Fine structure of sporadic sodium layer observed with a sodium lidar at Tromsø, Norway



[1] We report a sporadic sodium layer (SSL), in particular its fine structure, observed at 92–98 km between 20:00 and 23:30 UT (21:00–24:30 LT) on 11 January 2011 using a sodium lidar, which was installed in the European incoherent scatter (EISCAT) radar site at Tromsø, Norway (69.6°N, 19.2°E) in early 2010. The sodium lidar measurement with 5-sec time-resolution reveals the details of dramatic sodium-density increase as well as short-period wavelike structure in the SSL. The rate of increase of height-integrated sodium density at the beginning of the SSL event was 6.4–9.6 × 1010 m−2 s−1. Dominant oscillation periods in the wavelike structures were 7–11 min at 95–98 km and 3 min at 92–95 km. The calculated power spectral densities are well represented by power laws, implying the presence of the short-period waves and turbulence in the frequency range of 10−4–10−1 Hz.

1. Introduction

[2] The mesospheric and lower thermospheric sodium layer, distributed at heights around 80–110 km height, has been observed for more than 30 years by resonance scattering lidars. During these observations, researchers discovered the sudden formation of a dense thin sodium layer superposed on the normal sodium layer [e.g., Clemesha, 1995; Mathews, 1998, and references therein]. Such an enhanced layer is called a sporadic or sudden sodium layer (SSL). Typical features of the SSL are a thin layer with a full-width at half maximum (FWHM) of 0.1–2 km lasting for a few tens of minutes to several hours, and its peak sodium density a few to tens times larger than that of the background sodium density. Despite much study of SSLs, their cause is still an open question. A most likely mechanism producing sodium atoms would be the ion neutralization induced by a sporadic E (Es) descending to 90–95 km height [e.g., Cox and Plane,1998; Heinselman et al., 1998; Heinselman, 2000].

[3] Structures of the SSL have been used as tracers of the dynamic features in the upper atmosphere [e.g., Hansen and von Zahn, 1990; Pfrommer et al., 2009]. The characteristics of the SSL, the thin and dense sodium layer, are very suitable for investigating in particular fine structures of the atmospheric features, such as small scale waves. For example, at Andenes, Norway (69°N), Hansen and von Zahn [1990] analyzed sodium density data with time-resolution of 1-min, and demonstrated upward and downward movements of the SSL height with time scale of ∼20-min. They suggested that such movements are signatures of atmospheric gravity waves. A typical time-resolution of the previous sodium density measurements had been order of minutes; the time scale of the atmospheric features we can retrieve can be a few tens of minutes or longer.

[4] Recently high time-resolution measurements of the sodium density (40 msec to 1 sec) were made near Vancouver, Canada (49°N), and the atmospheric features with shorter time scale, such as the ∼4-min period oscillations in the sodium density, were found in the denser sodium clouds [Pfrommer et al., 2009]. They mentioned that the periodic oscillations of 4-min would be due to the atmospheric gravity waves. To the best of our knowledge, there is no publication made of time-resolution less than 1-min except for their result at mid-latitude region, making these short time scale phenomenon in the atmospheric dynamics poorly known.

[5] A new sodium lidar was installed in the European incoherent scatter (EISCAT) radar site at Tromsø, Norway (69.6°N, 19.2°E) in early 2010, and the first successful measurement of temperature and sodium density was performed on 1 October 2010. One of the scientific objectives of this sodium lidar is investigation of atmospheric features in shorter time-scale at high-latitude region. In this paper, we present an initial report of a SSL observed on 11 January 2011. In particular, fine structures of the SSL are investigated using data taken at a high time-resolution of 5-sec.

2. Tromsø Sodium Lidar

[6] Here we briefly introduce the new sodium lidar system at Tromsø, Norway. An all-solid-state Q-switched single-frequency light source tuned to the sodium D2-a line at 589.1583 nm is developed for the lidar laser. The light source is based on sum-frequency mixing two injection-locked Nd:YAG lasers oscillating at 1064 nm and 1319 nm, respectively, with a LiB3O5 crystal, which is used under 90° phase-matching condition at a temperature of 39.5°C. Performance of the laser is currently an average output power of ∼2 W at a repetition rate of 1 kHz. The FWHM of the laser pulse is ∼35 nsec, and the divergence of the laser beam is less than 1 mrad. The laser frequency is calibrated using the Doppler-free saturation fluorescence spectra of sodium atoms. An acousto-optic frequency shifter system is applied for the high-speed laser frequency switching. The lidar receiver mainly consists of a 355-mm diameter Schmidt-Cassegrain telescope and a high-speed photomultiplier tube of quantum efficiency of ∼40%. Pulses from the photomultiplier are amplified and then detected by a high-speed multi-channel scaler.

