Notice that one of the s/c zenith angles is missing. For whatever reason that image didn't correlate well and was excluded from the solution. This did not happen on all landmarks in this test, but it did happen for more than half.
Solution is surprisingly good given this is from a single sun angle (see Data Set above). However there are many details that SPC doesn't get right, such as an apparent oscillation away from the boulder, and a dip in the E-W transit on the right of the plot. The dip could be due an artifact from nearby topography, but it is not seen in the other Sun angle tests.
Notice that the height of the boulder in this solution is always less than the truth, and the surrounding topography is often greater than the truth.
Heat Maps (x and y is in units of pixels, z is deviation from truth and has units of meters)
Iteration 1 and 10: The solution starts out with a N-S tilt.
Iteration 20 and 30:
Iteration 50 and 75:
Iteration 100: The solution continually improves, but still has some pretty big excursions from truth, also note that the solution no longer has a N-S tilt, but is more like a convex curvature where the middle is higher than the edges.