Purpose and Scope

This report documents the distribution of streamflow and nutrient and suspended-sediment concentrations and loads in freshwater inflow entering Galveston Bay from the Trinity River, including its complex system of distributary channels, lakes, and interdistributary wetlands in the lower reaches, during 2016–19. Instantaneous streamflow data and continuous streamflow records from USGS monitoring stations in the delta of the Trinity River were used to quantify freshwater inflow into Galveston Bay and assess the distribution of streamflow in the delta of the Trinity River Basin. Water quality was characterized from in situ measurements of physicochemical properties and discrete nutrient and sediment samples collected predominantly during high-flow events at various USGS monitoring stations in the delta of the Trinity River. Water-quality data also were used to assess the influence of the distribution of streamflow in the delta of the Trinity River on nutrient and sediment inputs into Galveston Bay. This report helps address the need for streamflow and water-quality information in the delta of the Trinity River to improve our understanding of the processes that influence the delivery of freshwater, nutrients, and suspended sediment from the Trinity River into Galveston Bay.

Study Area Description

The Trinity River Basin covers approximately 18,000 square miles and is the largest river basin entirely in Texas. From its headwaters near Dallas, the Trinity River flows approximately 715 miles before emptying into the Galveston Bay (Trinity River Authority of Texas, 2021). For this study, the Trinity River Basin was divided into upper and lower parts (hereinafter referred to as the “upper Trinity River Basin” and “lower Trinity River Basin,” respectively) separated by Lake Livingston, a water-supply reservoir (fig. 1); the Trinity River was likewise separated by Lake Livingston into upper and lower parts hereinafter referred to as the “upper Trinity River and “lower Trinity River.” The hydrology in the lower Trinity River Basin is predominantly driven by releases from Lake Livingston as a result of rainfall-runoff events in the upper Trinity River Basin (Lucena and Lee, 2017). These rainfall-runoff events result in the release of large volumes of water from Lake Livingston for extended periods and are typically several orders of magnitude larger than events that cause localized surface runoff. In addition to periods of runoff-generated streamflow, there were ongoing small releases from Lake Livingston during the study period designed to mirror the streamflow entering Lake Livingston from the upper Trinity River Basin (Trinity River Authority of Texas, 2022 ); during these periods of small releases, tides control daily streamflow patterns in the extreme lowermost part of the Trinity River near Galveston Bay, and upstream movement is regulated by a saltwater barrier constructed and operated by the U.S. Army Corps of Engineers that is approximately 2 river miles downstream from the Wallisville site (fig. 2).

The study area (fig. 1) encompasses the lower Trinity River Basin, downstream from Lake Livingston and upstream from where the Trinity River delta meets Galveston Bay. The lower Trinity River flows downstream from Lake Livingston in one main channel for about 85 river miles. Approximately 4 river miles downstream from Moss Bluff (fig. 1), the lower Trinity River bifurcates from its main channel to form various distributaries (Phillips and Slattery, 2007), including the Old River and the Lost River, located west of the main channel of the Trinity River (fig. 2). The Old River and the Lost River flow through the delta plain until they converge at Interstate 10 (fig. 2) to form Old River Lake, a large, shallow open water body surrounded by freshwater wetlands and distributary channels. Downstream from Interstate 10, in the lower delta plain, freshwater inflow enters Galveston Bay through various hydrologically connected distributaries and estuarine wetlands.

Freshwater inflow from the Trinity River Basin may also enter Galveston Bay through Lake Anahuac (figs. 1, 2), south of Interstate 10 and east of the main channel of the Trinity River. Lake Anahuac is a reservoir used for municipal, irrigation, industrial, and mining supply purposes (TWDB, 2006). The reservoir receives water from Turtle Bayou, but when necessary, water is pumped from the Trinity River into Lake Anahuac through Big Hog Bayou (fig. 2). The reservoir’s spillway is at the south end of the lake, approximately 3 river miles downstream from the confluence between the Trinity River and the Old River Cutoff. A levee prevents Lake Anahuac from being hydrologically connected to the Trinity River, except at times when water is being pumped into the reservoir through Big Hog Bayou and when high water levels cause water to flow over the uncontrolled spillway into the Trinity River. Because Lake Anahuac is usually not hydrologically connected to the Trinity River, its overall contribution of freshwater inflow, nutrients, and sediment to Galveston Bay was not monitored as part of this study.

