Purpose and Scope

This report describes the substrate particle-size distribution, sediment temperatures, shallow subsurface dissolved-oxygen concentrations of beach sediments, and groundwater/surface-water exchange measured at Olsen’s Beach in Lake Ozette from October 2018 to May 2019. These data were collected to assess the effects of a potential strategy for improving the function of beach spawning habitat of sockeye salmon along Lake Ozette’s shoreline. Sediment, dissolved oxygen, and groundwater/surface-water exchange data were collected in three sampling areas of the shoreline—(1) unvegetated shoreline currently used for spawning, (2) vegetated shoreline not known to be used for spawning by sockeye salmon, and (3) shoreline where vegetation was manually removed in autumn 2018. Data from each of the sampling areas measured in October 2018 to May 2019 were compared to determine if the vegetation removal in historical sockeye salmon spawning areas of Lake Ozette decreased fine-grained sediment, increased dissolved-oxygen concentrations above levels needed to support the successful incubation of sockeye salmon embryos, and improved groundwater/surface-water exchange. Data were collected at Olsen’s Beach on the eastern shoreline of Lake Ozette, which is one of two areas of Lake Ozette’s shoreline that currently supports sockeye salmon spawning.

Description of Study Area

Lake Ozette is located in the northwestern part of the Olympic Peninsula of Washington State and has a 30.6-square kilometer (km²) surface area, which makes it the third largest natural lake and largest lake with an unregulated outlet in the State (fig. 1). The only surface-water outlet of Lake Ozette is the Ozette River, which begins at the northern end of the lake at an elevation of 10.3 m and maintains a low-gradient (0.0014 m/m), sinuous path for 8.5 km to its outlet at the Pacific Ocean at Cape Alava. The lake drains a 179-km² area of the coastal plain and foothills of the Olympic Peninsula with a maximum elevation of 591 m at the summit of Sekiu Mountain. Lake Ozette is entirely contained within Olympic National Park, but 75 percent of the drainage basin draining to Lake Ozette is privately owned with the rest publicly owned under the management of Olympic National Park (13 percent) and Washington State Department of Natural Resources (12 percent; Herrera Environmental Consultants, 2006). Except for the Lake Ozette basin in Olympic National Park, the privately owned land and public land managed by Washington Department of Natural Resources is mostly managed as industrial timberland. The largest tributaries to Lake Ozette include Big River, Crooked Creek, and Umbrella Creek, which have contributing drainage basin areas of 59.1, 30.9, and 29.9 km², respectively. Lake Ozette has a mean depth of 40 m and a maximum depth of 98 m (Dlugokenski and others, 1981). Mean annual surface-water elevation of Lake Ozette measured from 1982 to 1994 and 1998 to 2003 was 10.3 m and ranged from a minimum of 9.4 m to a maximum of 12.6 m (Herrera Environmental Consultants, 2006); the mean annual range in surface-water elevation during this time period was 2.3 m. The temperate, maritime climate of Lake Ozette is characterized by warm, dry summers and cool, wet winters where most precipitation in the basin occurs as rainfall. Mean annual precipitation averaged over the Lake Ozette basin from 1980 and 2010 was 260 centimeters (cm) (PRISM Climate Group, 2014) but was not distributed evenly throughout the year with the highest mean monthly precipitation in November (41 cm) and the lowest mean monthly precipitation in July (5.7 cm). Surface-water elevation of Lake Ozette fluctuates in response to seasonal changes in precipitation with median monthly lake levels highest in December after the onset of the autumn and winter rainy season and lowest in September following the prolonged, dry summer (Haggerty and others, 2009). Mean annual precipitation in water year 2019 over the study period was 210 cm (81 percent of the 1980 to 2010 mean annual precipitation). During the winter, winds are predominantly from the south and the west, which results in increased wave energy imparted on the northern and eastern shores of Lake Ozette.

