High-resolution bathymetries and shorelines for the Great Lakes of the White Nile basin

The survey of Lake Albert was conducted in February 2020. The Lake Edward and Lake George surveys were conducted in August of 2020. The surveys of Lake Victoria were conducted between September and November of 2017, 2018, 2019, and 2020. We assume no significant morphological change occurred in Lake Victoria across these 4-years. All collection periods correspond to the end of a traditional dry season and the transition period into the beginning of a traditional wet season. Water levels were monitored during the period of each Lakes’ survey. Benchmarks were installed during each Lakes survey aside from Lake Victoria, where an existing benchmark nail existed. Unmanned aerial systems (UAS) were flown during the Lake Albert survey to assess our shoreline delineation methodology.

Lake elevation levels

Lake Victoria utilizes spaceborne altimetry to ascertain its lake elevation. Lakes Albert, Edward, and George have no systematic high accuracy spaceborne altimeter measure of lake elevations. Therefore, Lake Albert, Edward, and George’s lake elevations are derived from statistical analyses of observed water levels.

Lake sounding datums

For Lake Albert, Lake Edward, and Lake George, visual water-level (WL) observations taken throughout the survey are averaged to obtain the lake elevation (LE), also known as the project sounding datum (SDp). The method for determining SDp is to observe the WL on a graduated board, often called a tide board or a staff gauge (G), securely attached to a piling or other solid vertical structure extending below the lake surface. The graduations are then marked relative to the gauge zero (G₀). The WL is read as the distance above or below G₀ where the water surface intersects the gauge.

A fixed, tamper-resistant benchmark (Bm) was installed or in operation at each Lake within the optical leveling distance of each gauge to achieve the conversion from local water levels to ellipsoidal heights and EGM 2008 elevations. First, each Bm’s horizontal and vertical position was measured using the

Global Navigation Satellite System (GNSS). Then, the vertical distance between the benchmark elevation (BmE) and G₀ is measured using standard optical or laser-based survey methods. This distance is the vertical gauge offset (VGO).

The Lakes’ elevation methodology is summarized in Fig. 2 and is defined in Eq. 1. At this point, SDp for Lakes Albert, Edward, and George is merely an ellipsoidal height; the ellipsoidal height is converted to EGM:2008 using Harmonic Synthesis at the horizontal coordinate location of each Bm^15,16.

Eq. 1 – Lake Albert, Lake Edward, and Lake George Sounding Datums

SDp is the lake elevation or project sounding datum, Bm is the benchmark elevation from RTK GPS, VGO is the vertical gauge offset derived by using an optical level, and WL is the water level obtained from the gauge reading.

Unlike Lake Albert, Lake Edward, and Lake George, due to Lake Victoria’s size, with a maximum diameter exceeding 375 km, hydrodynamic effects could readily negate the hydrostatic assumption that the lake surface is uniformly level. On Lake Victoria, wind setup, seiching, and the significant outflow into the Victoria Nile would result in hydraulic gradients that would make any single, nearshore water level gauge unrepresentative of lake levels at points distant from the gauge. To establish a meaningful SDp for Lake Victoria using nearshore water level gauges, at least three stations distributed equilaterally around the Lake’s perimeter would need to be established and operated simultaneously for long periods. However, this approach was deemed unfeasible primarily due to cost and logistical constraints. For example, creating a multi-country concurrent network of gauges would require at least three times the equipment, three times the labor, and three times the training.

The alternate approach utilizes Jason-3 spaceborne altimeter data. This method has been used in Lake Victoria and is supported by the USDA G-REALM program¹⁷. Jason-3 is a radar altimeter launched in January 2017. The primary goal of Jason-3 is to provide sea-level variations with accuracies under 2.5 cm at a repetition cycle of 10-days¹⁸. As Jason-3 passes over Lake Victoria, it can establish EGM 2008 elevations for the Lake from numerous measures towards the middle of the Lake. Jason-3 passes over 150 km of Lake Victoria. The collection path runs from approximately Nyabansari in Tanzania to Bugaia in Uganda. As the instrument is radar-based, climatic conditions rarely limit the data collection. The raw altimeter data collected by Jason-3 undergoes numerous corrections before a lake surface elevation is determined, including a dry tropospheric correction, a wet tropospheric correction, an ionosphere correction, and an instrument-specific bias adjustment¹⁹. Lake elevation observations were obtained from Jason-3 during Lake Victoria’s surveys across 2017, 2018, 2019, and 2020. The average of the Jason-3 readings from 2020, which itself is an average of many hundreds of observations, defines the SDp for the Lake Victoria surveys.

