Satellite-altimetry data have previously been extensively used by Getech to generate gravity data sets for areas of open sea and oceans, but a similar approach can be used to generate gravity data over inland lakes. We have used a technically innovative method to create new gravity data sets for twenty lakes around the globe, ranging in size from 82,000 km2 (Lake Superior – US/Canada) to 2,300 km2 (Lake Edward – DR Congo/Uganda). Included in this, is a new gravity data set for Lake Tanganyika in the Albertine Rift, Africa.
Gravity data have been derived from satellite altimetry since Haxby (1985) used data from SeaSat to create a global offshore gravity map. Since then new satellites and processing advances have dramatically improved the data coverage and have also led to new higher-resolution data sets (Green et al., 2019). There are two requirements for appropriate satellite missions to be used for this purpose: 1) they require an altimeter on board to measure the sea-surface height and 2) they must have close track spacing (less than ~10 km). The latter of these two requirements implies that the satellite is in a ‘geodetic’ orbit. Many satellites follow an ‘exact repeat’ orbit but these are rarely of use for this study, as their short repeat period results in large track spacing.
The surface of any liquid that is free to flow tends to form an equipotential surface of the gravity (Figure 1). The gravity at the lake surface is the vertical derivative of the gravity potential, which is itself proportional to the surface height. Thus, the gravity can be readily calculated from the surface height variation over an area, e.g. using Fourier transforms. The processing flow is largely similar to that applied by Getech to calculate gravity for the world’s continental margins and oceans. A geoid-to-gravity approach is used to convert the raw altimetry data to free-air gravity, including re-tracking the radar waveform data, isolating the residual water surface height, removing noise, and applying statistical levelling techniques: cross-over levelling and micro-levelling to reduce track-orientated noise. Finally, the residual geoid is converted to gravity and the long wavelength representation of the gravity is added back in to the model to produce free-air gravity. Bouguer and terrain corrections can then be applied to the grid.
Figure 1 – The water level at the lake surface is an equipotential surface and will rise in the vicinity of dense sub-surface bodies. This allows the water height to be converted into gravity data at the lake surface.
The raw data for Lake Tanganyika were obtained from radar altimetry data from the CryoSat-2, Jason-1 and Geosat satellites. Geosat, operated by the United States Navy, was the first geodetic satellite to provide global coverage of altimetry data. As the Geosat mission was primarily designed to measure the sea-surface height rather than other oceanographic effects, it was designed to have a large number of samples on the leading edge of the waveform to aid in the accurate interpretation of the water-surface height. Unlike later satellites, the Geosat satellite was allowed to drift rather than using propulsion to keep it on exact orbital paths; hence, the tracks are unevenly spaced.
The Jason-1 satellite was a joint sponsored mission by NASA and CNES to monitor global ocean circulation in a 10-day repeat cycle. After a mission duration of over 10 years in repeat orbit it was moved into its ‘graveyard’ orbit to protect the prime orbit path for future missions. This change in orbit resulted in a full year of a geodetic mission before the satellite finally failed in June 2013. As well as the additional coverage, the benefit of including Jason-1 in this study is its shallow orbit inclination of 66°, which adds a different orientation of tracks to the network. An additional benefit here is that the Jason-1 satellite was in a prograding orbit compared with the other satellites which were in a retrograding orbit. The alteration in the direction of the satellite orbit in certain situations provides additional data due to the slight delay in tracking the radar pulse as the satellite moves from land to water (whereas when it travels from water to land, a satellite can record right up to the coast).
The CryoSat-2 satellite was launched in 2010. The main aim of the satellite was to measure the polar ice extents, meaning the orbit has a high angle of inclination, giving coverage to within 2° of the poles. During one 369-day cycle, the satellite completes a full geodetic mission with orbital spacing of 7.5 km at the equator. CryoSat-2 has two unique features compared with the other satellites in this study: 1) it carries two altimeters and 2) it can record in three different modes. The first mode is common to all satellites in this study and is referred to by ESA as the low resolution mode (LRM). This mode uses a conventional pulse limited radar pulse that has a circular footprint with a diameter of ~5 km. The next mode is the synthetic aperture radar (SAR) mode, which uses delay Doppler techniques to improve along-track resolution. The cross-track footprint is ~5 km, like the LRM mode, but the along track footprint is much narrower at ~0.29 km. The final mode, SAR-Interferometry (SARIn) uses two antennas to form a cross-track interferometer. This mode is primarily used to measure elevation and cross-track slopes over land-ice surfaces. Over the African lakes, the CryoSat-2 satellite is often operated in SARIn mode. This mode is rarely used in open ocean areas as the accuracy of the data is degraded by surface slopes and roughness, however calmer water conditions are often found in lakes compared to open seas, so incorporating SARIn data into the workflow provides valuable additional coverage.
An example of processed lake data is shown in Figure 2 for Lake Tanganyika (Tanzania/DR Congo/Burundi/Zambia). The track coverage for the Geosat, Jason-1 and multiple years of CryoSat-2 SARIn data result in very good coverage of the lake-surface height (Figure 2a).
Figure 2 – Qualitative comparison between gravity data for Lake Tanganyika and the structures derived from seismic interpretation. a) Final satellite track coverage. b) Bouguer gravity overlain with structures from Rosendahl et al. (1988).
Figure 2b shows a high-pass filtered version of the final derived Bouguer gravity data. For comparison to other data sets, the structural interpretation from the PROBE seismic survey (Rosendahl et al., 1988) is overlain. This particular survey was conducted between 1983 and 1984, and involved the collection of ~2,000 km of 2D seismic data over Lake Tanganyika. The gravity data show a very good correlation with the seismic structures and can add value as they could be used to increase confidence in the interpolation of features away from and between the 2D seismic lines, as well as adding qualitative and quantitative information regarding the subsurface density structure.
Where available, the gravity results appear to match well with seismic interpretations and airborne gravity surveys where these are existing over the lakes. These data therefore have the potential to enhance the understanding of the exploration potential of key areas giving complete coverage over lakes that have previously been sparsely surveyed.
Written by Samuel Cheyney
Green, C., Fletcher, K., Cheyney, S., Dawson, G., Campbell, S. 2019. Satellite gravity – enhancements from new satellites and new altimeter technology. Geophysical Prospecting. 67. pp1611-1619.
GRAV-D Team. 2013. Gravity for the redefinition of the American vertical datum (GRAV-D) project, airborne gravity data; Block EN05. Available March 2017. Online at: http://www.ngs.noaa.gov/GRAV-D/data_EN05.shtml
Haxby, W. F. 1985. Gravity field of the world’s oceans. U.S. Navy, Naval Office of Research.
Rosendahl, B.R., J.W. Versfelt, C.A. Scholz, J.E. Buck, L.D. Woods. 1988. Seismic atlas of Lake Tanganyika. Project PROBE geophysical atlas series. V1. Duke University, North Carolina, pp. 120.
