Seismic refraction shootout: Blind test of methods for obtaining velocity models from first-arrival travel times

Click here to download the travel time data.

The data consist of P-wave travel times for a synthetic seismic refraction experiment through a velocity model that is intended to represent a realistic near-surface geologic situation. The site geology is known to consist of unconsolidated, or semi-consolidated, sediments overlying a non-flat bedrock surface. The water table depth is known to vary considerably in the area, between 20 and 100 m. The ground surface at the survey site is completely flat; that is, zero topography exists along the seismic transect. The survey objective is to identify low velocity zones within the sedimentary layers, as well as features of the bedrock surface and within the bedrock. Though not specifically intended to represent a karst environment, aspects of the geologic model are similar to what might be found in a karst environment.

The compressed ascii data file contains 3 columns. The first and second columns are the horizontal position of the source and receiver (m), respectively, and the third column is the travel time (ms). The elevations of the sources and receivers are zero. The travel times correspond to P-wave first arrivals, and were calculated assuming a 100 Hz wave. Uncorrelated Gaussianly-distributed noise with a standard deviation of 1 ms was added to the synthetic data. There are 101 shots from 0 to 300 m, and 100 receivers from 1.5 to 298.5 m. The shot and receiver spacing is 3 m. The total number of travel times is 10100.

To participate in the session at the SAGEEP 2011 meeting (April 10-14), you must submit an abstract by November 19. You will have until April 1 to develop your model(s) and submit them to Colin Zelt. The session at SAGEEP 2011 will consist of presentations from each submitter, intended to describe the approach taken and to present the resulting velocity model(s). This will be followed by a wrap-up presentation that reveals the true model and compares the submitted models at the same scale and with the same display parameters. The session is intended to be an informative opportunity for all interested scientists to better understand the options and trade-offs involved in creating seismic velocity models from refraction data. Submissions are encouraged from students new to seismic work, along with experienced experts.

Please email your final model(s) to Colin Zelt by April 1, 2011, in ascii format with 3 columns (x,z,v). Also, please provide the RMS misfit between the times from the true model and those predicted by your model(s).

More information about the session, SAGEEP meeting, and how to submit an abstract, is available in this PDF file.

Email me at if you have any questions.

Colin Zelt


The true model is now available for download. This is an ascii file with 3 columns, X, Z and velocity. The grid spacing is 0.5m.

Full waveform data is now available for download. This file is in SEGY format with 25 shot gathers evenly spaced (12.5 m spacing) across the model from left to right. The data were calculated using a 2D acoustic finite-difference code without Q attenuation. The file is 35Mb in size.