Bathymetric and topographic lidar surveys were done in 2017 (Andrews and others, 2018) and 2018, respectively, and integrated into a seamless topobathymetric digital elevation model (TBDEM; Poppenga and others, 2020) to determine storage capacity and areal extent of Lake Powell. This report (1) describes the methods used to modify an existing TBDEM of Lake Powell (Poppenga and others, 2020) to eliminate potential biases in storage capacity computations; (2) describes methods used to calculate the elevation-area-capacity relationships derived from the modified version of the TBDEM (Jones and Root, 2021); (3) presents the updated elevation-area-capacity relationships and associated error assessments; and (4) compares these updated elevation-area-capacity relationships with historical datasets and estimates from pre-impoundment (Bureau of Reclamation, 1963) and the 1986 survey (Ferrari, 1988). Herein, our results provide contextual basis for analyzing the expected reservoir life of Lake Powell, though any such presumptions are beyond the scope of this study and are not addressed. Explicit comparisons with operational elevation-area-capacity tables, which are independently calculated, are also not addressed because the calculation methodologies are not directly comparable. Error in elevation-area-capacity relationships is estimated, with vertical accuracy in the 2017 bathymetric dataset providing the most reasonable error for the TBDEM. Additional errors could be due, in part, to differing survey methodologies. Sediment deposition is recognized as the principal cause for storage loss over time; however, modes and rates of sedimentation are not considered.

Lake Powell is located on the border of Utah and Arizona, with maximum extents that flood nearly 200 miles upstream on the Colorado River. The lands surrounding the reservoir, the Glen Canyon National Recreation Area, are under the stewardship of the National Park Service and Reclamation (fig. 1). Commissioned in 1956, under the Colorado River Storage Project, the 710-feet (ft) Glen Canyon Dam was completed in March 1963 and rises to its spillway crest elevation of 3,717.91 ft above the North American Vertical Datum of 1988 (NAVD 88), with full-pool storage designated at 3,702.91 ft above NAVD 88. A historic maximum reservoir elevation of 3,708.34 ft above NAVD 88 occurred in spring 1983 when the Glen Canyon Dam spillways were damaged and temporarily closed during a period of particularly heavy runoff and high inflow (figs. 2A, B). Though the reservoir remained at nearly full pool for parts of the 1980s and 1990s, the elevation has fluctuated for much of Lake Powell’s history (fig. 2A).

The reservoir elevation for Lake Powell has been reported daily since December 28, 1963, by Reclamation, in addition to other parameters such as storage, inflow, and outflow. Reservoir elevation is recorded hourly, relative to the National Geodetic Vertical Datum of 1929 (NGVD 29), in a stilling well within Glen Canyon Dam, and the last value of each day, at 11:00 p.m. (Mountain Standard Time), is reported (Bureau of Reclamation, 2022; available for download at https://data.usbr.gov/). The USGS deployed a radar-based, water-stage recorder on the bridge next to the west forebay of Glen Canyon Dam on April 18, 2019 (USGS Station ID [STAID] 09379900, available at https://waterdata.usgs.gov/ut/nwis/inventory/?site_no=09379900; U.S. Geological Survey, 2021). Reservoir elevation is recorded at 15-minute intervals relative to NGVD 29 and NAVD 88. Independent survey errors will result in minor differences between reservoir elevations reported by Reclamation and USGS. A formal comparison of data collection methodologies and error quantification is outside the scope of this report; an evaluation of elevation-area-capacity relationships reported herein relative to real-time water surface elevations is not investigated. Additional surveying, which involved examining datums by perpetuating elevation to a variety of objective points from fiducial benchmarks, was done to characterize the differences in datums at Glen Canyon Dam (Gibson and others, 2021). This survey provided additional verification of vertical accuracy in the topobathymetric dataset (Poppenga and others, 2020).

The main tributaries of Lake Powell are the Colorado and San Juan Rivers, with substantially lesser contribution from the Escalante and Dirty Devil Rivers (fig. 1). The USGS has maintained streamgage records for all primary inputs dating to the early 20th century, providing a long-term streamflow record for reservoir monitoring and hydrological context above Glen Canyon Dam (U.S. Geological Survey, 2021). The combined discharges from the nearest gages to Lake Powell for the Colorado (STAID 09180500), Green (STAID 09315000), San Juan (STAID 09379500), Escalante (STAID 09337500), Dirty Devil (STAID 09333500), and San Rafael (STAID 09328500) Rivers approximate an expected total measured inflow (fig. 2B) that is independent from derived inflow by Reclamation (2022). Fluctuations in reservoir elevation have been historically proportional to fluctuations in combined discharge.

