Thermospheric density oscillations due to periodic solar wind high-speed streams
Abstract
[1] We report on periodic oscillations in thermosphere density, measured by the accelerometer on the CHAMP satellite during 2006, and relate these periodicities to oscillations observed in solar wind speed and Kp index. Common periodic oscillations at 4–5, 6–7, and 9–11 day periods are observed in the neutral density at 400 km in the 2006 data set, with the 7 day period being the predominant oscillation. Spectral analysis reveals that similar periodicities are present in both the solar wind and the planetary magnetic index Kp but not in the EUV solar flux proxy F10.7. We suggest that the periodic oscillations observed in thermosphere density are a direct response to recurrent geomagnetic activity and associated high-speed streams in the solar wind. The lack of response in F10.7 at the 7 day period enables storm effects on the thermosphere density to be isolated from solar flux effects. The Kp index for these events correspond to moderate levels of geomagnetic activity, and the resultant perturbations in thermosphere density are ±20–30% of background levels. Although these levels of perturbation are small compared to major magnetic storms, their much higher occurrence frequency and characteristic long recovery time may lead to a cumulative effect on the state of the thermosphere and ionosphere.
1. Introduction
[2] The Sun is the ultimate source for driving perturbations in thermosphere density; however, the type of energy and the complex pathways in transferring energy to the thermosphere remains a grand challenge in studying the geospace system. Periodic behavior in thermosphere neutral density can be used as a method to identify and isolate energy sources and can enable studies of their cause and effect. The 27 day solar rotation and 11-year solar cycle are characteristic periodicities in EUV flux that have been clearly correlated to changes in thermosphere density. Solar flares can appreciably enhance the EUV flux from the Sun and produce a significant thermosphere density response [Sutton et al., 2006] but are generally not repeatable events that could be considered periodic over multiple solar rotations. Thus, the magnitude of the EUV effect is variable, but the 27 day periodicity is repeatable and identifiable in the thermosphere density response.
[3] Meteorological influences from the lower atmosphere can propagate to the thermosphere and create periodic density perturbations. Atmospheric tides generate thermosphere density disturbances at harmonics of the 24-h rotation period of Earth, while planetary waves at 2, 5, 10, and 16 day oscillations are known to populate the lower thermosphere. The existence of oscillations in Earth's ionosphere at periods of 2, 5, 6–7, 9–10, and 12–18 days is well established [e.g., Altadill et al., 2003; Altadill and Apostolov, 2003; Forbes and Leveroni, 1992; Forbes et al., 1997; Laštovièka et al., 2003]. These are referred to as oscillations at “planetary wave (PW) periods” since they correspond to waves in the middle atmosphere that are characteristic of normal modes or resonant oscillations in the case of 2, 5, 9–10, and 12–18 day periods [Salby, 1984] or of maximum growth rates of unstable regions [e.g., Meyer and Forbes, 1997; Liu et al., 2004] and/or wave-wave interactions [Pogoreltsev et al., 2002] in the case of the 6–7 day wave.
[4] Solar disturbances that propagate through the interplanetary medium and lead to geomagnetic storms can be quite variable; however, natural periodicities of solar disturbances have been demonstrated and correlated to geomagnetic storm indices [e.g., Crooker et al., 1977; Musman and Altrock, 1978; Crooker and Cliver, 1994; Tsurutani et al., 2006]. Recurrent geomagnetic disturbances on a 27 day period have been associated with corotating interaction regions (CIRs) within the solar wind [e.g., Tsurutani et al., 1995]. CIRs occur in response to the interaction of fast and slow solar wind speeds and are related to the high-speed solar wind arising from coronal holes. Nonrecurrent geomagnetic disturbances have been associated with coronal mass ejections (CMEs) where solar material is catastrophically ejected from the Sun. These events are most numerous during solar maximum and typically do not persist beyond a solar rotation. CMEs may augment CIRs [Crooker and Cliver, 1994], but the recurring periodicity in CIRs enables a correlation analysis with periodicities in geomagnetic activity and changes in the thermosphere density.
