4. Discussion

[8]  Let us consider ionospheric data and the character of radio signals passing through the path. Figures 1-5 demonstrate data of the same type for different days of the analyzed storm from 14 May 1997 to 18 May 1997.

[9]  1. Figures 1a-5a give variations of the D f_oF2 values of the vertical ionospheric sounding (VIS) of the St. Petersburg station during the magnetic storm. Here the D f_oF2 value is the deviation of the critical frequency of the ionospheric F2 layer from the monthly median. Data of D f_oF2 characterize an ionosphere state at the receiving station of the radio path. Figure 1a demonstrates small variations (both negative and positive) of D f_oF2 during the quiet period and significant deviations of D f_oF2 during the disturbed interval (1600-2400 UT). So, before the bay-like disturbance with AE _max=500 nT at 1800 UT (Figure 1e) the D f_oF2 values increase from 1400 to 1800 UT. After the end of this disturbance there is an increase of D f_oF2 values from 1800 to 2300 UT. Figure 2a presents the extremely disturbed day of 15 May 1997. Variations of D f_oF2 have negative character and their intensities grow as the storm develops. After the substorm, negative values of D f_oF2 are being retained during two days: 16 May and 17 May 1997 (Figures 3a and 4a). A gradual approximation of D f_oF2 variations to zero occurs only by 18 May 1997 (Figure 5a). It should be noted, that during the growing of the disturbance a part of the VIS data begins to disappear because of absorption, scattering, screening, etc. Thus the number of St. Petersburg's ionograms with information for 14 May (a nearly quiet day) is i=22 from possible 24, for the extremely disturbed day of 15 May, i=13, for 16 May i=16, for 17 May i=13 and for the slightly disturbed day of 18 May i=18.

[10]  2. Figures 1b-5b illustrate the VIS data of Sodankyla observatory. It is located not far from the reflection point of the path Heiss Island-St. Petersburg and characterizes the ionospheric state near the reflection area. The VIS data by Murmansk station would be more suitable, but they are absent. Figures 1b-5b show that for the more high-latitude Sodankyla station there are ionograms with information less then for the St. Petersburg station. During the disturbances they are quite few in number.

[11]  3. Figures 1c-5c demonstrate variations of the frequency range MOF-LOF on the studied path. Let us consider these variations for every day from 14 to 18 May in more detail.

[12]  Figure 1c shows the MOF pattern during the period 0000-2400 UT on 14 May 1997 which repeat the D f_oF2 variations by Sodankyla station (Figure 1b), that is, in the area close to the reflection point of the path. These pattern during the interval 2000-2400 UT correlate with the D f_oF2 variations by St. Petersburg station (Figure 1a). The LOF values for quiet period from 0000 to 1600 UT are practically constant as well as the absorption A < 0.5 dB (Figure 1d). The first splash of a disturbance by the AE index from 1700 to 1900 UT causes a sharp increase of the absorption A=1.6 dB and therefore a growth of the LOF values. The frequency range MOF-LOF on the falling off of the storm from 1800 to 2000 UT becomes narrow sharply.

[13]  The extremely disturbed day on 15 May 1997 (Figure 2c) is characterized by considerable variations of the MOF and the LOF. About 3.5 hours before the moment T_o=0630 UT (the moment of a sharp increase of the AE index) there is a splash of the MOF values from 0300 to 0630 UT. A similar small splash of D f_oF2 during the same period takes place at St. Petersburg station. The disturbance grows sharply from 0700 to 1000 UT, therefore the MOF data are absent because of the extremely intensive absorption ( A=3-6 dB) at Sodankyla observatory (Figure 2d). On the falling off of the storm from 1300 to 2100 UT there is again a growth of the MOF values. A small hump of the MOF at 2200 UT is caused by a substorm at 2200 UT too (Figures 2e and 2d).

[14]  The second specific peculiarity of this day is that MOF and LOF correlate well. Increase of the MOF and the LOF from 0300 to 0700 UT and from 1200 to 2100 UT is caused by particle flows passing through the ionosphere in the area of the path's reflection point on the border of diffuse precipitation. This particle precipitation must increase the f_oF2 and f_oE_s values. For example, a calculation of the f_oF2 in the reflection area of the path by means of the well known "transfer curves" at 0400 UT of 15 May gives f_oF2=6.5 MHz. Whereas at 0400 UT of 14 May (the quiet period) the calculation gives f_oF2=4.1 MHz. Ionization occurs at both the upper and lower ionosphere [Rostoker et al., 1980]. In the F2 layer there is a growth of D f_oF2 and accordingly the MOF but in the lower ionosphere a growth of ionization causes the increase of absorption and therefore the LOF. As it was said above, a growth of ionization in the lower ionosphere can evoke an increase of the MOF due to intensive sporadic layers in the E region of the ionosphere. While the absorption level is extremely high from 0700 to 1000 UT (Figure 2d), the LOF and the MOF are absent (Figure 2c). The frequency range MOF-LOF on the falling off of the disturbances from 1000 to 1200 UT and from 2000 to 2100 UT is the narrowest.

[15]  The last but one disturbed day on 16 May 1997 follows the strong storm. From 0000 to 1100 UT any disturbance is absent (Figure 3e), significant variations of the MOF and the LOF do not occur for this period. Further, from 1200 to 1800 UT an intensive substorm happens which causes variations of the MOF and the LOF, an increase of these parameters during the growth and recovery phases of the substorm. From 1500 to 1800 UT the frequency range MOF-LOF is the narrowest. At midnight an intensive substorm with AE _max=900 nT (Figure 3e) develops. About 1-3 hours before T_o=2300 UT, values of D f_oF2 (Figure 3a) and the MOF (Figure 3c) grow significantly.

[16]  The last disturbed day on 17 May 1997 begins with an intensive substorm from 0000 to 0500 UT. There are a growth of absorption (Figure 4d) and a sharp narrowing of the frequency range MOF-LOF from 0100 to 0800 UT (Figure 4c). Similar events of narrowing the frequency range MOF-LOF occur during the periods 1200-1300 UT and 1900-2200 UT on the falling off of the small disturbances during periods 1100-1300 UT and 1700-1900 UT.

[17]  The slightly disturbed day on 18 May 1997 is characterized by smooth patterns of the MOF and the LOF which are the same as for the quiet period from 0000 to 1500 UT on 14 May 1997. The MOF and the LOF data from 1500 to 2400 UT in Figure 5c as well as the B_z values in Figure 5f are absent because of some technical reasons.