The current areal extent of the reservoir at full pool is approximately 250 square miles (mi²), reaching Cataract Canyon on the Colorado River and Clay Hills Crossing on the San Juan River (fig. 1). The reservoir and surrounding Glen Canyon National Recreation Area are classified as arid desert, with an average annual precipitation of 6 inches and summer temperatures up to 110 degrees Fahrenheit (°F; National Park Service, 2021b). Because of the climate and its porous sandstone substrate, losses to evaporation and bank storage are substantial (Myers, 2013; Friedrich and others, 2018). Stable isotope analysis of surface water entering Lake Powell indicates that little evaporation occurs in the upper Colorado River but increases moving downstream, particularly in Lake Powell and Lake Mead (Guay and others, 2006). Reclamation (1950) estimated an annual evaporation rate of 63.0 inches during the planning stages of Lake Powell. Using evaporation data collected between May 1973 and December 1974, the average annual total evaporation at Lake Powell was estimated at 69.48 inches by using mass-transfer methods (Jacoby and others, 1977). Reclamation (1986) estimated annual average evaporation at 68.32 inches for January 1965 through May 1979, incorporating the findings of Jacoby and others (1977) to calibrate its mass-transfer coefficient. Because evaporation at Lake Powell has not been recently studied, estimates at Lake Mead could be used as a proxy at Lake Powell given the reservoirs’ proximity and similar reservoir characteristics. Annual evaporation estimates at Lake Mead, using the eddy-covariance method, were 74.07–81.65 inches between March 2010 and February 2012 (Moreo and Swancar, 2013) and 74.65 inches between March 2010 and April 2019 (Earp and Moreo, 2021). Cumulative water loss to bank storage in Lake Powell is estimated at approximately 15,000,000 acre-feet through 2011 (Myers, 2013) or 300,000 acre-feet per year since Glen Canyon Dam closed in 1963. Releases from Glen Canyon Dam to Lake Mead account for approximately 90 percent of the flow into the Lower Colorado River Basin (Ostroff and others, 2017; Lukas and Payton, 2020; McCabe and others, 2020; Tillman and others, 2020).

The distinctive red rocks of the Colorado Plateau characterize the regional bedrock and range in age from Late Pennsylvanian to Late Cretaceous periods (318–66 million years ago). Cliff-forming units, primarily the Navajo Sandstone and Wingate Sandstone of the aptly named Glen Canyon Group, construct the deep and narrow canyons of Lake Powell. Incision of the Colorado River into the bedrock over the last 5 million years formed Glen Canyon (Anderson and others, 2010), which was renowned for its natural beauty (Powell, 1875). The region is culturally significant to Native Americans, and Navajo Mountain, a prominent laccolith south of the confluence of the Colorado River and San Juan River, is an important landmark in Navajo and Hopi history (Luckert, 1977; Bernardini and others, 2021; Navajo Nation Parks & Recreation, 2022).

Reclamation published two surveys with estimates of elevation-area-capacity relationships of Lake Powell: (1) the original, pre-Glen Canyon Dam storage estimates (Bureau of Reclamation, 1963) and (2) an extensive range-line survey completed in the summer and fall of 1986 (Ferrari, 1988).

Detailed contour maps of Glen Canyon were commissioned during the planning stages of Lake Powell to estimate storage capacity (Fairchild Aerial Surveys, Inc., 1947; Alster and Associates, Inc., 1959; Gessel and Rutledge, 1962). These contour maps were recently digitized and developed into a digital elevation model (DEM; Root and others, 2019). The Colorado and San Juan River arms were separately surveyed and had contour intervals of 10 and 20 ft, respectively. Area and storage capacity estimates were calculated at 1.0-ft and 0.01-ft elevation intervals, respectively. Linear interpolations were used for intermediate intervals in the storage capacity calculations (Bureau of Reclamation, 1963).