Five USGS monitoring stations where continuous streamflow data are collected in the Trinity River (fig. 1; table 1) are included in this study. Four of these stations are on the main channel of the Trinity River: 08066250 Trinity River near Goodrich, Tex. (hereinafter referred to as “the Goodrich site”); 08066500 Trinity River at Romayor, Tex. (hereinafter referred to as “the Romayor site”); 08067000 Trinity River at Liberty, Tex. (hereinafter referred to as “the Liberty site”); and 08067252 Trinity River at Wallisville, Tex. (hereinafter referred to as “the Wallisville site”). The Wallisville site is approximately 8 river miles upstream from where the Trinity River enters Galveston Bay. The Liberty site is approximately 30 river miles upstream from the Wallisville site and 16 river miles upstream from where the Trinity River branches off to form the Old River and Lost River distributaries; the Romayor site is approximately 40 river miles upstream from the Liberty site; and the Goodrich site is approximately 20 river miles upstream from the Romayor site. The Goodrich site is about 10 river miles downstream from Lake Livingston Dam. The fifth USGS monitoring station used in this study (08067070 Coastal Water Authority [CWA] Canal near Dayton, Tex., hereinafter referred to as “the CWA Canal site”) is at the intake of the CWA Canal approximately 10 river miles downstream from the Liberty site. The CWA Canal is used to transport water that is diverted out of the main channel of the Trinity River into a water-supply reservoir. Water is also diverted out of the Trinity River through the ungaged Devers Canal (fig. 1) and Big Hog Bayou (fig. 2).

In addition to the five sites where continuous streamflow data were collected (Goodrich, Romayor, Liberty, Wallisville, and CWA Canal sites), discrete (periodic) streamflow data were also collected at USGS stations 08067230 Old River Lake near Wallisville, Tex. (hereinafter referred to as “the Old River Lake site”) and 08067256 Old River Cutoff at Buckeye Island near Anahuac, Tex. (hereinafter referred to as “the Old River Cutoff site”). Water-quality monitoring data (including suspended-sediment samples) were collected at the Wallisville site and at two sites on distributaries in the delta of the Trinity River (Old River Lake site and USGS station 08067254 Old River at Buckeye Island near Anahuac, Tex. [hereinafter referred to as “the Buckeye Island site”]) (fig. 2; table 1).

Previous Studies

The USGS, in cooperation with the TWDB, has studied the lower reaches of the Trinity River since 2009 to evaluate the variability of freshwater inflow, nutrients, and sediment entering Galveston Bay during a variety of hydrologic conditions. During 2009–13, discrete measurements of streamflow were made during high-flow events at the Wallisville site to help quantify freshwater inflow to Galveston Bay. Lee (2010) evaluated two periods of relatively high streamflow in April 2009 and September 2009 in response to separate rain events. The peak streamflows at the Liberty site were approximately 27,000 cubic feet per second (ft³/s) in April 2009 and 60,000 ft³/s in September 2009. The corresponding peak streamflows at the downstream Wallisville site were approximately 16,500 ft³/s and 21,700 ft³/s, respectively (40 and 64 percent less than the peak streamflow at the upstream Liberty site) (Lee, 2010). An overall decrease in streamflow was also observed at the Wallisville site compared to the streamflow at the upstream Liberty site by using data from discrete streamflow measurements made during selected high-flow events in 2010–13 (USGS, 2021), possibly as a result of channel bifurcations causing water to divert into branching channels in the delta.

In 2014, the USGS installed a streamgage to continuously monitor streamflow at the Wallisville site in order to examine the hydrodynamics in the lowermost reaches of the main channel of the Trinity River. The continuous streamflow record during 2014–15 at the Wallisville site was used to evaluate the differences between the water volume measured at this site and the water volume measured at the upstream stations (Goodrich, Romayor, and Liberty sites) (Lucena and Lee, 2017). Lucena and Lee (2017) found that during 2014–15 only about 54 percent of the total volume of water measured at the Goodrich site was accounted for at the Wallisville site, indicating that at high streamflows, typically greater than 20,000 ft³/s, a large part of the inflow did not reach Galveston Bay through the main channel of the Trinity River. Results from that study also indicate that water likely flows out of the main channel of the Trinity River into surrounding wetlands and water bodies, such as the Old River Lake, and is being stored or discharged directly into Galveston Bay through other channels in the delta. The findings from that study led to the collection of additional data during 2016–19 to further assess the distribution of streamflow in the various channels that form the delta of the Trinity River and evaluate the effects of streamflow diversions into Old River Lake on the eventual supply of freshwater, nutrients, and sediments to Galveston Bay.

Quality Control

To evaluate the variability of sample processing and analysis in water-quality constituents, data from field-replicate samples collected at the Old River Lake and Wallisville sites from 2009 to 2019 were considered. The dataset consists of 21 split-replicate samples (table 1.1). Split-replicate samples are prepared from a single sample that is collected and then subdivided into two samples. Quality-control data collected outside of the study period can be used to increase the size of the dataset and improve the understanding of variability in the environmental data as long as the samples are considered to be in the same inference space (Mueller and others, 2015). It was determined that split-replicate samples collected at the Wallisville site during 2009–15 share the same inference space (Lorenzen and Anderson, 1993) as the samples collected during this study at the Old River Lake and Wallisville sites because they represent similar conditions in terms of potential variability. This determination was made because (1) consistent sampling methods and equipment were used for the collection of all samples, (2) hydrologic conditions and concentrations in the quality-control dataset are representative of those observed during the study period, (3) analytical methods for most constituents did not change throughout the sampling period, and (4) sample collection and processing were conducted in the same general area.