Lake Ozette and its tributaries support several populations of native and non-native, introduced fish that have both anadromous and resident life histories. Anadromous fish that spawn and rear within Lake Ozette include native salmonids, such as sockeye salmon (Oncorhynchus nerka), coho salmon (O. kisutch), chum salmon (O. keta), Chinook salmon (O. kisutch), steelhead (O. mykiss), and sea-run cutthroat trout (O. clarkii). The population of sockeye salmon within Lake Ozette was designated as an Evolutionary Significant Unit (ESU) in 1999 under the Endangered Species Act by NOAA-NMFS (National Oceanic and Atmospheric Administration, National Marine Fisheries Service). Kokanee, a non-anadromous, resident form of sockeye salmon (O. nerka) spawn in tributary streams of the Lake Ozette (Biological Review Team, 2003) but are not included in the Lake Ozette sockeye salmon ESU.

Particle-Size Distribution

The particle-size distribution of the upper 25 cm of substrate was characterized at each of the 12 sampling stations in October 2018 and May 2019 before and after the 2018–19 sockeye salmon incubation period and winter high-lake levels, respectively. At each station, a bulk sediment sample integrating the upper 25 cm of sediment was collected at about 1 m from the dissolved oxygen and subsurface-water temperature sensors to characterize a similar sediment environment. Sediment collection took place at a different location near each station on the first and last day of the sensor deployments; therefore, disturbance of sediment collection on shallow subsurface flow of water at the sensors was assumed negligible. At emergent vegetation stations, the top 25 cm of substrate was collected using a shovel and was transferred into a bucket. Particle sizes larger than 8 millimeters (mm) were sieved and weighed in the field, whereas smaller particles were collected for particle-size distribution analysis at the U.S. Geological Survey–Cascades Volcano Observatory (USGS–CVO). Results from the field and laboratory-measured particle-size distributions were combined to estimate the complete particle-size distribution at each station. Bulk-sediment samples were collected at submergent vegetation stations using a McNeil sampler inserted into the lakebed (McNeil and Ahnell, 1964) from which sediment was extracted by hand to a depth of 25 cm below the lakebed and residual water containing suspended sediment was sampled. Particle-size distribution of sediment larger than 8 mm was sieved and weighed in the field and sediment less than 8 mm was sent to the USGS–CVO sediment laboratory for particle-size distribution analysis. The volume of residual water in the McNeil sampler was measured and its suspended-sediment concentration was analyzed at USGS–CVO. All particle-size data are available in a companion data release (Sheibley and others, 2021).

Dissolved-Oxygen Concentrations

Dissolved-oxygen concentrations in subsurface sediments were recorded hourly from October 2018 to May 2019 using a submersible optical dissolved-oxygen sensor (HOBO U26-001, Onset Computer Corp.) at each of the 12 sampling stations. Dissolved-oxygen sensors were housed inside 50-cm long sections of slotted polyvinyl chloride (PVC) well screen that was capped at both ends and installed into the lakebed (fig. 3A). Part of the well screen was sealed with PVC tape so that the slotted part of the screen exposed to the subsurface environment was limited to a 10-cm interval from 15 to 25 cm below the lake surface. This open interval corresponded to typical depths of egg pockets within gravel nests called redds formed by sockeye salmon spawning (Brooks, 2010), and it was assumed that water flowing through the open interval of the housing of deployed sensors was indicative of subsurface conditions at this depth. The top of each sensor housing was installed flush with the sediment-water interface by removing lakebed sediment using a hand auger. The top of the sensor housing had a screwcap to allow access to the sensor and remove it if needed during the deployment.

