Giant submarine landslides in gas hydrate provinces: a comparison of the Nile and Amazon deep-sea fans – MEGA

The project is based on the integration of observation of natural phenomena using geophysical, geological and geochemical data, and numerical simulations.

1. Observation of fluid seepages/gas hydrate and mega-slides.

Characterising the seabed morphologies and subsurface architectures of fluid seepages associated with gas hydrates, and mega-slides from geophysical and geological data will enable us to understand the extent and characteristics of the gas hydrate system in each deep-sea fan, and to constrain their relationships with the distribution of structures facilitating the upward migration of fluids towards the seabed. The innovation will come from the use of deep learning to automatically detect and map fluid outflows on the seabed.

The mega-slides will be identified and mapped using geophysical data, then their stratigraphic periods of emplacement will be estimated and compared with climato-eustatic variations. This will make it possible to estimate their dynamics (single giant ruptures or multiple small and stacked landslides) and assess their translation mechanisms (fast or slow movement).

2. Modelling the dynamics of gas hydrates, the triggering of mega-slides and tsunamis.

The dynamics of gas hydrates will be modelled using an innovative combination of two software packages: on a basin scale using TemisFlow to estimate the quantities of microbial and thermogenic gases generated and their migration in the 10 km thick sedimentary pile; and on the scale of the GHSZ close to the seabed (<0. 5 km) using Tough+Hydrate to understand the dynamics of gas hydrates in response to the climato-eustatic forcing, and variations in heat and upward fluid/gas fluxes. The palaeo-temperatures of the bottom water will be estimated using Mg/Ca analyses carried out on the benthic foraminifera from the cores.

Simulation of the triggering of mega-slides will provide scenarios of slope failures associated with the dissociation of gas hydrates. Commercial finite-element software (OPTUM G2) will be used to quantify the reduction in resistance and pore fluid pressure conditions that could explain this triggering despite the low-angle slopes. Representative mechanical properties of the sediments will be obtained from existing boreholes and from geotechnical measurements carried out on the cores.

The formation of tsunamis in response to mega-landslides will be assessed in the Eastern Mediterranean and South Atlantic using previous results. This is of great scientific and societal importance, as historical mega landslides are unknown, as is the intensity of their impact on densely populated coastlines. This work will use the numerical code Shaltop, which simulates the displacement of landslides, coupled with the 3D Navier-Stokes code Freshkiss3d.

Halfway through the project, the first two tasks of observing natural phenomena have been completed for the two deep-sea fans of the Nile and the Amazon. The first task concerns the identification, mapping and characterisation of all the geological structures present at or near the seabed. The second task concerns the identification, characterisation and dating of mega-slides during the Plio-Quaternary period.

For task 1, we concentrated on the recognition and analysis of the same geological objects within the two study areas. These correspond to 1) fluid-seepage structures (mud volcanoes and pockmarks, presence of mudflows on the seabed and plumes of gas bubbles in the water column), 2) indicators of the presence of gas hydrates (BSR), 3) tectonic structures (faults, folds) linked to the gravitational collapse of the two deep-sea fans, 4) submarine landslide features (scars, erosive conduits, mass-transport deposits, compression ridges, debris flow deposits), 5) sedimentary architectures (canyons, channels, dunes).

For Task 2, using seismic data, we extended the stratigraphic interval studied to the Upper Miocene. On the Amazon, we have identified 25 very large mega-slides (200-20000 km3), the largest of which may correspond to the superposition of several smaller slides. On the Nile, 21 mega-slides of smaller volume (10-2000 km3) were identified, each corresponding to an individual event.

This work shows that the two study areas contain comparable structures, but in different proportions and with different characteristics. Mapping through the development of deep learning methods has been successful in analysing the fluid seepages associated with the escape of gas into the ocean, providing results that still require further investigation. Hundreds of fluid seepages have been recognised in the two study areas, forming distinct populations that are in different relationships with the BSR, which takes the form of discontinuous ‘patches’ associated with tectonic structures. On the Amazon cone, the BSR is strongly associated with the crests of anticlines, which are also the sources of the fluid overpressures at the origin of mud volcanism. On the Nile cone, the BSR tends to be associated with basins bounded by extensional faults, which contain little or no fluid seepages; on the other hand, pockmarks are abundant along the westward slope where faults do not cut the seafloor but are buried beneath stacked mega-slides. The seafloor of both deep-sea fans is heavily affected by landslide scars and associated deposits, the characteristics of which suggest that fast-dynamic landslides are predominant.

Once the tasks of observing and characterising natural phenomena had been completed, we began the first two modelling tasks.

For task 3, the typical architectures of the two margins of the Amazon and the Nile were finalised for import into TemisFlow.

For task 4, a synthetic depth section was validated for each of the two margins in order to begin the first slope stability modelling tests.

Giant submarine landslides (10-2000 km3) are found in the thick Quaternary sediment succession of passive continental margins. Their ages coincide with periods of sea-level fall and rise, but it is unclear how such vast failures can be triggered on low seafloor slopes (<2?) in the absence of a triggering factor such as seismicity. Key hypotheses involve excess pore pressures linked to reductions in gas-hydrate stability, driven by changes either in climate or in subsurface fluid flow. The MEGA project wants to explore such hypotheses through the first modelling of linked changes in gas hydrate and slope stability in response to ocean pressure and temperature changes, using an innovative comparison of the Nile and Amazon deep-sea fans that experience different forms of climate forcing over glacial-interglacial timescales. As such megaslides have never triggered in historical times, MEGA will provide input for the first modelling of their tsunamogenic consequences on coastal zones.