For nitrate plus nitrite, only 6 of the 21 replicate samples were considered because of a change in the analytical method in 2011 and because some results were less than the method detection limit, defined as the lowest concentration of an analyte that can be reliably distinguished from a blank for a given analytical method (Long and Winefordner, 1983; Christian, 2004) and computed following methods described in Foreman and others (2021). For the rest of the constituents, replicate samples were only used in the analysis when results were greater than or equal to the method detection limit. Samples were not analyzed for total dissolved nitrogen before 2016; therefore, only four replicate pairs were used in the analysis.

To determine variability in environmental samples, the relative percent difference (RPD) between each pair of duplicate samples was calculated by using the following equation:

The RPDs of replicate sample pairs are included in appendix 1 (table 1.1). The mean RPD of replicate concentrations was less than 8.5 percent for all analytes. This mean RPD, although low, is conservative; small differences in constituent concentration that were only slightly greater than method detection limit resulted in large RPD values in some instances. High RPD values from concentrations only slightly greater than the method detection limit were retained. Overall, the replicate samples indicate low variability in nutrient and sediment analyses.

Tests at the USGS National Water Quality Laboratory in 2007 and 2009–10 indicate that the alkaline-persulfate digestion of whole water samples resulted in a potential negative bias in total nitrogen concentrations in the presence of suspended sediment (Rus and others, 2013). This negative bias may be site specific and varies in magnitude, depending on the origin and concentration of suspended sediment. To assess the potential bias at the Old River Lake, Wallisville, and Buckeye Island sites, total particulate nitrogen (TPN) concentrations were measured following methods described by Zimmerman and others (1997) in a subset of samples that was also analyzed for total nitrogen using the alkaline-persulfate digestion method. Total nitrogen concentrations from the sum of total dissolved nitrogen and TPN were compared to total nitrogen results obtained from the alkaline-persulfate digestion method for 30 samples. Bias ranged from −0.28 to 0.14 milligram per liter (mg/L) (−30.9 to 14.1 percent) at the three sites, with a median bias of −0.03 mg/L (−3.7 percent) at the Old River Lake site, 0 mg/L at the Wallisville site, and 0.02 mg/L (2.1 percent) at the Buckeye Island site. Negative bias did not increase as sediment concentration increased; instead, bias was random and present in most samples. Random bias may be partially explained by the variability in analytical results for TPN, which can affect the bias and precision of computed total nitrogen values. The average relative standard deviation, based on three replicate pairs, was 11 percent for TPN and 6.5 percent for total dissolved nitrogen, resulting in a total variability that ranged from 7.0 to 8.9 percent, depending on the relative concentrations of each constituent in each sample. The mean relative standard deviation for total nitrogen determined by the alkaline-persulfate method was 3.1 percent. For this study, the effects of a potential bias from the alkaline persulfate method are considered negligible because the variability associated with both analytical methods was similar and no relation between negative bias and suspended-sediment concentrations was observed.

Suspended-Sediment Regression Equation Development

Regression equations that contain acoustic backscatter, turbidity, or streamflow can be used to compute suspended-sediment concentrations at a high temporal resolution (typically every 15 minutes), providing an advantage over discrete suspended-sediment concentration samples (Rasmussen and others, 2009; Landers and others, 2016). For example, during high-flow events, the concentrations of suspended sediment can vary considerably within a short time. This temporal variability with discrete samples is difficult to capture because these samples provide information only for the time in which the sample was collected, resulting in interpolation between sampling points that may not accurately characterize variability in suspended-sediment concentrations (Rasmussen and others, 2009).