A bathymetric survey completed by Reclamation (Ferrari, 1988) used the range-line method, as described by Blanton (1982), to collect 409 bank-to-bank cross-sections of Lake Powell between April 1986 and July 1987. Though previous range-line surveys had been done between 1968 and 1973 (Bureau of Reclamation, 1973), these surveys were limited in their coverage. The 1986 survey produced the first reservoir-wide bathymetric data since impoundment. The complexity of the environment and inherent limitations of early positioning systems warranted the use of a variety of techniques across the reservoir to georeference the measured reservoir depths, which are described in Ferrari (1988). A bathymetric contour map was developed from range-line data and used to calculate elevation-area-capacity relationships with the Reclamation Area-Capacity Computation Program (Bureau of Reclamation, 1985). Abbreviated tables from this study were published at elevation intervals of 20 ft (Ferrari, 1988).

Several studies have been done on sedimentation and storage loss in Lake Powell, though they do not include reservoir-wide estimates for storage capacity. To monitor sedimentation, Reclamation performed a series of sediment monitoring range-line surveys of the deltaic regions of the Colorado and San Juan Rivers between 1968 and 1973 (Bureau of Reclamation, 1973; Lazenby and Nelson, 1976). Other early surveys were done through the Lake Powell Research Project (Spydell, 1975; Condit and others, 1978; Potter and Drake, 1989), a National Science Foundation-funded, interdisciplinary collaboration that included natural and social scientists whose purpose was to observe the effects of water management in the Lake Powell region. As part of sediment and water-chemistry studies in Lake Powell, the USGS completed several single-beam, longitudinal profiles along the river thalwegs (Twichell and others, 2001; Hart and others, 2005; Hornewer, 2014). The first multibeam bathymetric survey was done in 2005 (Clarke and others, 2005; Pratson and others, 2008), though storage capacity estimates were not calculated.

A TBDEM is a continuous representation of subaerial topography and submerged bathymetry. Poppenga and others (2020) created a TBDEM of Lake Powell from four data sources: (1) 2017 2-meter (m) multibeam bathymetry (Andrews and others, 2018); (2) 2018 1-m topographic lidar point-cloud dataset contracted by Reclamation; (3) 2-m historical digital elevation model (DEM) interpolated from 1947 and 1959 contour maps (Root and others, 2019); and (4) 10-m interpolated topography where gaps exist between data sources (fig. 3). These datasets were mosaiced together by generating spatial seamlines with a blending width of 5 m. Where overlap between datasets existed, topographic lidar was given the highest priority, bathymetry the second highest priority, historical DEM third highest priority, and interpolated gap-filling topography had the lowest priority. Priority was based on cell size of the source data.

A dual-head Reson T20-P multibeam echosounder was used to collect reservoir-wide high-resolution bathymetry between October 8 and November 15, 2017 (Andrews and others, 2018). Reservoir water surface elevations ranged between 3,629.18 and 3,631.20 ft above NAVD 88 during the bathymetric survey. Differential Global Positioning System (DGPS) data and vessel roll, pitch, heave, and yaw were collected with an inertial navigation system and depth data collected with the multibeam echosounder were integrated during the survey using HYPACK data acquisition software (version 2017, 17.1.3.0; https://www.hypack.com/). Post-survey, the raw depth data were processed with Computer Aided Resource Information System Hydrographic Information Processing System (CARIS HIPS; versions 10.2 and 10.4; http://www.teledynecaris.com/en/products/hips-and-sips/), which included applying an adjustment for changes in sound velocity with depth and removing erroneous points, such as noise, in individual multibeam swaths. The raw DGPS data were processed with POSPac Mobile Mapping Suite software (version 8.1; https://www.applanix.com/products/pospac-mms.htm), which uses Applanix SmartBase technology to improve the horizontal and vertical accuracy of the navigation solution. The improved solution was applied to the depth data in CARIS HIPS. The post-processed bathymetric dataset was exported as 6.6-ft per pixel ASCII files referenced to Universal Transverse Mercator (UTM) Zone 12N, World Geodetic System 84 (WGS 84) and WGS 84 ellipsoidal heights. Vertical accuracy of the post-processed bathymetry dataset is approximately 1 percent of water depth, which ranged between approximately 15 and 500 ft. The inertial navigation system used in the survey has an additional theoretical vertical accuracy on the order of hundredths of an inch. Vertical transformations from WGS 84 to NAVD 88 during post-processing corrections introduced an additional ±3.0 inches of vertical uncertainty. Horizontal positioning of the raw data is accurate to 1.6–6.6 m but may be as accurate as less than 4 inches after post processing.