A Lake Victoria benchmark is still surveyed at a water level gauge to allow for past and future data integration, and the benchmarks are tied to the altimeter measures used. At this point, SDp for Lake Victoria is already in EGM:2008 as Jason-3 uses EGM:2008 as opposed to ellipsoidal elevations, so harmonic synthesis is not required as it is for the other Lakes.

Lake benchmarks

Benchmarks for Lake Albert (BmA), Lake Edward (BmEd), and Lake George (BmG) were installed along each of the three Lakes’ shorelines. Each benchmark is situated within a few meters and line-of-sight of a water level staff gauge. A preexisting benchmark nail (BmV) located above the gauge was utilized for Lake Victoria. Aside from Lake Victoria, each installed benchmark is an 8 cm diameter brass disc stamped with LEAF II. Each installed benchmark was anchored approximately 15 cm into a larger concrete pad using a twisted steel reinforcement bar. Each benchmark’s location was obtained using long-term GNSS averaging, captured by a Hemisphere GNSS receiver with Atlas satellite-based augmentation system wide-area corrections applied. Observations without a corrective signal were discarded. Ellipsoidal elevation, recorded to the millimeter level, was also captured by the GPS receiver. Conversion of benchmark ellipsoid elevations to EGM 2008 WGS 1984 Version used the harmonic synthesis coefficients provided by the U.S. National Geospatial-Intelligence Agency (NGA) EGM Development Team^15,16.

BmA was installed on 1/31/2020 within the UPDF Marine compound at Mbegu, approximately 6.5 km east-northeast of Kaiso, Uganda, on the eastern side of Lake Albert. Across seven days between 2/1/2020 and 2/20/2020, the horizontal location of the benchmark was recorded by a GNSS receiver with built-in averaging. The GPS unit averaged horizontal locations at the benchmark until it reached 95 percent confidence. In addition, the ellipsoidal height was collected on the surface of Lake Albert across the survey period and adjusted to the benchmark elevation using the vertical gauge offset and the water level readings. The total number of vertical observations is 35,550.

BmEd was installed on 2/13/2020 at the fish landing site in Katwe Village, Uganda, at the northern end of Lake Edward. Across portions of 8/5/2020, 8/10/2020, 8/13/2020, and 8/15/2020, one X, Y, and Z GPS location were recorded every 5-seconds, totaling 11,242 observations.

BmG was installed on 8/11/2020 at the landing site in Kahendero, Uganda, on the western side of Lake George. On 8/13/2020, one X, Y, and Z GPS location were recorded every 5-seconds, totaling 2,663 observations. Unfortunately, BmG does not have a full unobstructed 360° view of the sky and may require further refinement.

A preexisting benchmark nail (BmV) at the railroad dock in Jinja, Uganda, is used for Lake Victoria. The nail is located directly above the water level gauge and marked with a white paint X. Across portions of 3/22/2021 and 3/23/2021, one X, Y, and Z GPS location was recorded every 5-seconds, totaling 6,842 observations. Still, as noted earlier, altimetry data was used for the actual SDp.

Lake gauges

Within a few meters of each benchmark, a water level staff gauge was either installed or already existed. For Lake Victoria (GV), Lake Albert (GA), and Lake Edward (GE), preexisting gauges were used. At Lake George (GG), a temporary gauge was established for the duration of field operations.

GA is a staff gauge of unknown origin. The staff is a simple iron square tube painted decimeter intervals subdivided into 5 cm steps. The 100 cm subdivision at the top of the gauge was surveyed relative to the BmA (Fig. 2, Y_BG) using an optical level on 1/31/2020. Between 2/1/2020 and 2/20/2020, twelve lake level observations were collected. The water level only varied by 6 cm across the entire survey. The average of the 12-daily readings was used to help define the SDp for the Lake Albert bathymetric survey.

GE is a long-term gauge installed by the Ugandan Ministry of Water. The gauge is a stepped gauge consisting of three separate concrete pillars of increasing height with graduated measurement strips attached at the centimeter level. The water level on the gauge, relative to the BmEd, was surveyed using an optical level on 8/10/2020. Twice-daily Lake level observations continued throughout the 11-day survey operation between 8/5/2020 to 8/22/2020. The water level only varied by 3 cm across the entire survey. The average of the 11-daily readings was used to help define the SDp for the Lake Edward bathymetric survey.

