[14] The main results of our analysis can be summarized as following:
[15] 1. Geomagnetic activity thresholds expressed in ap index units for F2 -layer negative storms onset calculated for 14 midlatitude ionosonde stations exhibit a pronounced seasonal (winter/summer) variation with summer thresholds being larger than winter ones. There are also relative minima (or plateau) in the threshold during equinoctial periods.
[16] 2. Seasonal difference in the thresholds is due to seasonal difference in the thermosphere reaction to geomagnetic disturbances: a steep dependence of D (O/N2 ) on geomagnetic activity in winter but gently sloping dependence in summer. This needs higher geomagnetic disturbance level (higher threshold) in summer compared to winter to have the same perturbation effect in the thermospheric parameters.
[17] 3. A proposed explanation for the threshold seasonal variations is based on seasonal changes in neutral temperature and on seasonal difference in the spatial distribution of the perturbed neutral composition. Higher background Tn in summer needs stronger disturbance level (larger threshold) to have the same changes in O/N2 compared to winter when Tn is lower. The other effect is due to the interaction of seasonal (background) and storm induced thermospheric circulation. In summer the [O]/[N2] disturbance zone is known to extend over the whole hemisphere all the way from the polar to the low latitudes while in winter it is restricted to high latitudes only. Therefore more energy should be deposited in the auroral zone during a geomagnetic storm in summer compared to winter to have the same perturbation effect in the thermospheric parameters.
[18] 4. The equinoctial plateau in the threshold annual variations may be related to day-to-day changes in the thermospheric circulation and atomic oxygen abundance during the equinoctial transition periods.