A regression equation based on continuous acoustic backscatter data was developed by Lucena and Lee (2017) for estimating suspended-sediment concentrations at the Wallisville site (table 2) following methods described by Landers and others (2016). Continuous suspended-sediment concentrations were estimated from acoustic backscatter data collected at 15-minute intervals at the Wallisville site between April 19, 2014, and December 23, 2019. To compare sediment concentrations and loads at the Liberty and Wallisville sites during selected high-flow events in 2016–19, a regression equation was developed to estimate suspended-sediment concentrations at the Liberty site from streamflow and suspended-sediment concentration data collected between January 9, 2014, and July 14, 2020; the suspended-sediment concentration data were collected at the Liberty site in cooperation with the City of Houston. It is preferable to use continuous backscatter or turbidity data when developing regression equations to predict suspended-sediment concentrations (Rasmussen and others, 2009; Landers and others, 2016), but only continuous streamflow data were available at the Liberty site. Sixty-seven observations were used to calibrate the suspended-sediment regression equation (USGS, 2021) (table 1.2). Although 72 suspended-sediment concentration samples were collected at the Liberty site, the data from 5 samples were excluded because of potential issues with data quality indicated by an outlier evaluation based on a graphical analysis of residuals and leverage and influence statistics (Helsel and others, 2020). Outlier values were also identified by a further assessment of field measurements and observations to identify any potential errors that could have occurred during sample collection. Prior to 2018, streamflow data for the Liberty site are only available at streamflows greater than or equal to 10,000 ft³/s. To ensure the entire range of hydrologic conditions was considered and to minimize bias from high-flow observations, streamflow values less than 10,000 ft³/s that were missing for 29 of the remaining 67 suspended-sediment samples collected at the Liberty site were estimated based on streamflow data collected at the Romayor site by using the drainage-area method (Hirsch, 1979).

Streamflow and suspended-sediment concentration data used to develop the regression equation for suspended-sediment concentrations ranged from 785 to 83,700 ft³/s and from 15 to 550 mg/L, respectively (table 1.2). Because streamflow and suspended-sediment concentration data were not normally distributed, data were log transformed to approximate a normal distribution as required for linear regression (Helsel and others, 2020). The use of a log transformation introduces a bias when retransforming data to their original units. This bias was corrected with the nonparametric bias correction factor introduced by Duan (1983). The relation between log-transformed streamflow and suspended-sediment concentrations is depicted in figure 3. The suspended-sediment regression equation for estimating suspended-sediment concentrations at the Liberty site and associated regression diagnostics are included in table 2.

Residual and probability plots (fig. 4) indicate that the regression equation for estimating suspended-sediment concentrations resulted in residuals (discrepancies between the observed and estimated data) that met the required inherent assumptions of normality and constant variance (Helsel and others, 2020). The relation between regression-estimated and measured suspended-sediment concentrations (fig. 4D) indicates that, at suspended-sediment concentrations greater than 300 mg/L, the regression equation generally underpredicts concentrations. Because of this limitation, it is possible that some suspended-sediment concentrations included in this report are underestimated. Other predictive variables were unavailable to estimate suspended-sediment concentrations at the Liberty site, and attempts to develop a regression equation by using other methods, such as seasonal and flow-range methods (Glysson, 1987), were not appropriate or did not yield better results. Thus, the regression equation that was developed is used only to estimate concentrations and loads during selected hydrologic events and to make comparisons of suspended-sediment concentrations and loads between the Liberty site and the Old River Lake and Wallisville sites.

Distribution of Streamflow from the Trinity River into Old River Lake

To assess the distribution of streamflow and estimate the amount of water diverted from the main channel of the Trinity River into Old River Lake, discrete streamflow measurements were made at monitoring stations in the delta of the Trinity River (fig. 2) on 11 different days during a range of hydrologic conditions. Measurements were made when streamflows at the upstream Liberty site ranged from approximately 16,000 to 125,000 ft³/s (fig. 6; table 3).

Two of the 11 days when discrete streamflow measurements were made were during low-flow events when streamflow at the Liberty site did not exceed approximately 16,700 ft³/s. The corresponding streamflows at the Old River Lake site remained influenced by flood tides resulting in a reversal in the direction of flow (negative streamflow value) at the time of the measurements. During the low-flow event measured on April 13, 2017, when the streamflow at the Liberty site was approximately 16,700 ft³/s, the streamflow at the Wallisville site was 13,200 ft³/s (fig. 7). This reduction of 3,500 ft³/s (21 percent) from the Liberty site downstream to the Wallisville site is consistent with streamflow losses reported by Lucena and Lee (2017) from the main channel of the Trinity River into distributary channels, wetlands, and canals that branch off from the main channel for a similar magnitude of streamflow. The CWA Canal is one of several canals in the study area that divert water from the Trinity River and the only canal where streamflow is measured. During the low-flow event measured on April 13, 2017, approximately 830 ft³/s was diverted from the Trinity River into the CWA Canal. The remaining volume of water (approximately 2,670 ft³/s) was likely diverted out of the main channel of the Trinity River into the various diversion pathways throughout the delta, but it was not large enough to attenuate the tidal-flow pattern typically observed in Old River Lake. A similar streamflow pattern was observed during another low-flow event on December 7, 2016, when streamflows at the Liberty and Wallisville sites were 16,000 ft³/s and 12,600 ft³/s, respectively, a difference of 3,400 ft³/s in streamflow between the two sites (where approximately 780 ft³/s was diverted into the CWA Canal). Even though not all freshwater outputs were quantified, it is likely that during low-flow events a portion of the water diverted out of the main channel of the Trinity River was stored in the channels, lakes, and wetlands in the delta of the Trinity River.

