There are some limitations on such balloons already. For example, if the payload is 4-6lbs, stricter rules apply if the weight/size ratio is greater than three ounces per square inch (measured by the smallest surface on the payload).[1]

Also for larger balloons, any trailing antenna must break if subjected to an impact force of 50lbs, or the antenna must have colored streamers every 50ft.

The ideal measurement would be some sort of crash testing. eg: The payload is accelerated at some standard velocity towards some standard target that represents the weakest part of an airplane (either cockpit glass or leading edge of a wing) and must not damage the target beyond some threshold. But that seems like it would be expensive, since every change in payload would require re-testing. Limits on sectional density seem like a good compromise.

1. https://www.ecfr.gov/current/title-14/chapter-I/subchapter-F...