In our non-linear, post-yield, impact computer models we simply declare the ground "infinitely rigid". This typically results in a conservative answer for what we are interested in, which is container or "component" survivability. The container or "component" absorbs all of the impact energy. In the actual drop tests we use steel plates backed by many tons of concrete for the "infinitely rigid" ground.
As was pointed out, initial strike attitude is problematic. One interesting attitude, particularly for flanged lid containers is the "slap down". Consider a 2-D rectangular component of constant density. It falls such that initial impact is on one corner with the center of gravity CG overhanging the one corner. The rebound energy, energy not absorbed by local post-yield failure, may impart a rotational velocity about the CG. This can cause the second corner to strike with a local velocity greater than the CG velocity. Local forces at the second strike corner can be greater than the first strike corner. There will be a "sweet angle" that maximizes this effect. This attitude and many others (end on, side, corner with CG over the corner, drop on penetrator, etc.) are simulated and results evaluated to determine maximum damage attitudes.
For actual drop tests, the predicted worst case attitudes are attempted. As pointed out, it is difficult to manage the initial strike attitude. Internal accelerometers and strain gages, high speed cameras, etc., help to validate the model predictions. Also, as pointed out, post-impact assessment of damage is not trivial.