During the October 2018–May 2019 deployment, continuous dissolved-oxygen concentrations recorded by submersible optical dissolved-oxygen sensors were independently field checked using a field-calibrated sonde. The sonde also recorded other field parameters including water temperature, pH, and specific conductance. Water adjacent to the submersible optical dissolved-oxygen sensors was sampled through a 15-cm long sample tip made of rigid stainless-steel screen (0.14 mm mesh size) attached to a 0.9–1.2-meter section of polypropylene tubing and secured to the outside of the PVC well housings using a pipe clamp prior to installation into the lakebed (fig. 3B). A short flexible piece of tubing was attached to the end of the polypropylene tubing and sealed between field checks. Field checks took place at the beginning, middle, and end of the deployment by slowly pumping water from the polypropylene sample tubes using a field peristaltic pump at a rate of less than 100 milliliter per minute. The subsurface dissolved oxygen was measured using a field calibrated sonde (Aqua Troll 600; In Situ, Inc.) equipped with a low-volume flow through chamber to measure dissolved-oxygen concentrations in situ, with minimal disturbance. Dissolved-oxygen readings were recorded after several well volumes were removed and readings on the sonde were stable (within ±3 percent). All continuous dissolved oxygen and water temperature data from the submersible oxygen sensors are provided in a companion data release (Sheibley and others, 2021).

Sediment Temperature Monitoring and Modeling

Sediment temperature was analyzed to assess if thermal conditions at the locations of dissolved-oxygen measurements were adequate for salmon egg development. In addition, subsurface temperature profiles were determined at the sampling stations to estimate groundwater/surface-water exchange rates using a low-cost temperature rod (Briggs and others, 2014; Naranjo and Turcotte, 2015; Rosenberry and others, 2016). Temperature rods were constructed from 50 to 100 cm long piece of 1.9-cm plain galvanized pipe and four to five small temperature sensors (iButton model # DS1922L) (figs. 4A–D). Each section of pipe was crimped at the base of the temperature rod and four to five horizontal slots (about 0.6-cm thick) were cut into the pipe at 1, 4, 7, and 11 cm from the top of the temperature rod using a table router. For longer rods, an additional slit at 50 cm was included to better determine groundwater temperatures. Briggs and others (2014) and Rosenberry and others (2016) demonstrated that temperature sensors at these locations are optimal for modeling a range of groundwater exchange rates in aquatic systems. The pipe was filled with spray insulation to minimize thermal conduction within the pipe and a cavity excavated from each one of the horizontal slits to house the temperature sensors. Each temperature sensor was programmed to record temperature at 1-hour intervals and inserted into the horizontal slits of the pipe and sealed with silicone caulk to make them waterproof. The silicone was allowed to completely dry prior to deployment and a length of PVC tape was placed over the pipe where the horizontal slits were located to provide protection when inserting the rods into the lakebed (fig. 4D). Temperature rods were directly driven into lakebed sediments using a small sledgehammer so that the uppermost temperature sensor was installed to a depth of 1 cm. At some stations where coarse substrate prevented direct temperature-rod installation, a pilot hole was first created by driving a 1.9-cm diameter solid steel rod into the lakebed first before insertion of the temperature rod.

The deployment period for the temperature rods was shorter than the time period for dissolved oxygen because the internal memory of the temperature loggers reached capacity on April 9, 2019, and no further data were collected. During the final site visit in May 2019, the temperature rods were removed from the lakebed and returned to the office where sensors from each rod were removed and temperature data were downloaded. Prior to modeling the data, the temperature time-series sets were limited to the period when each station was fully submerged, as determined by the dissolved-oxygen data recorded at each sampling station.

These vertical profiles of subsurface temperatures were used to fit heat-flux models to infer rates of groundwater/surface-water exchange at each station. VS2DH heat-flux models (Healy and Ronan, 1996) that solve a one-dimensional heat-flux equation within saturated sediment were developed using the U.S. Geological Survey program, 1DTempPro V2 (https://water.usgs.gov/ogw/bgas/1dtemppro/), which provides a graphical user interface to run the VS2DH model allowing users to simulate subsurface-temperature profiles to infer groundwater-exchange rates (Koch and others, 2016). Input to each model included vertically distributed temperature data from at least three different depths measured by temperature rods and estimates of physical and thermal properties of saturated sediment obtained from published values (Irvine and others, 2020) and are provided in table 1. The 1DTempPro model did not automatically estimate variable exchange rates; therefore, we assumed constant groundwater exchange during each model simulation and limited the temporal extent of each model simulation to 5- to 7-day periods to reduce the chance of violating the assumption of constant exchange flux. All 1DTempPro model results were published in a data release (Sheibley and others, 2021). All simulations estimated the specific discharge in meters per day (m/d) from the field-collected sediment-temperature data. The 1DTempPro model domain defines upwelling groundwater, flow into the lake, as a negative flux, and downwelling, flow into the lakebed, as a positive flux.