The lidar topographic data were acquired during a 2-day airborne survey on April 2 and April 3, 2018, and completed by The Atlantic Group, LLC (https://www.atlantic.tech), under contract by Reclamation. This dataset is not independently published as of the release of this report. A Pacific Aerospace PAC750XL (N750VX) outfitted with a Leica ALS70-HP topographic lidar system was used for data collection, with a maximum flying height of 11,500 ft above ground level. The vertical accuracy of the point cloud and bare-earth data were assessed to the American Society for Photogrammetry and Remote Sensing Positional Accuracy Standards for Digital Geospatial Data (American Society for Photogrammetry and Remote Sensing, 2015). There were 28 distributed check points, including 21 non-vegetated and 7 vegetated, with ground-cover classifications that included open terrain, urban terrain, bare earth, brush, and high grass) measured to assess the accuracy of the lidar data. Vertical error of all check points at the 95-percent confidence level ranged between 0.092 and 0.134 m (0.3018–0.4396 ft) and root mean square error in the vertical direction (RMSEz) ranged between 0.048 and 0.068 m (0.1575–0.2231 ft). In the inundated river and delta regions, where lidar did not penetrate, the DEM was hydro-flattened so that the water surface behaved like a lake rather than a river with a gradient. A constant elevation value of 3,612.76 ft above NAVD 88, representing the water surface elevation of the reservoir during the survey, was applied to these areas.

Gaps between the 2017 bathymetry and 2018 lidar datasets were inevitable in shallow regions of the reservoir where the multibeam vessel was incapable of accessing and lidar could not penetrate the water surface. Interpolation errors in the TBDEM appear at the location of those gaps because of a lack of topographic or bathymetric data. To create a continuous surface, the gaps were seamlessly assimilated with the U.S. Geological Survey Coastal National Elevation Database methodology of interpolation (Danielson and others, 2016) or filled with a pre-Glen Canyon Dam DEM (Root and others, 2019). Additionally, a hydro-flattened elevation of 3612.76 ft above NAVD 88 was applied to the water surfaces of the Colorado and San Juan Rivers upstream from the deltas where water was too shallow for the multibeam vessel to access and lidar data recovery was poor. The hydro-flattened water surface in the rivers was not corrected by Poppenga and others (2020) because there was no alternate data source and further manipulation of topography would introduce additional error; thus, the hydro-flattened surface was the most transparent and decipherable solution.

Root and others (2019) digitized the pre-Glen Canyon Dam contour maps (Fairchild Aerial Surveys, Inc., 1947; Alster and Associates, Inc., 1959) and created a DEM of Glen Canyon prior to the impoundment of Lake Powell. The Colorado River arm, from the site of Glen Canyon Dam to Cataract Canyon, was surveyed in 1958 and 1959 at a contour interval of 10 ft. The San Juan River arm was surveyed in 1947 at a contour interval of 20 ft from the confluence of the San Juan and Colorado Rivers through Mexican Hat, Utah. Horizontal data were transformed from the State Plane Coordinate System, based on the North American Datum of 1927 (NAD 27), to Zone 12N of the Universal Transverse Mercator coordinate system, North American Datum of 1983 with the National Adjustment of 2011. Vertical data were transformed from the NGVD 29 to NAVD 88 using the correction rasters developed for the National Oceanic and Atmospheric Administration (NOAA) VDatum vertical datum transformation tool (National Oceanic and Atmospheric Administration, 2019). A hydrologically corrected, 2-m DEM was created from the digitized contours with the Topo to Raster tool in ArcMap (version 10.6.1; https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview). Vertical accuracy with respect to the original contour maps was reviewed, though a formal error analysis was not performed. The vertical error in this dataset is, at least, equal to the contour interval (10 or 20 ft) that represents that area of the DEM.

