Loch Lomond Reservoir, in Santa Cruz County, California, was created in 1960 through impoundment of Newell Creek by the Newell Creek Dam (fig. 1). The 182-foot-high dam allows the reservoir to store water for use by the city of Santa Cruz, which owns and operates the dam. Several surveys of the reservoir have been completed since the dam was constructed to assess changes in storage capacity due to reservoir sedimentation (Brown, 1973; Fogelman and Johnson, 1985; McPherson and Harmon, 2000; McPherson and others, 2011). McPherson and others (2011) presented an extensive review of the surveys completed prior to 2009, including developing error estimates for the previous surveys. The most recent survey in 2009 indicated a total reservoir storage capacity of 8,650±90 acre-feet (McPherson and others, 2011) and that the storage capacity had decreased by about 320 acre-feet since 1998. McPherson and Harmon (2000) estimated that storage-capacity increased between 1982 and 1998 by 141 acre-feet. In addition to storage estimates, McPherson and others (2011) also documented elevation increases along the reservoir thalweg on the order of 20–30 feet (ft) since dam construction, primarily in the upper reservoir and near the dam. In 2019, the U.S. Geological Survey (USGS), in cooperation with the city of Santa Cruz, completed a new reservoir survey to assess changes since 2009 and to provide information to the city of Santa Cruz for planned modifications to the dam inlet and outlet works. This report describes the results of the reservoir survey and provides further evaluation of changes in reservoir storage capacity through time.

The equipment and methods used to collect the topographic and bathymetric data and bed-sediment samples are similar to those described in Marineau and others (2020). The bathymetry and shoreline topography data were collected May 29–31, 2019; the bed-sediment samples and some additional survey points were collected on December 17, 2019. The following equipment was used to collect the survey data (fig. 2):

Data from this study are publicly available as a U.S. Geological Survey ScienceBase data release (Whealdon-Haught and others, 2021). The following results are provided: (1) 1-m² DEM points file; (2) tables of storage capacity and surface area as a function of water-surface elevation at 0.1-ft increments; (3) table of bed-sediment results; and (4) sound-velocity profiles.

Sound-speed profiles indicated differences in vertical structure, such that surface warming led to greater sound velocity near the surface and the presence of a well-defined thermocline in about the upper 30 feet of the water column.

Independent elevation measurements were in close agreement with elevations from the SLMS, as shown in table 1. Elevation differences ranged from –0.29 to 0.26 ft, with mean and median of signed differences equal to –0.08 and –0.13 ft, respectively. The mean and median of the unsigned differences (absolute values) were 0.16 and 0.18 ft, respectively. These values were used to estimate the uncertainty in storage capacity, as described in the “Discussion of Storage-Capacity Changes from 2009 to 2019” section that follows. Here, we assumed the error from depth measurements by the SLMS was less than the error from the vertical resolution of the GNSS PosPac corrections to the INS.

Figure 6 shows the DEM from the survey (1-m² grid spacing) along with the spillway elevation contour. Figure 7 shows a higher resolution (0.1-m² grid spacing) DEM for the region around the existing outlet works at the time of survey, along with the elevation contours for the new outlets installed in the summer of 2020.

Figures 8 and 9 show the relations between water-surface elevation and reservoir-storage capacity and surface area for the 2019 survey. For the spillway elevation of 580.87 ft above NAVD 88 (577.18 ft above NCDD), the total storage capacity of the reservoir below the spillway crest elevation was 8,770 acre-feet, and the surface area at full pool was 168 acres. The dead storage (below the lowest outlet elevation of 480 ft above NCDD) was 390 acre-feet, leaving 8,381 acre-feet of live (or usable) storage. The uncertainty in storage capacity was estimated by multiplying the largest difference between the SLMS and an independent elevation measurement from table 1 (0.29 ft) by the total surface area at full pool (168 acre-feet), which equals 49 acre-feet (0.6 percent of total storage capacity); thus, the 2019 estimate of total storage capacity is 8,770±50 acre-feet.

The results of the bed-sediment sampling are shown in figure 10, which indicates that the bed sediment in the reservoir (in the top 2 inches of the surface) is primarily composed of silt and clay sized particles, consistent with the findings of McPherson and others (2011) and Fogelman and Johnson (1985). At the very upstream end of the reservoir, the sediment is about 30–50-percent sand, but the sand content drops off quickly as the reservoir becomes deeper. The rapid downstream decrease in particle size is a common feature of depositional environments (Schleiss and others, 2016). One sample contained appreciable quantities of gravel-sized sediment, mid-way in the reservoir at site 17 in figure 4. Although it is not completely clear why gravel is exposed here, site 17 is in a narrow section of the reservoir with a tributary just upstream on river right that is a potential gravel source.