During 9 of the 11 days when discrete streamflow measurements were made, streamflow at the Liberty site exceeded 30,000 ft³/s, and streamflow at the Old River Lake site was large enough to attenuate the tidal-flow patterns. During these high-flow events, the combined streamflow measured at the Old River Lake and Wallisville sites was of similar magnitude to the estimated streamflow at the Liberty site (fig. 6). For example, on June 3, 2016, the streamflow at the Wallisville site was 22,300 ft³/s, and streamflow at the Old River Lake site was 64,200 ft³/s (fig. 8), resulting in a total of 86,500 ft³/s when combined. The streamflow measured at the Liberty site was approximately 81,000 ft³/s, only 5,500 ft³/s (less than 7 percent) less than the combined streamflow at the Old River Lake and Wallisville sites; this small discrepancy is likely attributed to a combination of factors not measured as part of this study, such as measurement uncertainty and local surface runoff downstream from the Liberty site. In addition, about 870 ft³/s was diverted from the main channel of the Trinity River into the CWA Canal. Considering all pertinent streamflow data from June 3, 2016, approximately 74 percent of the estimated freshwater inflow of 86,500 ft³/s was measured at the Old River Lake site, and approximately 26 percent was measured in the main channel of the Trinity River.

Data from discrete measurements made during the remaining high-flow events (fig. 6; table 3) indicate that a portion of the freshwater inflow, ranging from 30 to 74 percent of the total inflow, was delivered to Galveston Bay through pathways other than main channel of the Trinity River. Water that was diverted out of the main channel of the Trinity River south of Moss Bluff, Tex. (and was not stored or lost as a result of evapotranspiration, infiltration, or withdrawals) flowed into Old River Lake and ultimately into Galveston Bay through pathways that included the Old River and Lost River. As the magnitude of streamflow increased, the proportion of water diverted out of the main channel of the Trinity River also increased. During events when streamflow at the Liberty site exceeded approximately 40,000 ft³/s, the volume of freshwater traveling through Old River Lake exceeded the volume of water traveling through the main channel of the Trinity River. As a result, Old River Lake and its network of hydrologically connected channels likely became the primary pathways for freshwater inflow into Galveston Bay.

Distribution of Streamflow Downstream from the Old River Lake and Wallisville Sites

To assess the distribution of flow downstream from the Old River Lake and Wallisville sites, streamflow measurements were made at the Old River Cutoff site on 11 different days representing various hydrologic conditions (table 3). Results from these measurements indicate that the Old River Cutoff is a distributary stream that flows out of the Trinity River rather than a tributary stream that flows into the Trinity River. Approximately 3.5 miles downstream from the Wallisville site, a portion of the water from the main channel of the Trinity River flows into the Old River Cutoff and combines with Old River at the Buckeye Island site (figs. 7, 8). During the low-flow event measured on April 13, 2017, in which the streamflow at the Wallisville site was 13,200 ft³/s, the streamflow at the Old River Cutoff site was 8,760 ft³/s, approximately 66 percent of the total streamflow measured at the Wallisville site. Similarly, during the high-flow event on June 3, 2016, in which the streamflow at the Wallisville site was 22,300 ft³/s, the streamflow at the Old River Cutoff site was 10,500 ft³/s, approximately 47 percent of the total streamflow measured at the Wallisville site, flowing in the direction towards Old River Lake (fig. 7). During the measurements made between 2016 and 2019, including low-flow events, approximately 44 to 79 percent of the streamflow originating from the main channel in the lowermost reaches of the Trinity River traveled through the Old River Cutoff, mixed with the Old River at the Buckeye Island site, and then entered Galveston Bay through the distributaries and estuarine wetlands in the lower delta plain (fig. 8). The redistribution of streamflow into the Old River Cutoff is the consequence of sediment deposition in the form of a large shoal immediately downstream from the confluence between the Trinity River and the Old River Cutoff, which has caused a reduction of transport capacity in the main channel of the Trinity River (Letter and others, 2015). The Old River Cutoff (figs. 7, 8) has become the most hydraulically efficient pathway for inflow to Galveston Bay, and as the shoal continues to grow, the portion of streamflow that travels through the Old River Cutoff to Galveston Bay will likely increase, essentially making the Old River Cutoff the main pathway for inflow traveling from the main channel of the Trinity River into Galveston Bay (Letter and others, 2015).

