The great magnetic storm of April 6–7, 2000 generated ionospheric disturbances in the west Pacific region. Two ionosondes at Wuhan (30.6°N, 114.4°E) and Chung-Li (24.9°N, 121°E) observed the ionosphere during this period. The variations of the ionospheric parameters, NmF2 (plasma density of the F-peak), hmF2 (height of the F-peak) and h′F (minimum virtual height of the F-layer), show that a traveling atmospheric disturbance (TAD) affects the ionosphere at middle and low latitudes in this region. The propagation velocities deduced from the time delay of hmF2 and h′F recorded at the two stations are about 655 and 164 m/s, respectively. Furthermore, a method of deriving the vertical phase and group velocities is applied to the sequences of virtual heights at fixed sounding frequencies. The vertical and associated meridional velocities demonstrate that the upward motions of the ionospheric plasma are caused by a TAD.
 In the upper atmosphere, charged particles are minor constituents and firmly embedded in the neutral gas environment. Any perturbation of the neutral atmosphere will immediately entail corresponding perturbations of the ionosphere [Prölss, 1997]. During magnetic storms, heatings at high latitudes cause expansion of the neutral atmosphere. When the heating events are impulsive, they may drive traveling atmospheric disturbances (TADs) generating short-duration positive storm effects at lower latitudes [Prölss and Jung, 1978; Burns and Killeen, 1992; Fuller-Rowell et al., 1994; Bauske and Prölss, 1997; Prölss, 1997]. Prölss , Prölss , and Bauske and Prölss  show that a TAD is an impulse-like perturbation formed by a superposition of gravity waves, which propagates with high velocity (e.g. 600 m/s) in meridional direction and carries a meridional wind perturbation (e.g. 150 m/s) at middle latitudes.
 In this work, we are interested in TADs that propagate along equatorward direction in the West Pacific region. To further investigate their TAD effects, observations from two ionosondes in this region were analyzed for the April 6–7, 2000 magnetic storm. The ionograms were recorded simultaneously by the Digisonde at Wuhan (30.6°N, 114.4°E, dip angle of magnetic field: 43.8°) and the ionosonde IPS-42 at Chung-Li (24.9°N, 121°E, dip angle of magnetic field: 34.1°) with a time resolution of 15 minutes. Measurements of the F-peak plasma frequency foF2 (or plasma density NmF2), F-peak height hmF2, and the F-layer minimum virtual height h′F are used to document the ionospheric response to the storm. The meridional and vertical velocities deduced from the ionograms recorded by the two ionosondes during this storm time are examined and discussed.
2. Data Analyses and Interpretations
 With a sudden storm commencement (SSC) onset at 1640 UT on April 6, a magnetic storm occurs during April 6–7, 2000. The minimum Dst of less than −300 nT appearing at about 0000 UT on April 7 indicates a large magnetic storm (Figure 1). Figures 2a–2c illustrates that after the SSC, the hmF2 (h′F) at Wuhan starts to rise by about 150 km at about 1830 UT (1800 UT) on April 6. During the uplift phase, the NmF2 first decreases, then increases to reach a maximum at 0030 UT on April 7, and again decreases during the rest period. Note that the solid lines represent the observed values, while the dashed lines are the median values of April 2000. Figures 2d–2f displays that at Chung-Li, the sudden increase by about 200 km in hmF2 (h′F) begins at about 1845 UT (1900 UT) on April 6. During the uplift phase the NmF2 decreases rapidly then recovers to reach its maximum at about 0000 UT on April 7, and then decreases during the rest period. After the sudden increases in NmF2 at about 0000–0100 UT on April 7, the negative storm effects appear in the variations of NmF2 at the two stations. These behaviors at the two stations agree with the TAD effects suggested by Prölss  and Bauske and Prölss , which an uplifting of the F2 layer (ΔhmF2) can leads to a positive ionospheric storm effect (ΔNmF2). Note that at about 1800 UT on 6 April the true heights of the sudden hmF2 increase (the virtual altitudes of the sudden h′F increase) at Wuhan and Chung-Li are 300 and 290 km (250 and 200 km), respectively (see Figure 2). A meridional phase velocity VM (Vm) of about 655 ± 320 m/s (164 ± 40 m/s) can be deduced from the time lag of 15 minutes (about 1 hour) between the onset times in hmF2 (h′F) variations of the two ionosondes latitudinally separated by about 590 km (see Figure 2).