The storage capacity measured in 2019 was 8,770±50 acre-feet, which is 125 acre-feet, or 1.4 percent, greater than the published storage capacity from the 2009 survey (McPherson and others, 2011) of 8,650±90 acre-feet. Although this amount of change is within the uncertainty bounds of the surveys (the upper bound of 2009 is slightly higher than the lower bound of 2019), a reservoir gaining storage capacity over time is counterintuitive and thus warrants further exploration.

One potential explanation for the difference in storage is that the elevation of the spillway used to compute the 2009 storage capacity was too low. A spillway elevation of 577.5 ft NGVD 29 was used (McPherson and others, 2011) for the 2009 storage capacity calculation, but the leveling survey of Bowman and Williams (appendix 1) found the spillway elevation to be 578.14 ft NGVD 29, a difference of 0.64 feet. Extension of the 2009 elevation-storage curve to this updated spillway elevation yields an additional 111 acre-feet of storage capacity, increasing the total estimate of storage capacity to 8,760 acre-feet. This revised estimate is nearly identical to the 2019 storage capacity (only 14 acre-feet less). This indicates that the change in total storage capacity between 2009 and 2019 was less than the uncertainty in the survey methods (similar methods were used for both surveys). Consideration of the spatial variability of change, however, as well as information on watershed sediment yields, can provide further context for interpreting the differences between 2009 and 2019, as described in the “Discussion of Long-Term Reservoir Storage and Watershed Sediment Yield” section that follows. The spillway elevation of 577.5 ft NGVD 29 used in the 2009 survey also was used in the 1998 survey; however, the 1998 storage capacity was not revised here because the change based on the updated spillway elevation is much less than the storage uncertainty for that survey. Additionally, methods used in 1998 survey deviate from that of the 2009 and 2019 surveys.

Figure 11 shows the difference between the 2019 survey and the 2009 survey for overlapping areas. A positive value represents an increase in elevation since 2009. The results show that areas of deposition are in the upper reservoir, whereas the lower reservoir shows decreased elevation. Most of the reservoir shows elevation changes between –0.82 and 0.82 m, which is similar to the range of error in our survey (table 1 and earlier discussion). Additionally, any datum inaccuracies from the 2009 survey using a lower reference benchmark for the survey could bias the differences toward negative elevation changes.

The changes between 2009 and 2019 were further explored by examining the storage changes as a function of elevation. All previous surveys (1982, 1998, and 2009) documented sediment deposition at the upstream end of the reservoir, where depths are low and surface elevations are high (Fogelman and Johnson, 1985; McPherson and Harmon, 2000; McPherson and others, 2011; respectively). To compare changes as a function of depth or elevation, the WSE-storage curves for 2009 and 2019 were used to compute storage in 2-ft elevation increments (the increment from table 1 of McPherson and others, 2011), and the change in storage for each increment was computed. The results in figure 12 show that the change in storage was measurably different by depth. In shallow parts of the reservoir (depths less than about 30 ft), storage decreased from 2009 to 2019, indicating sediment deposition. In areas of the reservoir where depths exceeded 30 ft, small increases in storage were observed consistently over a wide range in depths. Although the storage losses in the shallow areas were greater per unit area, the area of deep water (110 acres) was much larger than the area of water less than 30 ft deep (58 acres), resulting in a near balance between losses in the shallows and gains at depth. As shown in figure 12, the storage loss at depths less than 30 ft was 68 acre-feet, whereas the storage gain at depths greater than 30 ft was 82 acre-feet, resulting in the 14 acre-feet total storage increase between 2009 and 2019 noted previously. The average vertical changes computed from the volume change and the area of reach region were +1.2 ft in shallow areas and –0.75 ft in deep areas.

Differences in storage between 2009 and 2019 surveys are further explicated in figure 13, which shows histograms of elevation changes based on differences between 1-m² DEMs for the shallow and deep areas (positive elevation change indicates deposition). For the shallow areas, the mean and median differences were small and slightly negative (–0.01 and –0.39 ft, repectively), but the large elevation differences in the upper reservoir were apparent in the right-tail of the distribution. For the deep areas, there were very few positive differences (deposition), and the mean (–1.3 ft) and median (–1.1 ft) values both indicated decreasing elevations between 2009 and 2019.

