Where are you getting your numbers? They sound to me like they come from the National Crime Victimization Survey. These are not reports to police, but the result of asking random people if they have been raped. I don't think that they sample prisoners, so they are probably highly biased against prison rape, but should catch some.