Ionosonde data from sixteen stations are used to study the semiannual and a
nnual variations in the height of the ionospheric F2-peak, hmF2. The semian
nual variation, which peaks shortly after equinox, has an amplitude of abou
t 8 km at an average level of solar activity (10.7 cm flux = 140 units), bo
th at noon and midnight. The annual variation has an amplitude of about II
km at northern midlatitudes, peaking in early summer; and is larger at sout
hern stations, where it peaks in late summer. Both annual and semiannual am
plitudes increase with increasing solar activity by day, but not at night.
The semiannual variation in hmF2 is unrelated to the semiannual variation o
f the peak electron density NmF2, and is not reproduced by the CTIP and TIM
E-GCM computational models of the quiet-day thermosphere and ionosphere. Th
e semiannual variation in hmF2 is approximately "isobaric", in that its amp
litude corresponds quite well to the semiannual variation in the height of
fixed pressure-levels in the thermosphere, as represented by the MSIS empir
ical model. The annual variation is not "isobaric". The annual mean of hmF2
increases with solar 10.7 cm flux, both by night and by day, on average by
about 0.45 km/flux unit, rather smaller than the corresponding increase of
height of constant pressure-levels in the MSIS model. The discrepancy may
be due to solar-cycle variations of thermospheric winds. Although geomagnet
ic activity, which affects thermospheric density and temperature and theref
ore hmF2 also, is greatest at the equinoxes, this seems to account for less
than half the semiannual variation of hmF2. The rest may be due to a semia
nnual variation of tidal and wave energy transmitted to the thermosphere fr
om lower levels in the atmosphere.