 For further study, the vertical phase (Vp) and group (Vg) velocities at the two stations are determined from the sequences of ionograms recorded at the two stations by applying the method developed by Liu et al. . To extract the virtual height variations as function of time, for a given plasma frequency, from a sequence of ionograms, this function is expanded into a Fourier series:
where i = 1…N measurements. The largest Fourier coefficients reveal the spectral component dominating the height fluctuations. A TAD can be represented as a superposition of plane waves
where kj, lj, and nj are the wave vector components of the wave with frequency Φj/2π = j/NΔt (Δt = 15 minutes) of the jth harmonic. The phase function
is a function of z only, since for vertical ionospheric sounding [Liu et al., 1998]
Differentiating Φj with respect to z gives the vertical component, nj:
The vertical phase and group velocities can then be calculated
for the strongest wave component σj in (2). At Wuhan the virtual heights at 3.3 ∼ 5.2 MHz from 1430 to 2230 UT (Figure 3a) are used to estimate the vertical velocities, while at Chung-Li the virtual heights at 2.6 ∼ 4.5 MHz from 1600 to 2400 UT (Figure 3b) are employed to calculate the velocities. Note that the interval of the sounding frequency in Figure 3 is 0.1 MHz. The reflection heights for the selected sounding frequencies are calculated by true height analysis programs POLAN [Titheridge, 1985] and ARTIST [Reinisch and Haung, 1983; Reinisch, 1996]. Tables 1 and 2 list the Vp and Vg of a 4-hour period (j = 2; the 2nd harmonic) wave simultaneously evaluated at Wuhan and Chung-Li at various heights. Our results show upward Vp and downward Vg at each altitude.
Table 1. The 2nd Harmonic Vertical Velocities Vp and Vg, and the Meridional Velocities Vpm and Vgm at Wuhana
Note that The positive sign denotes the upward and equatorward directions.
Table 2. The 2nd Harmonic Vertical Velocities Vp and Vg, and the Meridional Velocities Vpm and Vgm at Chung-Lia
Note that The positive sign denotes the upward and equatorward directions.
 In order to examine the vertical effects of a TAD, the Vp and Vg are projected to the meridional direction by the equation:
where Vpm and Vgm are the meridional velocities projected from Vp and Vg, respectively; I is the dip angle of the magnetic field. The Vpm and Vgm associated with the 2nd harmonic at the two stations are also shown in Tables 1 and 2. Since the heights in the tables are mean true heights of the time series obtained for associated fixed sounding frequencies, the discrepancy in height gradients of two successive frequencies between two stations could be due to different latitudes [Davies, 1990]. It can be seen that the phase velocities Vp and Vpm of two stations are upward and equatorward direction at each height. It is interesting to find that the projected Vpm (146–178 m/s) at about 229 and 201 km altitudes ranges are close to the Vm (160 m/s) directly deduced from the time lag between the two stations h′Fs. The coincidence of the Vm and Vpm suggests that the meridional TAD possibly result in waves in plasma propagating along the magnetic field lines. It is also interesting to note that both group velocities Vg and Vgm are downward and poleward, respectively.
3. Discussion and Conclusion
Prölss  and Bauske and Prölss  concluded that certain positive ionospheric storm effects are caused by TADs, which are pulse-like superposition of atmospheric gravity waves (AGWs). In this work, the VM (about 655 m/s) is deduced from the time lag between the onset times in hmF2 of the two ionosondes. Note that the onset time is when the hmF2 starts to rise, which means that a TAD begins to affect the local ionosphere. This VM is consistent with the propagating velocity (about 600s m/s) suggested by Prölss [1993, 1995]. Therefore, this VM could be attributed the propagating velocity of a TAD from high to low latitudes during the April 6–7 storm.
