2. Data Analysis

[4]  An analysis of the F2 -layer storms at middle latitudes indicates the existence of a threshold for the ionospheric storm onset expressed in ap index units, the threshold being seasonal-dependent. The following analysis aims to clarify the question. As the effect is expected to depend on latitude, only midlatitude stations with close invariant latitudes ( F _inv 50 5^o ) were considered (Table 1). The initial experimental material, available hourly f_oF2 observations at the stations listed in Table 1. A 27-day f_oF2 running median centered to the day in question rather than usual monthly median was used in the analysis. The advantages of using such median were discussed by Mikhailov et al. [2004]. Only long-duration (  6 hours) negative disturbances with d=(N_mF2/N_mF2 _med -1) 100% more than 40% were analyzed. Such ionospheric disturbances may be considered to be related to changes in the thermospheric composition and temperature. The ionospheric storm is supposed to begin if d > 40% takes place during 4 successive hours at least. If this requirement is fulfilled, 3-hour ap indices for the previous 24 hour period were analyzed. To separate geomagnetic activity induced F2 -layer disturbances from quiet time ones ( Q disturbances [Mikhailov et al., 2004]), at least one of eight ap indices for the previous 24-hour period should be larger than 15. This choice is used in accordance with Kutiev and Muhtarov's [2001, and references therein] results which show that the most probable state of the ionosphere corresponds to kp 3₀ ( ap= 15 ) and on average negative disturbances correspond to geomagnetic activity level higher than ap = 15. The threshold was calculated as an average over 8  ap indices for the 24-hour period prior the ionospheric storm onset. The choice of 24-hour period is based on the empirical estimations of the ionosphere reaction to the forcing geomagnetic activity (see earlier). The thresholds for all the storms at each station were ordered, and an average of over 5 of the smallest values was referred as a threshold for a given month. The same analysis was applied for 10 the smallest thresholds as well, but the results turned out to be about the same: Only the absolute values of the threshold were larger as cases for more disturbed conditions turned out to be included into the consideration. No separation on solar activity level was made for two reasons. On one hand, no pronounced and systematic dependence of the effect on solar activity has been revealed; on the other hand, analyzing the thresholds distribution over 12 months, the number of cases may turn out to be statistically insufficient in some bins for a particular month if solar activity gradation is applied.

Figure 1

[5]  The results of our analysis over 14 stations are given in Table 2. A well-pronounced seasonal (winter/summer) variation is seen for all the stations. The threshold is low in winter and much higher in summer. There is also a pronounced tendency for complementary minima to appear during equinoctial periods. To present these annual variations in a more explicit way, monthly deviations of the thresholds relative to the annual mean values were found for each station and the results are given in Figure 1.

[6]  This result implies that the level of geomagnetic disturbances should be higher in summer than in winter to result in the same F2 -layer negative storm effect. A decrease in the threshold during equinoxes is also an interesting result of our analysis. Let us consider possible explanations for the variations revealed.