Yes, that’s the correct procedure. ROHF applied to a closed-shell system simplifies to RHF. If you want to be paranoid, write out orbitals using Molden and verify that the open-shell orbital is largely localized on O2. As long as it is, you should be fine.