The 2017 bathymetry and 2018 lidar topography do not always overlap or abut. Sizeable gaps between the datasets were filled with interpolated topography to create a continuous surface. The edges of the bathymetry and lidar topography were converted to points and then interpolated within the gaps at a 33-ft (10-m) cell size using ArcGIS Topo to Raster. This methodology was primarily utilized in the northeast region of the dataset. Poppenga and others (2020) include spatial metadata of source inputs, and Jones and Root (2021) include an updated version of this metadata with the modified TBDEM.

The assimilation of interpolated data, incorporation of historical data, and hydro-flattening technique into the TBDEM provides the topobathymetric surface in the absence of data but will cause inconsistencies with the true bathymetric surface that are difficult to quantify. Though these inconsistencies may misrepresent a small fraction (less than 10 percent) of the total TBDEM area, these regions are expected to have the largest volumetric change due to sedimentation at the river deltas. In context of extensive sedimentation observed in the Colorado and San Juan River deltas and inflows (Lazenby and Nelson, 1976; Condit and others, 1978; Ferrari, 1988; Potter and Drake, 1989; Hart and others, 2005; Ferrari, 2006; Pratson and others, 2008; Hornewer, 2014), further modifications were made to the original TBDEM (Poppenga and others, 2020) prior to calculating storage capacity relationships (table 1; Jones and Root, 2021). Specifically, these modifications (1) address gaps in the dataset where the historical DEM (Root and others, 2019) was incorporated and where sediment has since been deposited and (2) replace the constant, hydro-flattened elevation in the river channel upstream from the Colorado and San Juan deltas with alternate topographic data sources.

Regions of the topobathymetric data that incorporated the historical DEM were modified with a reasonable expectation of sedimentation in Lake Powell. In many places in the reservoir, particularly where steep cliffs are predominant, the historical DEM was an acceptable substitution for data gaps. In other regions, with shallower topography or near the river deltas, the historical DEM does not account for large volumes of accumulated sediment. Steep, sharp drop-offs at the interface between bathymetry and lidar data are most pronounced at the edges of the thalweg where the reservoir narrows into the river channel, creating artifact trenches nearly 165 ft deep where they are, in fact, filled with reservoir sediment. To create a surface that best represents the topography in these areas, these artifacts were individually identified in the delta regions and filled using the Fill tool in ArcMap (version 10.6.1; fig. 4).

Two substitute elevation datasets were incorporated into the modified TBDEM to replace the zones in the original TBDEM where the river thalwegs are represented with a hydro-flattened elevation value of 3,612.76 ft above NAVD 88 (fig. 4). The first dataset is a 0.5-m DEM derived from lidar data acquired in October and November 2015 by the Utah Division of Forestry, Fire And State Lands and the National Park Service (Utah Geospatial Resource Center, 2017). The dataset covers 152 mi² of the Colorado, Green, and Yampa river channels in Utah and Colorado, with a sampling density of eight points per square meter. The vertical accuracy of the 2015 DEM was assessed with 72 total check points, including 40 non-vegetated and 32 vegetated. For non-vegetated check points, the 95-percent confidence level was 0.4406 ft (0.1343 m), with a RMSEz of 0.2247 ft (0.0685 m); for vegetated check points, the 95-percent confidence level was 0.1367 m (0.4485 ft), with a RMSEz of 0.0697 m (0.2287 ft). The second dataset is a 5-m DEM created from lidar that was collected in July 2006 (Utah Geospatial Resource Center, 2007) when the lake was at a comparable elevation to the TBDEM datasets (Poppenga and others, 2020). Vertical accuracy was tested on 190 check points, which resulted in 2 check points being removed due to unacceptable accuracy. The 95-percent confidence level is 12.64 ft (3.854 m), with a RMSEz of 6.453 ft (1.967 m). The course of the San Juan River, over subaerially exposed reservoir sediment, is similar to present day based on satellite imagery. The vertical difference of the water surface elevation of the San Juan River as it incises the reservoir sediment (since 2006) is unknown because other elevation datasets that could improve this estimate were not available for the region.