Data Quality and Quality Assurance

All field deployed sensors, including iButton temperature sensors and HOBO dissolved-oxygen/temperature sensors, were calibrated or checked prior to deployment. The calibration of iButton temperature sensors installed within temperature rods was verified prior to deployment using a five-point temperature calibration using a NIST-certified thermometer (Wagner and others, 2006; Conn and others, 2017). Temperature sensors that exceeded a ±0.2 degrees Celsius (°C) tolerance compared to the NIST thermistor were not used in the deployments. HOBO dissolved-oxygen sensors were either calibrated in the laboratory or in the field before the deployment depending on their availability. Seven of the 12 dissolved-oxygen sensors were calibrated in the laboratory and the remaining 5 were calibrated in the field using a similar procedure because they were sent to the manufacturer for service and battery replacement and were not available for laboratory calibration prior to the deployment.

For the laboratory calibration, seven dissolved-oxygen sensors were calibrated according to the manufacturer’s specifications, which included installation of a fresh sensor cap and calibrating of the sensor in saturated air and within a zero dissolved-oxygen solution. Dissolved-oxygen sensor calibrations were performed as closely to the field deployments as possible to ensure that expiration of the sensors, which occurs 7 months after their initial calibration according to the manufacturer, did not occur during the study deployment. After calibration, the dissolved-oxygen sensors were placed in a continuously aerated bucket along with a laboratory calibrated sonde (Aqua Troll 600) for several days to test accuracy (against the calibrated sonde) and sensor variability (when compared against each other). The remaining five dissolved-oxygen sensors that were sent to the manufacturer for service were tested in the field in a similar manner to the seven laboratory-calibrated sensors, except the bucket was not aerated, and the test only lasted about 1 hour before deployment.

Subsurface Particle-Size Distribution

Particle-size distributions of sediment collected from the upper 25 cm of the lakebed were estimated for each of the 12 sampling stations in October 2018 and May 2019. Statistics describing these particle-size distributions, including median (D₅₀) and 90th percentile (D₉₀), are presented in table 2. At each station, measured D₅₀ ranged from medium sand to coarse gravel and D₉₀ was coarse gravel. In general, the finest sediment for each station type (spawning control, treatment, and vegetation control) occurred at stations located farthest from the shore (SC-1, TR-1, and VC-1), and the coarsest sediment occurred at stations closest to the shore (SC-4, TR-4, and VC-4) as measured by both D₅₀, D₉₀, and the percentage of fine-grained sediment less than 4 mm. The change in D₅₀ in the treatment sampling area from October 2018 to May 2019 was less than 3 mm except for at station TR-4 where D₅₀ increased by 14 mm, which was the largest overall increase in D₅₀ at any station; D₉₀ commensurately increased by 18 mm at station TR-4 from October 2018 to May 2019.

Dissolved-Oxygen Concentrations and Other Field Parameters

Dissolved-oxygen sensors were deployed for a 7-month period from October 18, 2018, to May 8, 2019. Prior to deployment, dissolved-oxygen sensors were calibrated in the laboratory or the field depending on the availability of sensors. All laboratory-calibrated dissolved-oxygen sensors were within 0.2 mg/L of the calibrated sonde and 6 of the 7 sensors read within less than 0.1 mg/L of each other, with the 7th sensor always being within 0.2 mg/L of the rest (fig. 5). Field-calibrated sensors showed similar results to the laboratory-calibrated sensors with all field-calibrated sensors within 0.1 mg/L of each other and within 0.2 mg/L or less of the calibrated sonde after the initial few minutes of the test (fig. 6). All these differences were within the USGS specifications for continuous dissolved-oxygen measurements (±0.3 mg/L; Wagner and others, 2006).