GG is a temporary gauge installed for the duration of field operations. The gauge is a simple wooden gauge with painted centimeter intervals anchored to a galvanized steel pipe driven between 1 m and 2 m into the substrate. The water level on the gauge, relative to the BmG, was surveyed using an optical level on 8/12/2020. Once-daily Lake level observations were collected across the two days of the hydrographic survey and the day before and after the survey. The water was stable across the entire survey. The two average daily readings were used to define the SDp for the Lake George bathymetric survey.

GV is a long-term gauge installed by the Ugandan Ministry of Water. The gauge has graduated measurement markers at the two-centimeter level. The zero level on the gauge, relative to the BmV, was surveyed on 3/22/2021 and 3/23/2021. As BmV and GV are at the same horizontal coordinates, leveling is not required. Water level observations were not utilized from this gauge during the survey, as the Jason-3 altimeter was used to establish the Lake elevation level for Lake Victoria. Instead, the closest four Jason-3 measures across the survey dates are used to calculate the water level. The water level varied by 4 cm across the 2017 bathymetric survey, 9 cm across the 2018 survey, 5 cm across the 2019 survey, and 13 cm across the 2020 survey. The 2020 water level is used as the SDp to allow for as close as possible temporal consistency across all Lakes in the database.

Lake elevation data

Table 1 provides each lake’s SDp in the most common gravitational models and all input parameters to the lake elevation models. The SDp for Lake Edward is 915.77 m (EGM08), the E/SDp for Lake George is 915.74 m (EGM08), and the SDp for Lake Albert is 622.18 m (EGM08), and the SDp for Lake Victoria is 1136.92 m (EGM08). Measures of uncertainty are provided in the technical validation.

Lake bathymetries

The Lake Albert hydroacoustic survey was conducted across 14-days between February 1^st, 2020, and February 20^th, 2020. The Lake Edward hydroacoustic survey was conducted across 10-days between August 4^th, 2020 and August 22^nd, 2020. On August 13^th, 2020 and August 14^th, 2020, the Lake George hydroacoustic survey occurred during a Lake Edward Survey break. The Lake Victoria hydroacoustic survey occurred daily between September 8^th, 2017 and October 7^th, 2017, September 10^th, 2018 and October 9^th, 2018, September 15^th, 2019 and October 13^th, 2019, and finally between October 20^th, 2020 and November 25^th, 2020. The Lake Victoria soundings from 2017, 2018, and 2019 were vertically corrected to align to the 2020 water levels. The earlier year were adjusted by 1.28 m (0.03 m, 95CI), 0.975 m (0.06, 95CI), and 1.025 m (0.05 m, 95 CI), respectively.

The hydroacoustic survey transect designs were based on local topography, available bathymetry, and cost considerations. Both Lake Albert and Lake Edward had dominant relief patterns running from the Congolese highlands in the west to the Ugandan Plateau in the east, forming a deep U shape perpendicular to the Albertine Rift. The survey transects were designed to follow this axis of high relief across the Albertine Rift. Lake George and Lake Victoria have no discernable relief patterns, both being relatively shallow bowls situated across flat planes. Therefore, the survey designs were optimized to capture an adequate portion of these two Lakes while minimizing cost.

Lake soundings

Across Lake Albert, Lake Edward, and Lake George, a 9 m, V-bottomed, shallow draft research vessel was deployed with a Ugandan crew out of Jinja, Uganda. The echosounder used to collect the soundings was a dual-frequency sounder with a built-in data logger, external GNSS receiver, and a combined low-frequency (33 kHz) high-frequency (200 kHz) transducer. Both frequencies were operational and recorded during the survey, but only the high-frequency signal was processed to produce Lake Albert and Lake George’s soundings. Greater than 90 percent of Lake Edward also used the high-frequency sounder, but the instrument was switched to low-frequency in areas over 90 m deep. A speed of sound adjustment was made based on the water sampling that occurred on average twice each transect. Calibration was performed before the initial deployment.

For Lake Albert, Lake Edward, and Lake George, Hydromagic 9.1 software was used to record and process the acoustic soundings into tabular X, Y, and Z formats. The echosounder’s echogram was output in real-time to a laptop. A dedicated 12-volt battery, maintained by a 60-watt solar panel mounted on the cabin top, powered all equipment. Positions were obtained by a multi-frequency GNSS antenna connected to the echosounder. The transducer was mounted on an aluminum extension pole that supported the GNSS antenna directly above the transducer. The antenna received Atlas L-band satellite-based augmentation system (SBAS) correction signals that allow precise positioning.