Using the TBDEM spatial metadata as a mask to exclude areas where modifications were not necessary (Poppenga and others, 2020), the substitute elevation datasets were mosaiced over the hydro-flattened regions of the unmodified TBDEM using the ‘Mosaic to New Raster’ tool in ArcMap (version 10.6.1). Source inputs were not reassembled due to processing capacity limitations. A 10-m buffer accounted for the difference in cell size between the input datasets. Because the reservoir had not flooded this region between acquisitions of the 2015 and 2018 lidar (fig. 2), the full extent of the 2015 dataset, including areas outside of the hydro-flattened zone, was used to avoid potential conflicts with the 2018 TBDEM related to changes in topography over time. The resulting raster was matched by cell coordinate location to the original TBDEM, and the substitute datasets were resampled using a cubic convolution to conform with the 1-m cell size. The 2006 5-m DEM primarily is used on the San Juan River arm and had several anomalous sinks that were adjacent to vertical cliffs; these sinks were individually filled.

Elevation-area-capacity relationships were derived from a script written in the Python coding language (Python Software Foundation, Python Language Reference, version 2.7, available at http://www.python.org) that utilizes Esri’s Storage Capacity tool, part of the Spatial Analyst Supplemental Toolbox (version 1.4, https://www.arcgis.com/home/item.html?id=3528bd72847c439f88190a137a1d0e67). The script requires three inputs: (1) a DEM, (2) a polygon boundary of the spatial extent of the computation, and (3) the elevation range and interval to calculate the elevation-area-capacity values within. Starting at the minimum elevation value provided, the script iteratively tabulates the total area and volume within the provided polygon extent below each elevation increment. Due to the computational requirements of the analysis, the script was adapted for use on the Yeti high performance computing cluster at the USGS Advanced Research Computing Center in Denver, Colorado (https://www.usgs.gov/core-science-systems/sas/arc). The inputs used to compute elevation-area-capacity relationships for Lake Powell included the modified TBDEM (available for download through ScienceBase at https://doi.org/10.5066/P9H60YCF; Jones and Root, 2021) and a reservoir mask that excludes areas below the Glen Canyon Dam.

Units of the original TBDEM (Poppenga and others, 2020) are in the International System of Units (SI). Primary storage capacity calculations were made in SI, with an elevation range of 950.98–1,133.03 m above NAVD 88 (3,120.01–3,717.29 ft above NAVD 88) at increments of 0.10 m (0.33 ft). These results were then converted to the U.S. customary system (USCS), in accordance with Reclamation operational standards at Glen Canyon Dam and historical estimates of storage capacity. All subsequent calculations were done with USCS results.

Resulting 0.33-ft (0.10-m) relationships were used to generate elevation-area-capacity values at 0.01-ft increments by linearly interpolating values between each 0.33-ft (0.10-m) incremental pair. The 0.01-ft interpolated interval was calculated at the request of Reclamation.

Additional storage capacity calculations were done for error assessment at matching 0.01-ft increments for lower (3,160.00–3,161.00 ft above NAVD 88), middle (3,400.00–3,401.00 ft above NAVD 88), and upper (3,700.00–3,711.00 ft above NAVD 88) elevations. Values were generated following the same procedures outlined for the original 0.33-ft (0.10-m) increment calculations, thereby providing a one-to-one comparison dataset against the linearly interpolated values for these discrete elevation bands.

The TBDEM is referenced to the NAVD 88 using the Geoid12B geoid and horizontally referenced to the North American Datum of 1983 with the National Adjustment of 2011, UTM Zone 12 projection (Poppenga and others, 2020). Historical data (Root and others, 2019) were referenced to NGVD 29 and have been transformed to NAVD 88. Transformation from NGVD 29 to NAVD 88 is determined by adding a constant conversion value to elevations in NGVD 29 for the area of interest. Correction values are stored in a georeferenced grid (0.05 decimal-degree cell size) developed for the NOAA VDatum (v4.0) vertical datum transformation tool (National Oceanic and Atmospheric Administration, 2019). The conversion value for elevations at the dam used in this study is +2.913 ft from NGVD 29 to NAVD 88 and was selected from the land surface east of Glen Canyon Dam using the NOAA VERTCON tool (National Oceanic and Atmospheric Administration, 2021). This conversion was calculated for the USGS reservoir elevation station (09379901) located at 36.9369326, −111.4837694.

