000004285 001__ 4285
000004285 005__ 20191003080152.0
000004285 037__ $$aPOSTER-2019-0096
000004285 100__ $$aVan Camp, M.
000004285 245__ $$aComparing global seismic tomography models using the varimax Principal Component Analysis
000004285 260__ $$c2019
000004285 269__ $$c2019-12-10
000004285 500__ $$aCollaborative work with U. La Rochelle
000004285 520__ $$aClassical analysis of new tomography models consists in a comparison with others by correlation, spectral profiles, or localisation of patterns. To interpret the models in a quantitative, objective way and ease comparisons, we analyse the model information using principal component (PC) analysis. The varimax criterion applied to the PCs separates modes associated with different depth ranges. This enables determining the importance of different parts of the tomography models when reconstructing them. We apply this method to the isotropic part of 6 global shear-wave speed models: SAVANI (Auer et al., 2014), S20RTS (Ritsema et al., 1999), S40RTS (Ritsema et al., 2011), SEMUCB-WM1 (French & Romanowicz, 2014), SGLOBE-rani (Chang et al., 2015), and S362WMANI+M (Moulik & Ekstrom, 2014). According to the models, the method generates 7 to 15 independent varimax PCs, capturing more than 97% of the total information. A comparison of the PCs with the information extracted from the full models shows that no interpretable information is lost.  Each mode is composed of a vertical anomaly profile, to which we associate a horizontal pattern by orthogonal projection. The maximum of the depth profile and the geographical distribution of the horizontal pattern enable examining the key characteristics of the main components of the models.  For a fair comparison, we also compute a varimax PCA on a concatenation of the 6 models together. This imposes a projection of the average vertical profile, which allows for a mode-by-mode comparison between the model set. Similar main regions are identified when applying the analysis either to the individual tomography models or to the concatenation of the models: (i) Large Low Shear Velocity Provinces (LLSVPs); (ii) Mid-lower mantle (~1,200 km depth) showing some deep subduction signals and low-velocity anomalies beneath the Pacific and Africa, possibly associated with mantle plumes; (iii) Uppermost lower mantle (~800 km depth), also with deep subduction and low-velocity in the southeastern Pacific ocean; (iv) Transition zone (~400 km depth), showing subduction and low velocity anomalies beneath the Pacific and Indian oceans; and, (v) ridges and cratons in the uppermost upper mantle (~200 km depth). We discuss the significance and potential implications of these main regions that are identified.
000004285 594__ $$aNO
000004285 6531_ $$aSeismic Tomography
000004285 6531_ $$aPrincipal Component Analysis
000004285 6531_ $$aEarth's interior
000004285 700__ $$ade Viron, O.
000004285 700__ $$aFerreira, A.M.G.
000004285 773__ $$tAGU 2019 Fall Meeting, San Francisco
000004285 8560_ $$fmichel.vancamp@observatoire.be
000004285 980__ $$aCPOSTER