 Meanwhile, the Vm derived from time lag of the h′F observed from the two ionosondes is about 160 m/s. Although the two projected Vpm well agree with the Vm (about 160s m/s) at the lower ionosphere, we find that the TAD velocities VM and Vm derived in this paper are very different. One of possible causes is that the VM and Vm are derived from two different altitudes, which are 300–500 km and 150–200 km (the later one reduced due to the virtual altitudes), respectively. Prölss  and Bauske and Prölss  showed that the TADs drag the ionization up the inclined magnetic field lines, causing a transient increase in layer height and an increase of the ionization density. Note that it needs a certain amount of collisions between the neutral and plasma to activate the air drag. The collision frequency is a function of height [e.g. Davies, 1990], which could lead to different response times of the F region at hmF2 and at h′F.
 Another possible reason for the discrepancy between VM and Vm is the hmF2 being used. For simulations studies, it is very convenient and easy to locate the hmF2. However, it has been well known by ionospheric scientists that the ionospheric F2 region is both photochemical and transport dominated [Davies, 1990]. Even without any vertical motion, a slight shape change in the F2 region could significantly affect estimations of the hmF2. The NmF2 changes in the two stations could directly or indirectly affect the associated hmF2 observations.
 Many researches estimated horizontal or vertical (phase) velocities by means of the phase changes vs. distances or altitudes during storm periods [e.g. Hajkowicz, 1990 and 1991; Ho et al., 1998]. This paper reports the first result of the vertical group velocity (energy propagation) in the ionosphere during the storm period. The result shows that the vertical phase and group velocities of the 4-hour wave are respectively 45–135 m/s upward and 3–9 m/s downward during the 6–7 April 2000 magnetic storm. A rough estimation for the vertical velocities often employed by the previous scientists is to directly evaluate the vertical apparent (phase) velocity Va from the changes in h′F (or hmF2) in a certain time interval. Figures 4a and 4b depict at about 1800 UT on 6 April that the two h′Fs suddenly rise by about 200–300 km within 1–2 hours. The estimated Va is between 50s–100s m/s in the upward direction, which generally is close to the Vp (45–135 m/s upward).
Tables 1 and 2 illustrate that due to the TAD passage and the associated uplift of the ionosphere the phase velocities Vpm and Vp are in equatorward and upward direction, respectively. It is interesting to find the Vgm and Vg being opposite to the associated Vpm and Vp. On the other hand, the poleward direction of Vgm and downward direction of Vg at each height indicate that the energy transportations of the waves are from low to high latitude, and from high to low altitude, respectively. Although the causality of the poleward Vgm and downward Vg is not fully understood, the directions of wave propagation are consistent with the AGWs theory prediction that the phase progression in the direction is opposite to the energy transportation [Kuo et al., 1993].
 In this paper, in addition to three typical ionospheric parameters foF2 (or NmF2), hmF2, and h′F, a sequence of ionograms are employed to study the wave propagations during the April 6–7 ionospheric storm in the West Pacific region. The current observations agree with the previous simulation result that during the period of few hours of a TAD front passage, the hmF2 has a pulse-like motion, and the NmF2 almost simultaneously decrease but increase right after that. The different velocities observed at the peak and bottom of the F2 region suggest that the response to the TAD in the higher and lower ionosphere is different. The lower F region is more affected by air drag because the neutral-ion collision rates are higher. The time series of ionogram studies show that the energy transportation is directed opposite to the TAD phase progression. In conclusion, the similar passage features observed at Wuhan and Chung-Li during 1800 UT April 6 and 0200 UT April 7 (Figure 2) indicate that the TAD plays an important role during the magnetic storm.
 The research has been supported by the grant of National Science Council NSC90-2111-M-008-062-AP3. The authors would like to thank the referees for valuable suggestions that greatly improved the presentation of this paper, and to express appreciation to the World Data Center for Geomagnetism in Kyoto for providing Dst data.