h2 on liability scale in longitudinal studies

1. Does it make sense to calculate the h² on the liability scale if you have a longitudinal study?

2. What could explain a big difference between the observed and the linear transformed h², when the --prevalence is equivalent to the observed prevalence?

i.e.

(Proportion of cases in the sample = 0.103185; User-specified disease prevalence = 0.100000)
h2 observed scale (s.e.) = 7.8 % (6.5)
h2 liability scale (s.e) = 22.1 % (18.4)

My syntax looks like this:
qsub-bin-short gcta --reml --grm CIDI_new --pheno CIDI_nmiss.phen --prevalence 0.10 --qcovar Covar_CIDI.qcovar --covar Sex.covar --out CIDI --thread-num 10

Thank you in advance.

Raluca