Downstream from the confluence of the Trinity River and Old River Cutoff, the main channel of the Trinity River occasionally receives water from Lake Anahuac during localized runoff events in which water flows over the reservoir’s spillway. Because Lake Anahuac was not monitored as part of this study, its influence on freshwater, nutrient, and suspended-sediment delivery into Galveston Bay cannot be estimated. It is possible that its overall contribution of freshwater inflow into Galveston Bay is negligible when compared to the contributions of the main channel of the Trinity River and Old River Lake during high-flow events driven by Lake Livingston releases; however, during events primarily driven by local runoff, Lake Anahuac might be an important contributor of freshwater inflow into Galveston Bay. Additional data collected at this location during localized runoff events would allow for the quantification of Lake Anahuac’s contribution of freshwater inflow and associated sediment and nutrients into Galveston Bay.

Results indicate that even during low-flow events, the hydrologically connected wetlands, lakes, and distributary channels in the delta of the Trinity River play an important role in the delivery of freshwater inflow into Galveston Bay. During these low-flow events, Old River Lake (figs. 7, 8) and the surrounding wetlands likely provide storage for floodwaters, and the freshwater inflow to Galveston Bay is primarily delivered through the main channel of the Trinity River and the Old River Cutoff. During high-flow events, however, the Old River Lake and surrounding wetlands may be the primary pathways for freshwater inflow entering Galveston Bay through the delta of the Trinity River and not the main channel. Because the delta of the Trinity River is a dynamic system, water may be rerouted to other channels, causing the distribution of streamflow to continually change over time and continuously reshaping how freshwater inflow and associated sediment and nutrients are delivered to Galveston Bay.

Physicochemical Properties

Physicochemical properties of dissolved oxygen concentration, pH, specific conductance, water temperature, and turbidity were measured onsite during each sampling event of this study (table 1.3). The lowest dissolved oxygen concentrations were measured at all sites after Hurricane Harvey (August 2017) and ranged from 4.0 to 4.4 mg/L (table 5). During low-flow conditions on October 18, 2016, the specific conductance at the Old River Lake site was 9,470 microsiemens per centimeter at 25 degrees Celsius, a high value for the location, indicating an incoming tide at the time of sampling. A corresponding high specific conductance was not observed at the Wallisville site because the saltwater barrier operated by the U.S. Army Corps of Engineers approximately 2 river miles downstream from the Wallisville site prevents saltwater from entering the Trinity River.

The results from a one-way analysis of variance (ANOVA test; Helsel and others, 2020) for each parameter indicate that the means of the measured physicochemical properties were not significantly different (p-value greater than 0.05) between sites. Because physicochemical properties such as temperature and pH can affect the concentrations of nutrients, similarities in physicochemical properties between sites indicate that any potential differences in the concentrations of total or dissolved nutrients are likely not driven by these properties and might result from physical and biological processes, including sedimentation and biological uptake.

Nutrient Concentrations

Concentrations of ammonia, nitrate plus nitrite, nitrite, total nitrogen, total dissolved nitrogen, orthophosphate, and total phosphorus in samples collected at the Old River Lake, Wallisville, and Buckeye Island sites during 2016–19 are included in table 1.4.

Nitrogen Concentrations

Nitrate plus nitrite concentrations ranged from <0.040 to 0.663 mg/L at the Old River Lake site, from <0.040 to 0.658 mg/L at the Wallisville site, and from 0.122 to 0.673 mg/L at the Buckeye Island site (table 5; fig. 10A). Total dissolved nitrogen concentrations ranged from 0.40 to 1.20 mg/L at the Old River Lake site, from 0.38 to 1.11 mg/L at the Wallisville site, and from 0.50 to 1.10 mg/L at the Buckeye Island site (table 5; fig. 10B). Total nitrogen concentrations ranged from 0.62 to 1.19 mg/L at the Old River Lake site, 0.43 to 1.29 mg/L at the Wallisville site, and 0.73 to 1.29 mg/L at the Buckeye Island site (table 5: fig. 10C).

At the Old River Lake site, nitrate plus nitrite concentrations and total dissolved nitrogen concentrations were lower in samples collected during low-flow conditions than in samples collected during high-flow events (fig. 10A, B). For three of the four samples collected during low-flow conditions nitrate plus nitrite and total dissolved nitrogen concentrations at the Old River Lake site were less than those at the Wallisville site. These lower concentrations at the Old River Lake site may be associated with various physical and biogeochemical processes, including the transformation and biological uptake of nitrate, nitrite, and other species of nitrogen resulting from extended water residence times (Bowden, 1987; Jansson and others, 1994) and relatively small inputs of nitrogen during low-flow conditions (Lucena and Lee, 2017). Extended water residence times are not usually observed at the Wallisville site during low-flow conditions because of continuous releases from Lake Livingston. The releases may be a continuous source of nutrients in the main channel of the Trinity River during low-flow conditions and might also limit the potential for transformation and biological uptake that result in higher concentrations.

