If you have a hydraulic jump upstream of a culvert, that would imply that just upstream of the jump, you have supercritical flow. By definition, you cannot have a backwater extend upstream through supercritical flow. Therefore, you cannot have a rise upstream of the jump. If this is a natural river/stream, I wonder whether you really have supercritical flow, perhaps your n values are too low. In any case. If the rise 900 feet upstream (assuming your mean just downstream of the hydraulic jump, where the flow is subcritical) decreases as your culvert size decreases, something is wrong with how the model is set up, or something else is changing as well, when the culvert size decreases. This is all assuming I correctly understand your problem.