MODEL RESULTS

The input to RINSE is a simple cosine shaped mountain with an elevation of 800 m and a width of 1500 m. The model consists of a 50 m thick low velocity layer which follows the surface topography. The low velocity layer has a velocity of 1200 m/s and overlays a half space of velocity 2300 m/s. The synthetic data is split spread with 200 shots, 96 groups per shot. The shot and group intervals are both 30 m.

Two sets of synthetic data are generated for this model. In the first data set, only refracted head wave arrival times are calculated. The second data set is the realistic simulation because the transmitted arrival times are calculated.

The results of the refraction inversion are presented in Figure . The top plot is the RINSE refractor model generated by inverting on the unrealistic head wave arrival only data. This results in a nearly perfect reconstruction of the input model. The bottom plot is the RINSE refractor model generated by inverting on the realistic transmitted arrival data. The stars on the plot represent shot location and the lines emanating from the shots represent the location of the refractor as determined by RINSE. The presence of transmitted arrivals in the data causes the pronounced thickening of the refractor model over the top of the mountain. The erroneous thickness is about 170 m, instead of the actual thickness of 50 m. This results in a static error of up to 45 ms in one way time.

This simple model demonstrates that the thickening over mountains observed in refraction inversions is due to the fact that the data does not fit the assumed refraction model. When only refracted head wave arrivals are included in the inversion (top of Figure ) , the assumptions of RINSE are satisfied and the program works well. When the transmitted arrivals are included in the analysis (bottom of Figure ), the program fails.

The effect of transmitted arrivals on refraction inversion varies with elevation, spread length, and velocity. It is found that models identical to the one in Figure  but with elevations of 200 m and 400 m result in one way static errors of up to 7 ms and 17 ms, respectively. If spread length is decreased, the effect is the same as decreasing the elevation (or curvature) of the mountain. The static error decreases. The error in the inversion increases in regions of lower velocity and decreases in regions of higher velocity. This is because the difference in arrival time between transmitted arrivals and the assumed refracted arrivals is less when the velocity is higher.