Crustal structure of southern Madagascar from receiver functions and ambient noise correlation: Implications for crustal evolution

Abstract

The Precambrian rocks of Madagascar were formed and/or modified during continental collision known as the Pan‐African orogeny. Aborted Permo‐Triassic Karoo rifting and the subsequent separation from Africa and India resulted in the formation of sedimentary basins in the west and volcanic activity predominantly along the margins. Many geological studies have documented the imprint of these processes, but little was known about the deeper structure. We therefore deployed seismic stations along an SE‐NW trending profile spanning nearly all geological domains of southern Madagascar. Here we focus on the crustal structure, which we determined based on joint analysis of receiver functions and surface waves derived from ambient noise measurements. For the sedimentary basin we document a thinning of the underlying crystalline basement by up to ∼60% to 13 km. The crustal velocity structure demonstrates that the thinning was accomplished by removal or exhumation of the lower crust. Both the Proterozoic and Archean crust have a 10 km thick upper crust and 10–12 km thick midcrust. However, in contrast to the typical structure of Proterozoic and Archean aged crust, the Archean lower crust is thicker and faster than the Proterozoic one, indicating possible magmatic intrusions; an underplated layer of 2–8 km thickness is present only below the Archean crust. The Proterozoic mafic lower crust might have been lost during continental collision by delamination or subduction or thinned as a result of extensional collapse. Finally, the Cretaceous volcanics along the east coast are characterized by thin crust (30 km) and very large VP/VS ratios.

The structure of the crust and uppermost mantle beneath Madagascar

“…In contrast, there are no significant differences in Poisson's ratio between the Archean crust (0.25 ± 0.02) and Proterozoic crust (0.26 ± 0.04) or in average shear wave velocities (3.7 km s -1 for both), though the average thickness of the mafic lower crust is slightly greater for Archean terranes (7 km) than it is for Proterozoic terranes (4 km). Our results are in broad agreement with Rindraharisaona et al (2017). They concluded that the Archean crust is also slightly thicker (38-43 km) compared to the Proterozoic crust (33-38 km).…”

“…S4, in this study), a clear phase, presumably the Ps wave converted from the Moho discontinuity, is observed at ∼4 s. This indicates thinner crust (∼36 km) as reported in this study and earlier ones using receiver functions (e.g. Rindraharisaona et al 2013Rindraharisaona et al , 2017, which used more than one station along the eastern coast and considered broader back azimuth ranges. In support of this interpretation, crustal thickness estimates from gravimetry imply that the crust thins towards the eastern coast (Fourno & Roussel 1994;Rakotondraompiana et al 1999).…”

“…The surface-wave tomography study of Pasyanos & Nyblade (2007) found crustal thicknesses in Madagascar to vary between 25 and 35 km, and the global CRUST 1.0 model shows crustal thicknesses ranging from 36 to 45 km (Laske et al 2013). Recently, the crust in the southern part of Madagascar was investigated by Rindraharisaona et al (2017) using the full SELASOMA data set, as well as data from additional stations in southern Madagascar. Using joint inversion of receiver functions and ambient-noise derived surface waves, they revealed a thinning of the crystalline crust to 13 km in the Morondava basin, and a slightly greater thickness of the Archean crust (∼39 km) compared to the Proterozoic crust (∼35 km).…”

Crustal Radial Anisotropy and Linkage to Geodynamic Processes: A Study Based on Seismic Ambient Noise in Southern Madagascar

“…We used the results of Rindraharisaona et al () to construct starting velocity models for the Rayleigh wave inversion. For the subsequent Love wave inversion, the final velocity structure derived from the Rayleigh wave inversion was used as the starting model (R L ordered inversion).…”

“…In order to test the dependence of the inversion on the starting model, we performed R L and L R ordered inversions using six different velocity models: two gradient models and four realistic models representing the subregions of southern Madagascar, that is, the Morondava basin, Proterozoic domains, Archean domains, and Cretaceous volcanics (along the east coast). The realistic models were generated from the results of Rindraharisaona et al () by taking the median velocity for each layer from the models for all stations in each subregion. Although each of the grid nodes is located in just one of the four subregions, we always tested all six starting models to assess the uniqueness of the inversion.…”