At least two possible reasons exist to explain the unexpected storage increases in deep areas of the reservoir: (1) surveying error in the vertical measurements and (2) consolidation of deposited sediment over time. With respect to surveying error, the mean and median differences in the deep areas were both about 1 ft, which is greater than the average vertical survey error reported for the 2009 and 2019 surveys of less than about 0.5 ft. Given the similar magnitude of the changes and the survey errors, the possibility that the storage changes in deep areas are a result of survey error should not be ruled out. A second possibility is that the consolidation of deposited sediment is gradually lowering bed elevations over time. Consolidation is a process known to occur in reservoirs over long periods (Morris and others, 2008), whereby the bulk density of the deposited sediment increases over time. Consolidation is greatest when the deposited sediment is silt- and clay-sized (Lara and Pemberton, 1963), such as in Loch Lomond Reservoir (particularly in the deeper areas closer to the dam; fig. 11). Thus, one potential interpretation of the increased storage (for deep areas but also just slightly for shallow areas) is that consolidation is gradually lowering bed elevations, and the consolidation rate is greater in deep areas, where the deposited sediment is the finest. This interpretation should be considered speculative because the survey error hypothesis is also very probable. Further investigation of the reservoir deposits, such as by coring in order to assess the vertical variability in bulk density, could help resolve this issue.

Two methods can be used to compute storage losses in a reservoir: (1) directly as the difference in storage between two surveys, as detailed earlier for 2009 to 2019, and (2) indirectly from information about sediment inflows to the reservoir, deposit density, and reservoir trap efficiency. The direct method is more comprehensive when modern methods have been used for the repeat surveys. As noted in McPherson and others (2011), however, older surveys can have substantial uncertainty that often precludes meaningful storage-change assessments. The indirect method can provide context for interpreting the direct storage-change calculations, particularly for older surveys, depending on the information available about watershed sediment yield and reservoir sediment.

To assess sediment yields delivered to the Loch Lomond Reservoir, the USGS scaled sediment yields in the San Lorenzo River watershed to that of the Loch Lomond Reservoir. The period from 2009 to 2019 was reasonably similar to the period of record hydrologically. The mean flow at the San Lorenzo River at Big Trees streamgage (USGS station 11160500) for the entire period of record (1936–2019) was 131 cubic feet per second (ft³/s); the mean flow for 2009–19 was 121 ft³/s but included the highest single-year mean flow of 405 ft³/s in 2017 (U.S. Geological Survey, 2019). Although the highest mean flow on record was during 2017, the 2017 peak flow of 19,000 ft³/s is only the sixth highest on record and was substantially less than the historic peaks during December 1955 (30,400 ft³/s) and January 1982 (29,700 ft³/s). Daily reservoir WSEs and daily flows at the Big Trees streamgage for the period 2009–19 are shown in figure 14, illustrating the effects of low inflows during 2014–15 as well as the spilling events during wet years, particularly in 2017. There was at least one notable runoff event (2017) between the surveys that may have delivered substantial sediment from the watershed to the reservoir, causing decreased storage capacity.

Estimates of annual sediment yields available for the San Lorenzo River are based on measurements of suspended-sediment loads at the USGS streamgages at Big Trees and Santa Cruz (USGS stations 11160500 and 11161000, respectively; U.S. Geological Survey, 2019) and represent two time spans: 1973–82 (Best and Griggs, 1991) and 2009–17 (Conaway and others, 2013; East and others, 2018). The earlier period is generally considered to have had higher sediment yields than the more recent period, reflecting anthropogenic activities and the effects of the 1982 flood in the record (East and others, 2018). If it is assumed that the watershed-sediment yields for the San Lorenzo River can be scaled by the Loch Lomond Reservoir drainage area to estimate sediment inflows to the reservoir, the storage loss in the reservoir can be computed from the following equation: SL= Y s A LL T D×100 (1)where SL

is the storage loss (volume/time),

The inherent assumptions with this method are that the yields from San Lorenzo River represent that of Loch Lomond Reservoir and that the Loch Lomond Reservoir does not influence downstream yields. Loch Lomond Reservoir is of the same geology, vegetation, and topography as that of the larger, surrounding San Lorenzo River watershed, which indicates that samples from the San Lorenzo River should be representative. Nevertheless, we use a large uncertainty envelope to account for any unknown bias in our assumption. Lastly, because Loch Lomond Reservoir watershed is so small relative to San Lorenzo River watershed (approximately 6 percent), it seems safe to assume that downstream influence on sediment yields is minimal.