Results from the 2017–18 survey indicate that the total storage capacity and areal extent at full pool (3,702.91 ft above NAVD 88) are 25,160,000 acre-feet and 159,200 acres, respectively. These values represent a decrease in storage capacity of 1,833,000 acre-feet or 6.79 percent from 1963 to 2018 and a 1,048,000 acre-ft or 4.00 percent decrease from 1986 to 2018. The decrease in areal coverage is 2,142 acres or 1.33 percent from 1963 to 2018 and 1,536 acres or 0.96 percent from 1986 to 2018. The elevation-area-capacity relationships are summarized in figures 5 and 6, and table 2. Calculated elevation-area-capacity relationships at 0.33-ft (0.10-m) calculated increments and 0.01-ft interpolated increments are available for download at https://doi.org/10.5066/P9O3IPG3 (Jones and Root, 2022).

Error analysis of the interpolated relationships was performed at 0.01-ft intervals across three elevation bands: (1) lower (3,160.00–3,161.00 ft above NAVD 88); (2) middle (3,400.00–3,401.00 ft above NAVD 88); and (3) upper (3,700.00–3,711.00 ft above NAVD 88). The interpolated values are comparable to the calculated values, with differences ranging from −0.01 to 0.03 percent and 0 to 0.02 percent for area and storage capacity calculations, respectively. The error analysis for all interpolated values is available for download at https://doi.org/10.5066/P9O3IPG3 (Jones and Root, 2022).

Intermediate reservoir elevations include critical benchmarks for the management of Lake Powell and operations at Glen Canyon Dam. The dead storage elevation (3,372.91 ft above NAVD 88), which is the reservoir level below all hydroelectric penstocks and other outlet works, has decreased in storage capacity by 152,900 acre-feet or 8.11 percent from 1963 to 2018 and 27,100 acre-feet or 1.54 percent from 1986 to 2018. The inactive storage level (up to 3,492.91 ft above NAVD 88) represents the lowest reservoir elevation that all hydroelectric penstocks are submerged. Storage capacity at the inactive storage level has decreased by approximately 599,900 acre-feet or 9.84 percent from 1963 to 2018 and 333,000 acre-feet or 5.72 percent from 1986 to 2018.

These results reflect the elevation-area-capacity relationships determined in this study and are not directly comparable to values used for operations at Glen Canyon Dam. Direct comparisons to Reclamation operational relationships are outside the scope of this report.

Sources of error in the TBDEM and elevation-area-capacity calculations are difficult to identify and quantify. Random errors in topographic and bathymetric surveying are small and are ignored here. Systematic errors are difficult to identify, though each dataset was independently assessed for error prior to being assembled into the TBDEM. Interpolation errors are quantifiable, but this was not calculated because of the number of datasets and range of acquisition dates for each survey.

Because the TBDEM comprises sources with different resolutions, coverages, and known errors, the vertical error can be reasonably quantified. The data sources that most represent the TBDEM by area are the 2017 bathymetry (Andrews and others, 2018) and 2018 lidar. The bathymetry includes a majority of storage capacity in Lake Powell below the minimum reservoir elevation during the survey (3629.18 ft above NAVD 88). This elevation corresponds to 61.05 percent storage capacity at full pool. The remaining volume is primarily defined by the 2018 lidar, with a minority contribution from the historical DEM and interpolated topography. The vertical error of the bathymetry is 1 percent of water depth or up to 5 ft of error at the maximum water depth (approximately 500 ft). The average water depth of the bathymetric survey, using the highest reservoir elevation to difference the bathymetry, is 142 ft. This average depth equates to a 1.42 ft error but is likely to vary with location in the reservoir. Vertical error of the lidar at 95-percent confidence level is less than 0.5 ft. The greatest vertical error all associated datasets is in the 2006 statewide auto-correlated DEM (Utah Geospatial Resource Center, 2007), with a 95-percent confidence level of 12.64 ft and RMSEz of 6.453 ft. These errors are substantial but only representative of approximately 3.44 percent of the TBDEM area and only in select regions where hydro-flattened topography was used (see spatial metadata for Jones and Root, 2021). Thus, the bathymetric dataset is most likely to provide a reasonable error for elevation-area-capacity relationships. Conservatively, a vertical error in reservoir elevation of ±1.42 ft is suggested for the results provided here. At full pool (3,702.91 ft above NAVD 88), this error approximately corresponds with storage capacity between 24,940,671 and 25,392,938 acre-feet or a range of 452,267 acre-feet.