I certainly don't know the answer in this case, but to understand why sacrifice might be rational, you have to really contort your thinking and imagine your cost vs. benefit analysis when depriving your thought from centuries of advances in philosophy of science:
For the early humans many things obvious to us now were simply not known. The idea of an omnipotent sun god who controls the seasons isn't a priori crazy or anything. Why would it be? The normalization of how the solar system works only seems rational now because it's been scientifically presented to us.
In that sense a prediction that some entity controls this phenomena isn't necessarily superstitious. It's a fine prediction. But how do we get from there, to sacrifice?
Well, if we don't understand the mechanism of action for how these systems work, we can still try to make inference on them using statistics. For example: Sacrifice someone, then observe if anything changes. Before anyone ever ran this experiment, it's not obvious it wouldn't work. It's only obvious to us now because (a) we know the mechanism of action has no plausible link to human sacrifice, and (b) because people ran the experiment before, and it didn't work.
At the time, however, the benefit to this working would be huge, and the cost to society as a whole relatively small. This type of proto-statistical experimentation is, I think, what eventually led humanity to its core discoveries. In that sense we can't retroactively say the experiments that didn't work were superstitious or irrational, but the ones that did work were clever. For example, much of modern medicine arose from vast amounts of superstition and random experimentation.