Nitrate plus nitrite and total dissolved nitrogen concentrations were similar between the Wallisville and Buckeye Island sites and were generally higher in samples collected during high-flow events than during low-flow conditions. Total nitrogen concentrations were similar among the three sites for all samples (fig. 10C). For samples collected on September 2 and 6, 2017, concentrations of the three nitrogen species were the lowest among all samples collected during high-flow events. These samples were collected after the peak streamflow caused by Hurricane Harvey. Low concentrations in samples collected during these two dates may be attributed to various factors, including the timing of sample collection and the origin of the water flowing through the system. Because sample collection occurred after the peak streamflow (fig. 11), it is possible that some of the nitrogen mobilized during the storm had already been flushed into Galveston Bay. Also, during Hurricane Harvey, approximately 1,690,000 acre-ft of water were measured at the Liberty site, whereas 1,046,000 acre-ft of water (as measured from the Goodrich site) were released from Lake Livingston. This difference in water volumes between the Goodrich and Liberty sites indicates that approximately 38 percent (644,000 acre-ft) of the measured streamflow at the Liberty site originated predominantly from localized surface runoff. During most high-flow events in the lower Trinity River Basin, effects of localized runoff are typically minor when compared to the volumes of water released from Lake Livingston (Lucena and Lee, 2017). The large amount of localized surface runoff generated by Hurricane Harvey, however, may have affected nitrogen concentrations through dilution, resulting in nutrient concentrations that were less than those observed during high-flow events almost entirely driven by Lake Livingston releases.

To determine the effect streamflow variability may have on the relative amounts of ammonia, nitrate, nitrite, and organic nitrogen at each site, the percentage of the composition of each species of nitrogen was computed for each sample. Censored values for ammonia, nitrate, and nitrite were substituted by using one-half the method reporting limit, a widely used practice, albeit one that has no theoretical basis (Helsel, 1990; Helsel and others, 2020). Maximum likelihood and robust methods are recommended over simple substitution for censored values (Helsel and others, 2020); however, the chosen substitution method was used because censored results represent a relatively small percentage (less than 1 percent for ammonia and nitrite and less than 5 percent for nitrate) of the total nitrogen in each sample and thus do not appreciably affect the overall proportion of each species of nitrogen.

For most samples, organic nitrogen was the predominant nitrogen species, ranging from approximately 40 to 95 percent at the Old River Lake site, 46 to 94 percent at the Wallisville site, and 46 to 82 percent at the Buckeye Island site (fig. 12). Nitrate was the second most predominant nitrogen species, ranging from approximately 3 to 55 percent at the Old River Lake site, 5 to 52 percent at the Wallisville site, and 16 to 52 percent at the Buckeye Island site. The proportion of nitrite and ammonia ranged from approximately 0.1 to 2 percent and 0.1 to 7 percent, respectively, at all three sites.

Lucena and Lee (2017) documented that increases in streamflow associated with Lake Livingston releases typically cause a decrease in the proportion of organic nitrogen and an increase in the proportion of nitrate at the Wallisville site. Shifts in the nitrogen composition are hypothesized to result from changes in the source of nitrogen inputs to the lower Trinity River, which include natural inputs derived from wetlands and forested areas as well as nitrogen from fertilizers carried in runoff in the upper Trinity River Basin. Similar shifts were also observed at the Old River Lake site (fig. 12A). During high-flow events, the proportions of each nitrogen species were similar among the three sites (fig. 12), indicating that the travel path through wetlands and channels surrounding Old River Lake likely does not affect the relative concentrations of the various nitrogen species present in freshwater inflow to Galveston Bay. It is possible that water residence times in Old River Lake during high-flow events are too short to allow transformation and biological uptake of nitrogen at appreciable levels. During low-flow conditions, the proportion of each nitrogen species varied among the three sites in response to the magnitude of the streamflow released from Lake Livingston. When releases from Lake Livingston were not large enough to influence the tidal flow at the Old River Lake and Wallisville sites, such as during October 18, 2016, the proportions of nitrogen species were similar between sites. However, when releases from Lake Livingston were large enough to attenuate tidal flow at the Wallisville site, but too small to divert water into Old River Lake, such as during December 7, 2016, the proportions of organic nitrogen and nitrate were considerably different between the Wallisville and the Old River Lake sites. Results indicate that, similar to the suspended-sediment distribution, the predominant source of the nitrate and organic nitrogen being delivered to Galveston Bay during low-flow conditions is the main channel of the Trinity River and that both the Old River Lake and the main channel of the Trinity River are pathways for nitrate and organic nitrogen entering Galveston Bay during high-flow events.

Phosphorus Concentrations

Total phosphorus concentrations ranged from 0.113 to 0.179 mg/L at the Old River Lake site, from 0.049 to 0.204 mg/L at the Wallisville site, and from 0.107 to 0.200 mg/L at the Buckeye Island site (table 5; fig. 13A). Orthophosphate concentrations ranged from 0.016 to 0.066 mg/L at the Old River Lake site, from 0.012 to 0.060 mg/L at the Wallisville site, and from 0.022 to 0.061 mg/L at the Buckeye Island site (table 5; fig. 13B).

