflow events when the stage exceeded approximately 4.82 ft. The samplers were serviced 18 times from the initial deployment on September 13, 2018, to the final retrieval on March 16, 2022, representing 18 events. Bulk concentrations of cadmium and lead exceeded the PEC, TEC, and TET in all passive samples, whereas bulk concentrations of zinc exceeded the PEC in all passive samples, the TEC in 17 of 18 samples, and the TET in 15 of 18 samples (tables 10 and 14).

Bulk concentrations of barium, cadmium, lead, and zinc in the 18 passive samples were compared to bulk concentrations from the discrete suspended sediment stormflow event samples (tables 11 and 14; fig. 14). The median bulk concentrations of barium, cadmium, lead, and zinc in sediment from the passive samplers were 523, 13.6, 1,870, and 745 mg/kg, respectively (table 14). The median bulk concentrations of barium, cadmium, lead, and zinc from discrete stormflow event samples were 485, 7.90, 1,070, and 500 mg/kg, respectively (table 11). Wilcoxon rank-sum tests indicate that, although similar in magnitude, concentrations of barium, cadmium, lead, and zinc were significantly larger in the passive samples compared to the discrete stormflow event samples (p-values equal to 0.00824, <0.001, <0.001, and 0.00238, respectively). It should be noted that the passive samples were dry sieved during processing at the sediment laboratory, but the discrete samples were wet sieved. Because of the possibility of aggregated fines being included in the sand fraction after dry sieving, the fine fractions also were compared between the passive and discrete samples to avoid the potential bias caused by the difference in methods. Wilcoxon rank-sum tests of the fine fractions indicated similar conclusions as the bulk comparison with the concentrations of cadmium, lead, and zinc being significantly greater in the passive samples compared to the discrete suspended sediment samples (p-values equal to 0.00563, 0.00183, and 0.0156, respectively). Concentrations of barium were not significantly different when comparing the fine fractions in passive and discrete samples (p-value equal to 0.8247).

The most probable explanation for the higher trace element concentrations in the suspended sediments collected using the passive samplers is because the passive samplers are integrating sediment during a longer duration of the stormflow event (at about stage 4.82 ft and above), whereas the discrete cross-section samples are generally collected at much higher stages and later from the start of the individual event when a team can be present to obtain the sample. As such, discrete samples represent only a short period of time during the full hydrologic event and often miss high concentration early sediment transport. Each runoff event also has much variability especially if a bank of highly contaminated sediments sloughs into the river. The early part of a stormflow event at Bonne Terre has a larger proportion of locally derived sediment mobilized from the streambed and runoff from the OLB area in addition to the fine materials deposited from the previous runoff event. As the stage increases, the proportional contribution from the drainage basin upstream from the OLB gradually increases. Concentrations of mining-related trace elements in the fine fraction of the discrete samples collected on the rising limb of stormflow events during both phases of the study are inversely correlated with stage with correlation coefficients of −0.38 (cadmium), −0.26 (cobalt), −0.40 (lead), −0.63 (nickel), and −0.56 (zinc). In contrast, concentrations of barium in the rising limb samples had no correlation with stage (correlation coefficient of 0.00). Barium is not a trace element associated with mining in the OLB but is widely distributed across the middle and upper Big River drainage basin as barite. The largest lead concentration of 3,060 mg/kg detected in any discrete suspended sediment sample collected during both phases of the study was collected at a stage of only 4.74 ft, shortly after a small rise began on October 29, 2020 (event 6). Widespread projected heavy rains never materialized for that stormflow event. As previously discussed, the 24- and 48-hour rainfall estimates obtained from the Iowa State University Mesonet (Iowa State University, 2023) indicated the heavier rainfall was localized, with the highest totals (mean of about 2.1 in.) happening in the immediate area of the major mine waste piles within the OLB and Park Hills area. The stage peaked on October 29, 2020, at slightly less than 6.0 ft about 6 hours after it began rising. It is hypothesized that the initial rise of all stormflow events likely contains these higher lead concentrations derived from within the OLB and major mine waste areas and the fine materials deposited from the prior runoff event; however, during larger events, they are diluted by sediments derived from mostly nonmined areas from upstream parts of the drainage basin. During the rising limb of hydrologic events, the combination of high concentrations of lead in suspended sediments and the high concentrations of suspended sediments results in larger fluxes of lead at the beginning of events with SSC commonly peaking before or at the same time as the hydrologic peak (Guy, 1970).

The trace element concentrations in passive sediment samples were used to estimate total annual loads of barium, cadmium, lead, and zinc. The daily mean SSL calculated from the regression model (table 8) was used to calculate the SSL represented by each passive sampler deployment. The sum of the deployment period daily SSLs was then multiplied by the bulk trace element concentrations in the passive sample from that deployment period to estimate the total suspended sediment trace element load (table 15). Although the samplers were not submerged when stage was below 4.82 ft, the concentrations for the respective deployment period were used to calculate trace element loads for days when the sampler was not submerged and collecting sediment. This methodology was used knowing that most of the sediment and trace element load at the Big River below Bonne Terre sampling site is transported during stormflow events. Additionally, sediment samples were collected using a continuous flow centrifuge during base-flow conditions (March 5, 2021, and August 16 and 19, 2021) as part of another USGS project in the region to examine solid-phase associations of trace elements and trace element concentrations in particulates. These samples were collected at the Missouri Department of Conservation access at Cherokee Landing about 2 mi upstream from the Bonne Terre streamgage site (fig. 1). Generally, median bulk concentrations of barium, cadmium, lead, and zinc (553, 19.5, 2,130, and 1,646 mg/kg, respectively; Markland and others, 2024) fell within the range of concentrations observed in the discrete samples collected during stormflow conditions (table 11). The results from the continuous flow centrifuge sample further support the use of the passive sampler trace element data for the duration of the deployment period, even when the samplers were not submerged in the water column.

The passive samplers were submerged for approximately 330 days. The annual loads of barium and cadmium ranged from 41.4 to 58.6 tons per year and 1.17 to 1.43 tons per year, respectively. The annual load of lead ranged from 152 to 194 tons per year, and the zinc load ranged from 68.1 to 76.5 tons per year. Water year 2019 had the largest loads for all four trace elements (table 15). The loads were primarily transported during days with a median daily streamflow associated with the approximate stage above which the passive sampler was submerged. Less than 5 percent of the total lead load calculated for phase two was transported when daily mean streamflow was less than 455 ft³/s, which further highlights that most of the sediment and trace element loads are transported during stormflow events. The use of passive samplers to obtain suspended sediment samples is an effective way to monitor the transport of lead and other trace elements in the Big River drainage basin.