[5] In this paper we report periodic oscillations at 4–5, 6–7, and 9–11 day periods in the thermosphere neutral density at 400 km from the 2006 CHAMP accelerometer data set. This work complements the discovery by Lei et al. [2008] of a 9 day oscillation in the CHAMP 2005 thermosphere density. We demonstrate similar variations in the geomagnetic index, Kp, and in the solar wind speed measured at L1 with the near 7 day periodicity being the predominant oscillation in the 2006 data sets. Periodic CIRs and high-speed solar wind streams modulate energy transfer into the magnetosphere-ionosphere-thermosphere system [Tsurutani et al., 2006], and we hypothesize that this energy is manifested in the form of joule heating rates that drive the observed density variations.
2. Data Analysis and Results
[6] In this study we use the total mass densities inferred from the accelerometer measurements on the CHAMP satellite, which was launched into a near-circular orbit with an inclination of 87.3° on 15 July 2000 [Reigber et al., 2000]. CHAMP provides a density measurement every 80 km (10 s) along the orbit. The complete thermosphere density derivation procedure applied to these data is given by Sutton et al. [2005, 2007], and the measured neutral densities at satellite altitudes are normalized to a constant altitude of 400 km using the NRLMSISE-00 empirical model [Picone et al., 2002]. The CHAMP data from 1 January to 31 December 2006 are used in this study.
[7] In Figure 1, the orbit-averaged CHAMP thermosphere densities for 2006 are compared with the corresponding variations of EUV solar flux proxy, F10.7, solar wind speed measured by the ACE satellite, and planetary magnetic activity index, Kp. The 27 day periodicity is observed in both the ascending orbit–averaged (Figure 1d) and descending orbit–averaged (Figure 1e) thermosphere density. Higher-frequency peaks can also be seen within each solar rotation period (around 27 days) throughout the year of 2006.
[8] Figure 1a is a plot of the daily values of F10.7 and illustrates a regular periodicity of about 27 days associated with the average rotation rate of the Sun. As F10.7 is a proxy for EUV radiative flux from the Sun, the thermosphere will absorb this radiation and respond by increasing thermosphere density. Obviously, the higher-frequency periodic oscillations observed in thermosphere density are not seen in the F10.7 values.
[9] The variations of solar wind speed and the planetary 3-h magnetic activity index Kp, shown in Figures 1b and 1c, respectively, illustrate that they both contain periodic oscillations with periods of less than 27 days and that these shorter-period oscillations are modulated by 27 day solar flux oscillations. Moreover, a one-to-one correspondence of the periodic oscillations among thermosphere density, solar wind speed, and Kp index can clearly be seen over the entire year of 2006. Their similarities with respect to periodicity are further confirmed by the Lomb-Scargle analysis [Lomb, 1976; Scargle, 1982] in Figure 2, where spectral amplitudes are plotted against period.
[10] As shown in Figure 2, the spectrum indicates the presence of multiple peaks above the 95% significance level at periods, in order of highest to lowest amplitude, of 6–7, 9, 11, and 4–5 days. Note that the spectral amplitude at 6–7 days is most distinct during this 2006 time period. The solar wind speed and Kp index have very similar spectra to that observed in thermosphere density. The periodogram results for the F10.7 values indicate that the periodic oscillations in thermosphere density are not introduced by variations in the solar EUV flux.
[11] In order to examine the timing and evolution of the periodic oscillation events, a wavelet spectral analysis [Torrence and Compo, 1998] was applied to the time series presented in Figure 1 (i.e., time series in F10.7, solar wind speed, Kp index, and thermosphere density). Note that oscillations in thermosphere density with periods longer than 32 days are removed first to exclude long-term variations associated with seasonal, semiannual, and local time sampling effects (Figure 1). Density residuals from the low-frequency variations (<1/32 d−1) of neutral density are then subjected to Morlet wavelet analysis. The corresponding Morlet wavelet power spectra for periods from 2 to 20 days are shown in Figure 3 as a function of day of the year. The white solid contours indicate a significance level higher than 95% in the spectral power, and the periods of 5, 7, 9, and 13 days are marked with horizontal white dashed lines for better visualization.
