The thermochemical structure of the lithosphere and upper mantle beneath south China: Results from multiobservable probabilistic inversion

Abstract

We study the thermal and compositional structure of south China by jointly inverting Rayleigh wave dispersion data, geoid height, topography, and surface heat flow with a probabilistic (Bayesian) Monte Carlo method. We find that the lithosphere is thin (85-150 km) beneath the South China Fold system and thickens over the Yangtze Craton to maximum thicknesses of up to 250 km beneath the Sichuan Basin. Our inversion predicts that the lithospheric mantle beneath the South China Fold system and Yangtze Craton is highly fertile (Mg#~88-90), in agreement with independent geochemical signatures from East China mantle xenoliths. Such fertile compositions, together with the relatively thin lithospheric thickness in the area, point toward a widespread metasomatism/refertilization event. We suggest, as others have, that a flat-subduction episode and subsequent slab removal may have triggered both the delamination of the lowermost part of the subcontinental lithosphere and the generation of asthenospheric melts that metasomatized (refertilized) the remaining lithospheric mantle. Inconsistencies among geophysical observations and anomalously fertile compositions for the Sichuan Basin indicate that this region may be currently affected by small-scale convection or delamination processes. Alternatively, the anomalous observations may be associated with an eastward push of Tibetan lithosphere beneath the Yangtze Craton.

Physical State and Structure of the Crust Beneath the Western‐Central United States From Multiobservable Probabilistic Inversion

“…In this study, we employ a recently developed multiobservable probabilistic inversion approach (Afonso, Fullea, Griffin, et al, ; Afonso, Fullea, Yang, et al, ; Guo, Afonso, et al, ; Jones et al, ; Afonso, Rawlinson, et al, ; Qashqai et al, ; Shan et al, ) based on a Bayesian inference framework. In this framework, prior information on both data and model parameters is encoded in a joint probability density function (PDF) denoted by ρ ( d,m ).…”

“…Since the parameterization of the crust and the upper mantle have been fully described elsewhere (e.g., Afonso, Fullea, Yang, et al, ; Afonso, Rawlinson, et al, ; Qashqai et al, ; Shan et al, ), here we only summarize the most relevant aspects for this study. Our 3‐D domain is made up of a collection of adjacent and nonoverlapping 1‐D columns coincident with the location of the seismic stations.…”

“…Discretization scales used to model the 1‐D Earth structure beneath each station (Triangles in Figure a), modified from Afonso, Fullea, Yang, et al (); Shan et al (). The finest mesh with vertical grid step of 2 km (red dots) is used to solve the forward problems associated with elevation, geoid, and surface heat flow data (see text).…”

“…The discretization of the upper mantle is similar to that used in Afonso, Fullea, Yang, et al (); Shan et al (); Guo, Chen, et al (); and Afonso, Rawlinson, et al (). The entire 1‐D model (crust + upper mantle) is discretized with a finite‐difference mesh with a resolution of 2 km (red dots in Figure ), to solve the forward problems associated with geoid height, elevation and the computation of lithospheric geotherms (see next section).…”

“…The sublithospheric temperature structure is entirely controlled by three unknown parameters (or temperatures) at different node‐depths (Figure ‐ T b u f f e r , T i n t e r , and T b o t t o m ). Once these temperatures are randomly sampled by the MCMC sampler (see their ranges in Table ), the temperature profile for the sublithospheric mantle is constructed by linearly interpolation between T 1250 , T b u f f e r , T i n t e r and T b o t t o m (e.g., see Afonso, Rawlinson, et al, ; Shan et al, ). Since the focus of the present work is on the crust, we use a prior for LAB depth based on the results in (Afonso, Rawlinson, et al, ), who focused on the mantle thermochemical structure in the same region using a full 3‐D multiobservable thermochemical tomography approach.…”