Depending on the methods you are using, you will find a set of 2k spin orbitals formed by direct multiplication of K spatial orbitals by either alpha or beta spin function. Then, at first approximation, the negative of orbital energies is a good choice for ionization potentials (Koopmans theorem), in that sense one can label the highest occupied molecular orbital as HOMO (in your case 31 spin up or alpha orbital) and LUMO the corresponding lowest unoccupied molecular orbital (30 alpha in your case). When it occours degeneracies on the frontier orbitals you may get exactly the same energy values in the set alpha or beta (for example H2 molecule). Thus the criteria one thinks is useful is to consider the energy amounts of spin orbitals and compare them in order to say wich one are highest or lowest and finally the occupancy for a clear labeling. To clarify some ideas you may read Szabo and Ostlund chapter 2 and 3. Hope it helps 

Oh, I wanted to say 32 for LUMO sorry for confusion. :)

Jose Alberto Guerrero Cruz, Is it correct to calculate the HOMO-LUMO energy gap in the same way for a metal complex with doublet as spin state?

i.e., LUMO-alpha (32) - HOMO-alpha (31)?