Having realized that there is a problem, we must now clearly define it before we can try to solve it. Based on the work done by Muerdter et al. 1996, we hypothesize that we are seeing an illumination problem. Raytracing is an obvious way to investigate where the energy is going.
Creating raypaths is a straightforward process. We first used a paraxial raytracer to shoot rays from each source location to the section on the reflector being investigated. We then matched the rays at the reflector that obeyed Snell's Law, looping over the offset range in increments of 25 meters. The resulting raypaths for the entire offset range have already been displayed in Figure 4.
We have already shown the existence of a shadow zone in Figure 4. This figure shows the coverage obtained with the full offset range used in the synthetic model. Comparison with Figure 5 shows a definite lack of density of rays in the area where the reflector disappears from the stacked section.
It is clear that the primary problem in this area is poor illumination. However, there is some energy reaching this portion of the reflector. The next task was to determine what range of offsets contributed to the energy received from this area. From Figure 6 it appears that there is some coherent energy in a mid-offset range. Figure 7 shows that the near offsets (160 m - 1560 m) do not contribute, nor do the offsets beyond 2360 m (Figure 8). Figure 9 shows the mid-offset range that does contain the energy from the area we are interested in. The offset range that does contribute is between 1560 m and 2360 m. This compares favorably to the coherent energy visible in Figure 6.
Since we know that some energy gets through to this portion of the reflector, the next question is where does it go? To answer this, we placed a point source on the reflector at the point of the mysterious gap and traced rays from the reflector to the surface. Figure 10 shows the resulting irregular energy spread at the surface. This figure seems to indicate that illumination coverage may be improved by using wider offsets. However, this is an acquisition decision, not a processing one.