Cadence spectre phase noise plots for oscillators

Phase noise plots for oscillators using Cadence spectre (PSS+PNoise) in dBc/Hz show noise values greater than 0 at close in frequencies to the carrier (including several plots in their user guide)

How is this possible?

Close in PN is in fact "Shot Noise Contribution" and Shot Noise can not be modelled well due to its natural cause is not well defined.But simulators should do an extrapolation and that's why close-in PN may be greater than zero.

Thank you Bigboss!