[12] The most striking feature in Figure 3 is the predominant oscillation, with a period near 7 days, in both the ascending and descending thermosphere densities from days 90 to 140 and days 250 to 360. The 9–11 day oscillations appear within five time intervals: from days 20 to 40, days 70 to 120, days 150 to 170, days 210 to 250, and days 280 to 360. In addition, the 4–5 day oscillation can be seen clearly from days 80 to 110 and days 340 to 360. The time evolution of the power in neutral density perturbations matches well those in both solar wind speed and Kp index. Finally, the wavelet spectrum of F10.7 shows again that the short-period oscillations observed in neutral density are not present in F10.7 any time during the year, ruling out solar flux variations.
[13] To see the latitudinal structure in the response of neutral density to Kp index, Figure 4a shows the variations of neutral density at 80 km intervals along the satellite ascending orbit from pole to pole for days 260–360 of 2006 with the 3-h Kp index superimposed on the plot. The 6–7 day periodicity is dominant during this 100 day interval (see Figure 3), and the correlation between Kp and the thermosphere density is also clearly recognized. The thermosphere density response due to geomagnetic disturbances at this frequency is clearly observed at all latitudes, with some indication of greater change at high latitudes than at low latitudes. Note that the values of the Kp index generally do not exceed 4, indicating that these density oscillations are associated with periods of moderate to low geomagnetic activity.
[14] Percent residuals of the ascending orbit–averaged density after band-pass filtering the data to the 9 day running mean are shown in Figure 4b (for the same time interval as in Figure 4a). The band-pass filter was centered at the period of 7 days, with half-power points at 5 and 9 days. The filtered Kp index is plotted as the dashed line in Figure 4b. The 7 day oscillation in thermosphere density during this period can reach ±20–30% of background levels. As can be seen in Figure 4b, the percent change in neutral density follows the perturbations in Kp extremely well, with a linear correlation coefficient of r = 0.92. The phase lag between density changes and Kp perturbations corresponds to the response time of neutral density to geomagnetic forcing.
3. Discussions and Conclusions
[15] The spectral analysis (Figures 2 and 3) performed on the 2006 CHAMP density data at an altitude of 400 km indicates that oscillations in thermosphere density with periods of 4–5, 6–7, and 9–11 days are present, recur throughout the year, and are not associated with variability in the EUV solar flux. Furthermore, these periodicities show strong correlation with periodicities in the solar wind speed and the Kp index. Investigations of periodicities in high-speed solar wind streams have revealed oscillations at periods of 27, 13.5, 9, 7, and 5.5 days [e.g., Prabhakaran Nayar et al., 2001; Temmer et al., 2007; Verma and Joshi, 1994; Verma, 2001]. Recently, Temmer et al. [2007] related the periodic modulation of solar wind speed in a straightforward manner to the periodic occurrence of coronal holes, illustrating a predominant 9 day period in solar wind data during 2005. The occurrence of a given periodicity in solar wind speed at any specific time near Earth is related to the surface distribution of coronal holes on the Sun and their extension to the ecliptic plane. Lei et al. [2008] explore this connection in describing thermosphere density oscillations observed with a 9 day periodicity in 2005. Mlynczak et al. [2008] also report a connection between the longitudinal distribution of coronal holes and a 9 day periodicity observed in the thermosphere infrared emissions.