During the deployment, three field checks of the dissolved-oxygen sensors were completed—(1) pre-deployment (October 18, 2018, and October 19, 2018), (2) mid-deployment (March 5, 2019), and (3) post-deployment (May 8, 2019). During each of these field checks, subsurface-water temperature, dissolved-oxygen, specific conductance, and pH were determined using a laboratory calibrated sonde. Results from field checks showed that the specific conductance, pH, and temperature of subsurface water was similar across the three sample areas (table 3). Dissolved-oxygen concentrations were low, less than 1.0 mg/L at most stations during the mid- and post-deployment checks. For the initial checks when lake stage was near its annual minimum, adequate water was only present at the most lakeward station of each area (SC-1, TR-1, and VC-1) to obtain a reliable dissolved-oxygen reading, which all exceeded 1.0 mg/L of oxygen. Compared to the subsurface water, lake water was similar in temperature, but much lower in specific conductance, and much higher in dissolved-oxygen concentration (table 3).

At the end of the dissolved-oxygen sensor deployment in May 2019, dissolved-oxygen sensors were retrieved from the lakebed and submerged within a single bucket of lake water to perform a post-deployment comparison of all dissolved-oxygen sensors. Sensors continued to record data for several days before measurements were discontinued and data were downloaded. Variability across the sensors was similar to the pre-deployment inter-sensor comparisons. In general, all sensors were within 0.5 mg/L of each other for the duration on the post-deployment inter-sensor comparison (fig. 7). In addition, the temperature readings from the submersible oxygen sensors during this post deployment were within 0.2 °C of each other. When the dissolved-oxygen sensors were downloaded, communication could not be established with the sensor from station TR-4, the sensor highest up the bank at the vegetation treatment area. Therefore, the data in figure 7 only show results from 11 sensors.

These tests collectively demonstrated that the dissolved-oxygen sensors were accurate when compared to the calibrated sonde and had low variability when compared across the individual sensors. Furthermore, the sensors remained accurate throughout the duration of the deployment as shown by the post-deployment comparison. However, all these tests were made under relatively high oxygen conditions; at no time during these tests did the dissolved-oxygen concentration decrease to less than 6.0 mg/L. For reference, the manufacturer specifications for the accuracy of these sensors was ±0.2 mg/L for dissolved-oxygen concentrations as high as 8.0 mg/L.

Temporal changes in dissolved-oxygen concentrations measured from October 2018 to May 2019 were similar within all three sampling areas (fig. 8). Except a short period at the beginning and end of the 7-month deployment, when the lakebed in the vicinity of the sensor was not fully inundated, all 12 dissolved-oxygen sensors recorded 0.0 mg/L of dissolved oxygen for most of the deployment period. This interpretation is supported by the progressive shoreward inundation of the 12 sensors as lake levels rose during the weeks following the October 2018 dissolved-oxygen-sensor installation—for example, dissolved-oxygen sensors at stations SC-1, TR-1, and VC-1 were submerged in the lake upon installation and dissolved-oxygen concentrations decreased to 0.0 mg/L within days or less at these three stations. In contrast, stations SC-4, TR-4, and VC-4, which were the most shoreward and highest elevation stations, were completely dry during installation. At these stations, dissolved-oxygen concentrations decreased to 0.0 mg/L over a period of 2 months as lake stage increased due to increased precipitation and resulting tributary inflow to the lake. At the end of the deployment, dissolved- oxygen sensors recorded increased concentrations at the highest elevation, most shoreward stations that became emergent in the spring. As lake levels began to decline as early as March 2019, stations SC-4, TR-4, and VC-4 became emergent and samples could not be collected due to the lack of subsurface water.

