Reflection geophysics, or at least how we apply reflection in the modern day, has made me think quite a bit of how they used to do it back in the day. I have a fair idea how gravity and magnetic anomalies were interpreted and modelled in those dark times, but when we get to reflection and refraction, similar construction models are already quite a bit more computationally intensive.
If anyone does know how the first reflection surveys were modelled and interpreted in the ’20’s and ’30’s, please do inform me, but until I learn some more either through educated guesses or finding it out, I’ll try to bring up what strikes me as the difficult part there, at least when I look at it and see how we do it in the modern day.
Firstly, a number of shots in different points along a line are shot and recorded. I can imagine this recording device being a simple seismograph at some point in the beginning (or a device operating like a seismograph would, with a needle mapping the perturbations).
Secondly, we need to stack these shots, after removing noise. Now this is the difficult part — admittedly, noise removal would not be that complex although I would believe there’s trouble to be had when we’d try to use a seismograph to map a direct arrival vs a refracted/reflected one. But, this could be possible depending on how it is set up. But stacking? Without the ability to automatically add up different shot results from various locations, the method that remains possible is manually going through seismograph records and adding up the measured amplitudes to create an approximated stack (which would be, naturally, wrong in a variety of ways due to the inherent inaccuracy of manually re-graphing data which is not that accurate by itself, but it probably went through a function like this).
Thirdly, once we do have a stack, we should actually be able to base something on it, and try some interpretation.
Or, it could also be that the guys involved knew that the equipment and method was not refined enough and they met the above process somewhere half-way through with a proposed model that they then tried to fit into the ground. Say, for simplicity and ease of explanation that the scientist thought there would be largely four different layers, one of which relevant to the purposes of what he was looking for. He would then, based on the locations of shot-points and shot-gathers, calculate what those shots would look like if the subsurface actually looked like what he thought it looked like.
I am unsure how many tries and tests the second way would have to go through, but I guess that it would truly be part of one of those older things that people speak of when science was half art, and the best scientists could guess and then approximate based on their guesses so that in essence they were just proving themselves correct with data. And if it was anything like this, then I am both sorry and happy that I can use a variety of complex programs for the same purpose. Happy because I have quite a bit easier a life for myself with less chance of going horribly wrong for fifty times in a row; and sadder because the cultivation of such a skill as guessing anything as complex as that must have been very very interesting.