[16] CIRs lead high-speed solar wind streams and are the manifestation of high-speed streams interacting with the slower, upstream solar wind. The magnetic fields within the CIR can be quite variable, producing highly irregular geomagnetic storm main phases. In the peak and trailing portions of the solar wind high-speed stream exist Alfvénic fluctuations in the interplanetary magnetic field Bz component that have been related to high-intensity long-duration continuous AE activity [Tsurutani and Gonzalez, 1987; Tsurutani et al., 1995]. Tsurutani et al. [2006] provide a review of CIR-generated storm behavior, and Borovsky and Denton [2006] and Denton et al. [2006] contrast the CIR-driven storms with nonrecurrent CME-driven storms. Applicable to our study are the CIR and high-speed stream storm characteristics: recurrent behavior at periods of 27 days and shorter, moderate levels of geomagnetic activity, very long (days to weeks) geomagnetic storm recovery phase, and preferential occurrence in solar minimum and during the declining phase of solar maximum. These characteristics all reflect the conditions under which the CHAMP 2006 density data are being studied. The 2006 time period corresponds to solar minimum conditions. The thermosphere density has been shown to respond, at all latitudes, with periodicities that correlate well with oscillations in solar wind speed and Kp index (Figures 3 and 4). This behavior strongly suggests that the observed periodicities in thermosphere density are produced by recurrent geomagnetic activity caused by CIR and high-speed stream–Alfvénic wave structure in the solar wind. The geomagnetic activity leads to heating of the thermosphere, through currents and particle precipitation, that results in a global increase in neutral density at a fixed altitude of 400 km. The long recovery phase of high-speed-stream-driven geomagnetic activity of days to weeks can lead to overlapping geomagnetic conditions given the observed 5, 7, and 9 day periodicities observed in the solar wind data for 2006.
[17] Planetary waves propagating from the lower atmosphere are an alternative source of periodic variability in thermosphere density. A key problem is that planetary waves are not capable of propagating above about 100–110 km [Pogoreltsev et al., 2007]. Yet, F region ionospheric observations have demonstrated periodic oscillations that have been related to planetary wave periods. Ionosphere studies that reveal PW periodicities often utilize coincident measurements of PW activity at the same wave periods below 100 km as evidence that the two phenomena are connected [e.g., Borries et al., 2007]. Although several mechanisms have been proposed (see review by Forbes [1996]), the manner in which planetary waves may drive ionospheric variations still remains unknown, and therefore more indirect scenarios, such as modulation of dynamo-generated electric fields that map into the F region, appear to be required. The mechanism to generate thermospheric density perturbations at 400 km altitude due to recurrent geomagnetic forcing is much more direct than the complex pathways required with planetary wave propagation from the lower atmosphere.
[18] Hence our current work, and that of Lei et al. [2008] and Mlynczak et al. [2008], reveals a new solar-terrestrial connection between thermospheric density oscillations and periodic high-speed solar wind streams that may be rooted, at least in part, in rotating coronal holes. There are three additional and important implications. First, the periodic nature of the connection between coronal hole, solar wind speed, geomagnetic activity, and thermosphere density suggests some element of predictability. Moreover, the periodic behavior enables thermospheric density response to geomagnetic effects to be separated from its response to varying solar EUV flux. Second, the density variations reported here and the accompanying wind perturbations will modulate F region ionospheric loss rates and vertical transport, and therefore plasma densities (however, the degree of this modulation remains to be quantitatively established). This more direct modulation of the F region ionosphere may explain some of the planetary wave–like oscillations observed in ionospheric data sets. Although, features such as nonzonal structure in these ionospheric data sets require a more complete analysis of ionosphere response to recurrent geomagnetic activity coupled with the associated thermospheric response. Third, although the geomagnetic activity levels associated with CIRs and high-speed streams are moderate, their recurrent nature on periods of 9, 7, and 5 days may serve to cumulatively impact the state of the thermosphere and ionosphere as the recovery phase for these events can be many days to weeks [Tsurutani et al., 1995]. The CHAMP density data therefore represent an important resource for quantitatively isolating a driver of thermosphere-ionosphere disturbances.
Acknowledgments
[19] This work was supported by the AFOSR MURI award FA9550-07-1-0565 and by grant ATM-0719480 from the National Science Foundation as part of the Space Weather Program. The Kp and F10.7 indices were downloaded from NGDC database, and the ACE solar wind data were obtained from the GSFC/SPDF OMNIWeb interface at http://omniweb.gsfc.nasa.gov.
[20] Wolfgang Baumjohann thanks Bruce Tsurutani and Alexander Pogoreltsev for their assistance in evaluating this paper.