Field-check data (table 3), collected during the mid- and post-deployment checks, confirm that in situ dissolved-oxygen data were low, with values ranging from 0.46 to 0.97 mg/L in all cases except at station VC-3 in March 2019 and stations SC-4 and VC-4 in May 2019. These field-check data and pre- and post-deployment inter-sensor comparisons showed that all sensors were within 0.5 mg/L of each other. Field check data and inter-sensor comparisons suggest that the dissolved-oxygen sensors were working correctly and that low dissolved-oxygen concentrations during the October 2018 to May 2019 deployment were representative of subsurface conditions. Collectively, these data show that subsurface dissolved-oxygen concentrations did not exceed the 3 mg/L threshold required for successful egg incubation during most of the deployment period.

Sediment Temperatures

Sediment temperatures, as measured by the dissolved-oxygen sensors, were similar within and across each sampling area (fig. 9). After the initial period of variable saturation, sediment temperatures slowly decreased from about 15 °C in October 2018 to about 8 °C in February–March 2019. In March 2019, temperatures from stations SC-4, TR-4, and VC-4 became more variable due to a decrease in lake stage, and temperatures at the other sensors all slowly increased until the end of the deployment in May 2019. At the end of the deployment, sediment temperatures reached similar levels as the beginning of the deployment at about 15 °C and there was no difference among the spawning control, vegetation treatment, or vegetation control areas.

Groundwater/Surface-Water Exchange

Temperature rods were installed at 11 of the 12 sampling stations to measure vertical temperature profiles to estimate groundwater/surface-water exchange. Temperature rods were not installed at station VC-4 due to equipment shortages at the time of installation. At the other 11 stations, continuous vertical temperature profiles were measured at 60-minute intervals. These temperature data were used to calibrate 1DTempPro models to estimate rates of groundwater/surface-water exchange at 8 of the 11 sampling stations where temperature rods were deployed. At station VC-3, temperature loggers were compromised, and data could not be retrieved. At stations TR-2 and TR-4, data from only two of the four loggers were available to model and did not meet data input requirements of 1DTempPro model, which needs continuous temperature data measured at three or more depths.

Results from 1DTempPro model indicated spatial and temporal variability in the rates and direction of groundwater/surface-water exchange during the October 2018–May 2019 deployment. In the spawning control area, groundwater upwelling was simulated at stations SC-1 and SC-4 throughout the entire deployment (fig. 10A). At the two intervening stations, upwelling was modeled at station SC-2 until mid-February 2019 when downwelling was modeled. At station SC-3, downwelling was modeled throughout the deployment period. Besides station SC-4, where specific discharge ranged from -4.97 to -1.08 m/d, specific discharge was relatively constant throughout the deployment period with average specific discharge rates of –0.204, –0.080, and –0.130 m/d for stations SC-1, SC-2, and SC-3, respectively. At the vegetation control sampling area, station VC-1 upwelling was modeled throughout the deployment with specific discharge averaging –0.048 m/d with little variation (fig. 10C). Upwelling also was modeled at station VC-2 throughout the deployment, but there was more variability in specific discharge, which ranged from –2.87 to –0.036 m/d. The vegetation treatment area showed similar behavior as the other two sampling areas (fig. 10B). Station TR-1 indicated upwelling throughout the deployment, and average specific discharge was –0.142 m/d. In contrast, station TR-3 indicated downwelling throughout the deployment with an average specific discharge of 0.109 m/d.

Across area comparisons among the three shoreline classifications were made with stations SC-1, TR-1, and VC-1 because they were under water the entire deployment, had the longest time period of data, and showed a consistent direction of specific discharge throughout the deployment (upwelling, or into the lake). Throughout the incubation period (November 2018–March 2019), the greatest upwelling rate was modeled at the spawning control station (SC-1) and the least amount of upwelling was modeled at the vegetated control station (VC-1), with the vegetation treatment in the middle station (TR-1; fig. 11). Average specific discharge values during this period were –0.216, –0.142, and –0.043 m/d for stations SC-1, TR-1, and VC-1, respectively.