Lake Victoria soundings were collected by the stern trawler RV Lake Victoria Explorer by members of the Hydroacoustics Regional Working Group of the Lake Victoria Fisheries Organization. This group is based out of Jinja, Uganda, Kisumu in Kenya, and Mwanza in Tanzania. This group has conducted twenty-three acoustic surveys of Lake Victoria since 1999 under an established protocol²⁰. The RV Explorer is a 17 m research vessel and a V-shaped hull with a draft of 1.8 m. The echosounder used on the RV Explorer is a dual-frequency system operating at 70 kHz and 120 kHz, respectively. The transducers are mounted on a protruding instrument keel under the boat and powered by the vessel’s electrical system. Calibration was performed immediately before each daily survey. The GPS logger used on this system is not differentially corrected.

For Lake Victoria, Echoview 8.0 software was used to record and process the soundings into tabular X, Y, and Z formats. After noise was removed from the raw signal and adjustments were made to correct the beam angle, the initial lakebed soundings were obtained using the best bottom candidate algorithm²¹. A CTD probe was used at each calibration site to determine the local environmental conditions. The average water temperature at the calibration site was input into the system to predict the sound speed. Lake Victoria’s survey’s calibration protocol is detailed in the Standard Operating Procedures for Hydroacoustics surveys on Lake Victoria²⁰.

Across all Lakes, either a certified coastal engineer or an individual with relevant expertise processed the echograms from the echosounder. The process essentially involves detecting the average bottom in the echogram and digitizing through small peaks and pits caused by the boat’s motion. A narrow interpretation is needed on calm days, and the automated extraction of the lake bottoms often suffices. On days with rough water, manual digitization of the trace is required. Sometimes, the signal may reflect off anything in its path to the bottom, including suspended sediment, debris, animals, subaquatic vegetation, silt, mud, or a harder compacted layer beneath a softer surface layer. The digitization process removes such anomalies as well as smoothing over dropouts and other noise. Finally, the digitized trace is exported to tabular soundings for use in GIS and other software. Figure 3 represents the soundings across all Lakes.

Lake bathymetries data

For Lake Albert, Lake Edward, and Lake George, the output spatial and tabular data contains; the date of the sounding, the horizontal position of the sounding, and corrected depth using a local-verified speed of sound adjustment for both high-frequency and low-frequency soundings when applicable, the vessel speed at the time of the sounding, the vessel heading at the time of the sounding, and a field indicating if the GNSS was operating in uncorrected or corrected mode for each sounding. For Lake Victoria, the output spatial and tabular data contains the date of the sounding, the time of the sounding, the horizontal position of the sounding, corrected depth using a local-verified speed of sound adjustment, and a field indicating if the GNSS was operating in uncorrected or corrected mode for each sounding. Depth zero corresponds to the LE /SDp for each Lake as already defined.

Across Lake Albert, 290,018 soundings were collected (Table 2), resulting in 53 soundings per square kilometer. Across Lake Edward, 225,528 soundings were collected (Table 2), resulting in 101 soundings per square kilometer. Across Lake George, 59,281 soundings were collected (Table 2), resulting in a density of 211 soundings per square kilometer. Finally, across Lake Victoria, 17,958,859 soundings were collected (Table 2), resulting in a density of 269 soundings per square kilometer. The water volume and mean depth are calculated using constrained Delaney Triangulation, whereas the maximum depth is the deepest collect sounding. The summary information for each Lakes’ bathymetry is shown in Table 2 and is compared against values from the (WLD) World Lakes Database²² unless otherwise noted.

Lake shorelines

For each of the Lakes, we constructed high-resolution shorelines from spaceborne imagery at a combination of 15 m, 10 m, 5 m, 3 m, 50 cm, and 30 cm. Accuracy statistics were generated using UAS-derived imagery at 10 cm.