Bulk density was computed from the Lara and Pemberton (1963) method using the sediment-size fractions from the bed-sediment sampling (fig. 10); density ranged from about 62.4 pounds per cubic foot (or 1,000 kilograms per cubic meter; kg/m³) in the middle and lower reservoir to about 74.9 pounds per cubic foot (or 1,200 kg/m³) in the upper end, where sand content is higher. The median of all samples (65.5 lb/ft³ or 1,050 kg/m³) was used for all subsequent calculations. Trap efficiency was assumed to be 100 percent, which is consistent with Brown (1973), who estimated trap efficiency of 95 percent using the Brune (1953) method. The reservoir is actively managed for water quality, which results in a high trap efficiency, because the city of Santa Cruz switches to a different elevation outlet if sediment is observed to be passing through the dam (Ryan Bassett, Water Resources Analyst, city of Santa Cruz, written commun., June 10, 2020). Substituting the values for A LL , D , and T in equation 1 and applying the necessary unit conversions results in the following: SL =0.0165⋅ Y s (2)where

The highest annual sediment yield for the San Lorenzo River watershed was 5,068 t/yr/km², in 1982, measured at the Big Trees streamgage (USGS station 11160500), which equates to 84 acre-feet of storage loss when equation 2 is applied. For the 10-year period from 1972 to 1983, when sediment yield was measured at the Big Trees streamgage, the total storage loss computed from equation 2 was 160 acre-feet, or 16 acre-feet per year. For 2017, the year with the highest sediment yield in the recent measurements at the Santa Cruz streamgage, the sediment yield was 1,440 t/yr/km², which equates to 24 acre-feet of storage loss. For the period from 2009 to 2017, for which there are 6 years of available sediment yield data (2009, 2010, 2014–17; Conaway and others, 2013; East and others, 2018), application of equation 2 results in a total storage loss of 29 acre-feet, or about 5 acre-feet per year. These estimated storage losses reinforce the difficulty of detecting decadal-scale storage changes from repeat surveys: the decadal-scale changes can be similar in magnitude to survey uncertainty (86 acre-feet for 2009 survey and 49 acre-feet for 2019 survey).

To make an estimate of the total storage loss since dam construction, the available sediment yield data from the San Lorenzo River watershed and equation 2 were used to hindcast reservoir storage capacity from 2019 back to 1960 as follows. First, the storage loss for all years with sediment-yield data (16 years) was computed using equation 2. Second, relations between annual storage loss and the mean annual flow at the Big Trees streamgage (USGS station 11160500, U.S. Geological Survey, 2019) for the two periods for which there were data (1973–82, 2009–17) were developed (linear regressions to log transformed data; R²=0.96 and 0.98, respectively, for the two periods). These two relations (fig. 15 inset) were then used to fill years without data as follows: 1960–72 filled with the 1973–82 relation; 2009–19 missing years filled with the more recent relation; 1983–2008 filled with a weighted average of the two relations that linearly transitions between the two. Computed storage-loss data are provided in table 2. This procedure provided estimates of the storage loss for every year from 1960 to 2019 (fig. 15), which were then used to hindcast the storage capacity from the measured capacity in 2019. For the period 2009–19, the hindcasting produced an estimated total storage loss of 42 acre-feet, which is broadly consistent with the 68 acre-feet of storage loss computed for shallow areas based on the repeat surveys.

The results of this hindcast are shown in figure 15 along with previous estimates of storage capacity summarized by McPherson and others (2011, table 3). Although the uncertainty in the hindcasted storage capacity is unknown (but likely large), a ±50-percent uncertainty envelope is included on figure 15 to illustrate how uncertainty in the 2019 storage-capacity estimate compares to the uncertainty in the previous estimates of storage capacity. The older surveys have a much larger uncertainty due to survey methods and the spatial uncertainty associated with those methods. The hindcasting produced an estimate of about 9,300 acre-feet for the original capacity of the reservoir, ranging from about 9,000 to 9,600 acre-feet for the 50-percent uncertainty envelope embedded into the load estimates (which is cumulated in the hindcast). This hindcasting corresponds to a storage loss of about 500 acre-feet (or 5 percent) from 1960 to 2019, or an average annual loss of about 8–9 acre-feet per year. The annual loss ranged from near-zero to as high as 80 acre-feet in 1982–83, with a more recent high of 24 acre-feet in water-year 2017. A water-year is defined as the period from October 1st of one year through September 30th of the following year and is named for the year in which it ends.

In May 2019, the U.S. Geological Survey, in cooperation with the city of Santa Cruz, surveyed Loch Lomond Reservoir to assess the storage capacity and sedimentation rates in the reservoir. Survey methods combined sonar soundings to measure bathymetry, lidar scans to measure near-shore topography, and bed-sediment samples to understand reservoir bed-material size. The survey produced a bare-earth digital elevation model of the reservoir at a resolution of 1 square meter or better to elevations at or above the reservoir spillway elevation and provided the coverage needed to estimate storage capacity. Additionally, we compared the 2019 survey to historical surveys and provided a hindcast estimate of sediment yield to the reservoir to evaluate long-term changes in storage-capacity. The results are listed as a summary.