Introduction

The crust and upper mantle in southeastern Asia is highly heterogeneous, presenting a challenge for path calibration, but it is well surrounded by earthquake sources, and a significant number of high quality broadband digital stations exist. Using a finite-frequency 2D approximation (NACT, Li and Romanowicz, 1995), we have already developed a 3D radially anisotropic model in a large region (longitude 30 to 150 degrees and latitude -10 to 60 degrees) from the existing long period waveform database in the range of 60 s to 400 s. The database was collected at Berkeley over the last 10 years for the construction of global mantle tomographic models (Li and Romanowicz, 1996; Megnin and Romanowicz, 2000; Gung et al., 2003; Panning and Romanowicz, 2004), and to it are added data from  20 new events in the period up to to 2005.

A data set of 38826 3-component waveforms recorded at 169 stations from 393 events was used in the waveform inversion. The data have been processed with an automated algorithm, which removes glitches and checks for many common problems related to timing, poor instrument response, and excessively noisy windows. A weighting scheme has been applied to ensure even distribution of data across the region. The model is parameterized laterally in spherical spline level 6, which corresponds to lateral resolution of $\sim 200$ km. And the corresponding radially anisotropic model is parameterized in the spline level 5, which corresponds to a lateral resolution of $\sim 400$ km. In this study we will use a newly developed "non-linear" 3D Born approximation (N-Born) to refine the above NACT model (Panning et al., 2007).

Figure 2.50: Raypath coverage in the subregion. The background grey raypath coverage is denoted for NACT and the black raypath coverage is for N-Born. Stations and events are required to be in the large region for the N-Born inversion, and so the N-Born raypath coverage is a subset of that of NACT.