Sentinel-2 imagery

Sentinel-2 is designed to map and monitor water cover, inland waterways, and coastal areas²⁴. The baseline spaceborne imagery used to delineate the shorelines across Lake Albert, Lake Edward, and Lake George is Sentinel-2. Sentinel-2 is a European Space Agency (ESA) wide-swath, high-resolution (HR), a multi-spectral imaging system that consists of two satellites flying in the same orbit but phased at 180°²³. The system carries an optical instrument payload that samples thirteen spectral bands: four bands at 10 m resolution, six bands at 20 m resolution, and three bands at 60 m resolution²⁵. The four bands at 10 m resolution are centered on the wavelengths 0.490 µm, 0.56 µm, 0.665 µm, and 0.842 µm, respectively. These wavelengths correspond to the blue, green, red, and near-infrared portions of the electromagnetic spectrum. These spectral properties of Sentinel-2 allow for color composites and false color composites of each of the Lakes at 10 m resolution. Furthermore, as the radiometric signal in the near-infrared band is almost entirely absorbed by open water, it can assist in delineating a water-terrestrial edge boundary.

The Sentinel-2 data granules used to delineate the Lake Albert shoreline are:

• S2B_MSIL1C_20190403T080609_N0207_R078_T36NUH_20190403T110906, S2B_MSIL1C_20190503T080619_N0207_R078_T36NTG_20190503T112849, S2B_MSIL1C_20190503T080619_N0207_R078_T36NTH_20190503T112849, S2B_MSIL1C_20190503T080619_N0207_R078_T36NUG_20190503T112849

  The Sentinel-2 data granules used to delineate the Lake Edward shoreline are:

• MSIL1C_20170702T081009_N0205_R078_T35MRV_20170702T082404, MSIL1C_20170821T080959_N0205_R078_T35MQV_20170821T082855

  The Sentinel-2 data granule used to delineate the Lake George shoreline is:

• S2B_MSIL1C_20191229T081239_N0208_R078_T35NRA_20191229T100818

Landsat imagery

The baseline spaceborne imagery used to delineate the Lake Victoria shoreline is Landsat-8. Landsat-8 is a USGS/NASA, high-resolution (HR), multi-spectral imaging system. Landsat-8 uses a push-broom Operational Land Imager and Thermal Infrared Sensor to collect data with a spatial resolution of 30 meters in the visible and near-infrared regions of the electromagnetic spectrum. The relevant bands at 30 m resolution are the blue band located between 0.45 µm to 0.51 µm, the green band located between 0.53 µm to 0.58 µm, the red band located between 0.64 µm to 0.67 µm, and the near-infrared band located between 0.85 µm to 0.88 µm. As the infrared band is almost entirely absorbed by open water, it can assist in delineating a water-terrestrial edge boundary. In addition, a 15 m panchromatic band is located between 0.64 µm to 0.67 µm and is used to pansharpen the 30 m bands to allow for feature digitizing at 15 m resolution. These spectral properties of Landsat-8 allow for color composites and color-infrared composites of Lake Victoria at 15 m resolution when pan-sharpened.

The Landsat data are listed below.

LC81700602020049LGN00, LC81700602021003LGN00, LC81700612020001LGN00, LC81700612020049LGN00, LC81700622020017LGN00, LC81700622020049LGN00. LC81710602020040LGN00, LC81710602021026LGN00, LC81710612020040LGN00, LC81710622020040LGN00, LC81720602020047LGN00

Very high-resolution planetscope eye imagery

In highly dynamic vegetative areas where Sentinel-2 or Landsat-8 cannot delineate a clear shoreline, very high resolution (VHR) imagery was obtained and used (Table 3). For example, the southern wetland of Lake Albert across both the DRC and Uganda uses 50 cm Worldview 2 (WV2) and 30 cm Worldview 3 (WV3) imagery as opposed to Sentinel-2 (Table 3), as this region has ephemeral floating grasses, sub-aquatic vegetation, and therefore shows a reflected signal response in the near-infrared bands of the satellite imagery. Thus, the wetland areas of Lake Albert are of substantially higher resolution than the rest of the Lake Albert shoreline.

Sub-meter resolution UAS

Finally, sub-meter resolution (SMR) UAS was flown over Lake Albert to ascertain the shorelines’ positional accuracies. Once the accuracy statistics were calculated, the UAS data was incorporated back into the shorelines for these areas. These UAS-derived shorelines are the regions around Kaiso, Butiaba, and Ntoroko on Lake Albert in Uganda.

Shoreline digitization

The initial step of the shoreline delineation was selecting the required satellite scenes—the selected scenes needed to meet the following criteria, be mostly cloud-free over the Lakes, and have suitable flags indicating high-quality data. The ESA Copernicus Hub and USGS GLOVIS sites were searched until the images met the above criteria. The selected granules were then subset only the Blue, Green, Red, and near-infrared bands, and the Landsat-8 imagery was pan-sharpened. Once composited, each 4-band raster is represented as a color-IR composite and a visible color composite. Before digitizing began, the resolution was set to 1:20,000 for all Lakes aside from Lake Victoria, which was set to 1:30,000.