Concentrations of total phosphorus and orthophosphate were lower at the Old River Lake site than at the Wallisville and Buckeye Island sites in three of the four samples collected during low-flow conditions (fig. 13A, B), possibly as a result of longer water residence times allowing a decrease in concentration associated with the settling of the carrier sediment and biological uptake of phosphorus. During the low-flow conditions on October 18, 2016, phosphorus concentrations at the Old River Lake site were at least two times greater than at the Wallisville site. This spike in phosphorus concentrations can likely be attributed to surface runoff from a localized precipitation event near Old River Lake that occurred during the week prior to sample collection (National Oceanic and Atmospheric Administration, 2021).

During high-flow events, concentrations of total phosphorus followed a similar pattern at the three monitoring stations (fig. 13A). This pattern of high total phosphorus concentrations during high-flow events is consistent with previous findings by Lucena and Lee (2017) in samples collected during 2014–16 at the Wallisville site. Total phosphorus concentrations increased proportionally with suspended-sediment concentrations, likely as a result of adsorption of dissolved phosphorus onto sediment particles. Because total phosphorus concentrations are normally closely tied to suspended-sediment concentrations in freshwater and estuarine systems (He and Xu, 2018), and because suspended-sediment concentrations are consistently lower at the Old River Lake site than at the Wallisville site, it would be reasonable to expect total phosphorus concentrations at the Old River Lake site to be less than those at the Wallisville site. However, during high-flow events, total phosphorus concentrations were similar at the Old River Lake and Wallisville sites. Phosphate sorption capacity decreases as particle size increases because fine-sized (smaller than 0.0625 mm in diameter) particles provide a larger surface area (Wang and others, 2006). The concentrations of total phosphorus may also be affected by the size distribution of suspended-sediment particles. At the Wallisville site, suspended-sediment concentrations were greater than at the Old River Lake site, but the proportion of fine-sized suspended-sediment particles ranged from 32 to 60 percent during high-flow events (table 1.4). At the Old River Lake and Buckeye Island sites, fine-sized sediment particles were predominant in all samples collected during high-flow events, ranging from 58 to 100 percent and from 79 to 88 percent, respectively (table 1.4). This variability in particle size distribution likely results in similar concentrations of total phosphorus among monitoring stations during high-flow events. At all monitoring stations, total phosphorus concentrations increased as the concentration of fine-sized particles increased (fig. 14). When fitting a linear regression equation to the data points from each monitoring station, the resulting slopes were similar (0.00033 to 0.00044), indicating that total phosphorus concentrations at the observed range increase at a similar rate at each site in response to an increase in concentrations of fine-sized particles. Because total phosphorus concentrations are closely tied to the fine-sized suspended-sediment concentrations, the main channel of the Trinity River and Old River Lake are likely important sources of phosphorus during high-flow events.

Nutrient Loads

Instantaneous loads of each constituent were computed at the Old River Lake and the Wallisville sites when samples had a corresponding streamflow measurement and reported constituent concentrations that were greater than the method detection limit. To obtain the load amount, in tons per day, the constituent concentration was multiplied by the corresponding instantaneous streamflow and a conversion factor (0.0027) (Porterfield, 1972) (fig. 15).

Instantaneous loads of all nutrients were higher at the Old River Lake site than at the Wallisville site during seven of the nine high-flow events (fig. 15). Concentrations at the two sites were similar, but streamflow was substantially higher at the Old River Lake site. During the remaining two high-flow events (April 25, 2016, and March 1, 2018), nitrogen and phosphorus loads were lower at the Old River Lake site than at the Wallisville site, resulting from lower streamflow and concentrations at the Old River Lake site and indicating that the roles of the main channel of the Trinity River and the Old River Lake in nitrogen and phosphorus delivery to Galveston Bay are likely tied to the magnitude of the event. When streamflows at the Liberty site were less than approximately 20,000 ft³/s, the main channel of the Trinity River was the primary pathway for nitrogen and phosphorus to Galveston Bay. However, once streamflow at the Liberty site exceeded 20,000 ft³/s, the contribution of nutrient loading through Old River Lake to Galveston Bay increased proportionally to the nutrient loading in the main channel; when streamflow at the Liberty site exceeded approximately 50,000 ft³/s, Old River Lake likely became the primary pathway for nutrient delivery into Galveston Bay.

The characterization of the role that the Old River Lake plays in nutrient delivery to Galveston Bay described in this report is based on a few discrete samples representing suspended-sediment and nutrient loads on specific days. Additional continuous monitoring data at the Old River Lake site would allow for a more complete assessment of how the diversion of freshwater into Old River Lake affects the quantity and timing of nutrients entering Galveston Bay from the Trinity River.