Premium
ITERATIVE TOMOGRAPHIC METHODS TO LOCATE SEISMIC LOW‐VELOCITY ANOMALIES: A MODEL STUDY 1
Author(s) -
KRAJEWSKI CHRISTINA,
DRESEN LOTHAR,
GELBKE CHRISTOPH,
RÜTER HORST
Publication year - 1989
Publication title -
geophysical prospecting
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.735
H-Index - 79
eISSN - 1365-2478
pISSN - 0016-8025
DOI - 10.1111/j.1365-2478.1989.tb02231.x
Subject(s) - slowness , tomography , inversion (geology) , smoothing , geology , tomographic reconstruction , grid , synthetic data , geometry , algorithm , optics , geodesy , mathematics , physics , seismology , tectonics , statistics
The possibilities for reconstructing seismic velocity distributions containing low‐velocity anomalies by iterative tomographic methods are examined studying numerical and analogue 2D model data. The geometrical conditions of the model series were designed to generalize the geometrical characteristics of a typical cross‐hole tomographic field case. Models with high (30%) and low (8%) velocity contrasts were realized. Traveltimes of 2D ultrasonic P‐waves, determined for a dense net of raypaths across each model, form the analogue data set. The numerical data consists of traveltimes calculated along straight raypaths. Additionally, a set of curved‐ray traveltimes was calculated for a smoothed version of the high‐contrast model. The Simultaneous Iterative Reconstruction Technique (SIRT) was chosen from the various tomographic inversion methods. The abilities of this standard procedure are studied using the low‐contrast model data. The investigations concentrate on the resolving power concerning geometry and velocity, and on the effects caused by erroneous data due to noise or a finite time precision. The grid spacing and the source and receiver patterns are modified. Smoothing and slowness constraints were tested. The inversion of high‐contrast analogue model data shows that curved raypaths have to be considered. Hence, a ray‐tracing algorithm using velocity gradients was developed, based on the grid structure of the tomographic inversion. This algorithm is included in the SIRT‐process and the improvements concerning anomaly localization, resolution and velocity reconstruction are demonstrated. Since curved‐ray tomography is time‐consuming compared with straight‐ray SIRT, it is necessary to consider the effects of grid spacing, ray density, slowness constraints and the