Fishnets were constructed that covered the entirety of each Lake. The shoreline in each cell of the fishnet is manually digitized in a heads-up manner. The first pass of each cell digitizes the exterior shoreline of the Lake. The second pass of each cell digitizes all islands in the cell, and the third pass digitizes potential nearshore obstructions. Once each cell is complete, a second cartographer verifies the digitization and sends all questions back to the original digitizer, making the required updates. The final stage is to combine all the individual shoreline cells of the fishnet into a singular whole for each Lake and then verify the constructed shoreline feature’s topology.

Resolution and scale

Using Tobler’s rule of scale and resolution²⁶, it is possible to create a shoreline that approximates 1:20,000 scale from the 10 m Sentinel-2 images and 1:30,000 from the Landsat-8 imagery using appropriate error monitoring and control. The Planet Scope imagery at 3 m resolution would equate to 1:6,000, the WV2 imagery at 50 cm resolution would equate to 1:1000, the WV3 imagery at 30 cm resolution would equate to 1:600, the UAS imagery at 10 cm resolution would equate to 1:200. For these reasons, the Lakes Albert, Edward, and George shorelines can be considered at a minimum 10 m resolution or a 1:20,000 scale product. The Lake Victoria shoreline can be regarded as a minimum 15 m resolution or a 1:30,000 scale product. We report the coarsest resolution as the shoreline’s resolution from the coarsest instrument, but large portions of the shorelines are higher resolution from less coarse instruments.

Lake shorelines data

We find the surface area of Lake Edward, Lake Albert, Lake George, and Lake Victoria to be 2,241,119,039 m², 5,423,949,967 m², 281,121,696 m², and 66,792,882,259 m², respectively. We find the shoreline lengths of Lake Edward, Lake Albert, Lake George, and Lake Victoria to be 241,395 m, 484,454 m, 89,204 m, and 3,063,755 m, respectively. The summary information for each Lakes’ shoreline is shown in Table 4, and the data are compared to the Global Self-Consistent, Hierarchical, High-Resolution Geography Database (GSHHG)²⁷, considered the current best available consistent across these Lakes²⁷.

Hardware and Software

Soundings were collected and processed using Eye4Software Hydromagic or Echoview Software Pty Ltd, Echoview software. The sounding collection system used for Lake Albert, Lake Edward, and Lake George was the CEESystems CEESCOPE. High-frequency soundings for Lake Albert, Lake Edward, and Lake George were collected using a 200 Khz transducer from CEE Systems. The low-frequency soundings for the deep-water portion of Lake Edward were collected using a 33 kHz transducer from CEE Systems. The sounding collection system used on Lake Victoria before 2020 was a Simrad EK 60 dual frequency echo sounder with a 7° beam angle connected to 70 kHz and 120 kHz general-purpose dual transducer produced by Kongsberg Maritime AS. For 2020, the sounding collection system was changed to a Simrad EK80 dual frequency echo sounder, which operated at the same frequencies. The GNSS system used on Lake Albert, Lake Edward, and Lake George was a Novatel Hemisphere GPS. The Hemisphere Atlas system provided the SBAS L-Band GPS real-time correction. The Hemisphere Atlas system provided the SBAS L-Band GPS real-time correction. GNSS system used on Lake Victoria was a Globalsat Technology Corporation GPS.

ESRI ArcGIS ArcPro²⁸, GDAL/OGR²⁹, and QGIS³⁰ were used to perform all horizontal coordinate transfers, conduct geostatistical analysis, produce cartographic outputs, digitize shorelines, post-process the soundings, and analyze the soundings. Microsoft Excel was used to process and transform the SDp GPS data. Harmonic synthesis transformation for data conversion to EGM 2008 was conducted in the Harmonic Synth WGS 84 Fortran code provided by the NGA¹⁵.

Sentinel-2 and PlanetScope were the primary data sources for the satellite imagery The SenseFly EBee + UAS³¹, with the SODA survey camera³², was used to fly the data and then assess the accuracy of the shoreline delineation. SenseFly Emotion³³ software was used to plan and fly all UAV missions. Pix4D³⁴ was used to process all UAV imagery.

Tinfour 2.7.1³⁵ to triangulate mass bathymetric soundings and calculate each Lakes’ mean depths and volume.

Source: Resources - nature.com