[7] A basic observational mode is the two-frequency mode, which is used to derive the temperature and sodium density data [She et al., 1990]. On 11 January 2011, we made the two-frequency mode observation; the selected laser frequencies were −651.4 MHz (i.e. the D2-a peak frequency) and −151.4 MHz when the central wavelength is 589.15826 nm. In the case of measurement with the −151.4-MHz laser frequency, the sodium emission intensity is fairly insensitive to the temperature (∼6% change in the sodium absorption cross section per 100 K change in the temperature). Therefore the data taken by the −151.4-MHz laser frequency are used for determination of the sodium density variation. Individual soundings at 1-kHz are integrated for 5 seconds to give a single height profile of the sodium density with height-resolution of 96 m. These profiles are taken at one laser frequency 12 times, for a total duration of one minute. The system then switches laser frequency and repeats the process for another minute. This cycle is then repeated indefinitely. These basic data can then be post-integrated in time and in height for temperature and more accurate sodium density measurements. (The post integration being to improve signal/noise ratio.) As the result we can obtain the temperature and the more accurate sodium density data with time-resolution of several to tens minutes and height-resolution of a few km.

3. Observational Results and Discussion

[8] In nighttime on 11 January 2011, a Es appeared above Tromsø, which was not only detected by the ionosonde [Hall and Hansen, 2003] at the Tromsø but also the co-located MF radar [Hall, 2001] with the transmit frequency of 2.8-MHz (see Figure 1a). According to the ionosonde data, there was no ordinary E region after 17:15 UT; the Es established itself at around 18:30 UT and disappeared at 22:30 UT. The Es formed a descending layer from 105 km at 19:15 UT until 20:30 UT, thereafter fluctuating between 95 and 100 km. We can see strong echoes in the MF-radar received power data. The strong echoes, well corresponding to the bottom of the Es (i.e. the hEs), were descending with time for 19:00–20:00 UT, and reached below 100 km around 20:00 UT. At almost the same time, a dramatic increase was clearly seen in the sodium lidar received emission intensity at 96–97 km height (see Figure 1b). It is considered that this increase is mainly due to the sodium density enhancement, because such an increase was seen in the case of both two laser frequencies. After that, the strong echoes were seen from around 95–105 km until 22:30 UT, after which the strong echoes disappeared. The layer of enhanced sodium density at 95–98 km lasted for ∼3.5 hours. At higher altitudes (around 100 km and above), such sodium density-enhanced layers appeared at 19:30–20:30 UT and 21:00–22:00 UT. (It may be difficult to see the enhanced layers around 19:30 UT due to data gaps in the temporal coverage.) After 22:30 UT, such an enhanced layer was seen also below 95 km. These sodium density-enhanced layers are considered to be the SSLs. The relationship between the emergence of the SSL at 95–98 km and the descending Es is similar to the observational result by Heinselman et al. [1998], supporting that the SSL generation mechanism would be the ion neutralization from the Es. Of particular interest are clear wavelike structures with time scales of several minutes in the observed SSLs, found in the high resolution data of 5-sec and 96-m. Such short-period wavelike structures were not detected in the strong echoes from the MF radar data with the resolution of 5-min and 3-km.

Figure 1.

(a) The MF-radar (2.8 MHz) received power data with a resolution of 5-min and 3-km, at 74–109 km for 19:00–24:00 UT on 11 January 2011. Diamond symbols indicate hEs determined with the ionosonde at Tromsø. It should be noted that the height information of the strong echoes and the hEs could be affected by a time delay, i.e. group retardation, in the radio propagation. (b) The sodium lidar received emission intensity data with a resolution of 5-sec and 96-m. It should be noted that the emission intensity includes an artificial variation due to changing of the laser frequency at every 1-min. (c) The peak height and (d) the FWHM in the emission intensity of the sporadic sodium layer. Red symbols indicate those at 95–98 km, and blue symbols indicate those at 92–95 km. The FWHM is also described by vertical gray bar in Figure 1c.

[9] We have extracted the peak height and the FWHM of the SSLs at 92–99 km and 19:00–24:00 UT (see Figures 1c and 1d), avoiding the artificial emission intensity variation due to the laser frequency change. The short-period wavelike structures were seen in the peak height during the entire period at 95–98 km and after 2250 UT at 92–95 km. Multiple peaks occasionally appeared in narrow a height range of 1–2 km around 20:20 and 21:30 UT. The FWHM gradually increased with time perhaps due to diffusion and/or mixing, but seemed to be extremely enhanced during the multiple-peak appearances. The increasing rate of the FWHM was more significant at 92–95 km than at 95–98 km, implying that the diffusion and/or the mixing were more effective at the lower height. In the event beginning (20:00–20:10 UT), the FWHM was small (∼0.5 km) and was fairly constant. Thus, in the first 10 minutes of the SSL event, the diffusion and/or the mixing did not become effective yet. This feature may imply that the dramatic increase of the emission intensity during the first 10 minutes was induced mostly by the sodium atom generation above Tromsø rather than the advection of the SSL.

[10] Figures 2a and 2b show an enlarged part of Figure 1b and height-integrated emission intensity for 94–98 km during 19:55–20:20 UT, respectively. As shown in Figure 2a, the high resolution data of 5-sec and 96-m detected that the SSL within narrow height range of ∼1 km had a periodic motion in height with time-scale of several minutes. The increase of the emission intensity seemed to begin around 19:58 UT. Here, to determine the sodium density variation, we focus on the height-integrated emission intensity data taken by the −151.4-MHz laser frequency, i.e. solid line in Figure 2b, which fairly correspond to the sodium density (see Section 2). Before the SSL event, at 19:56:57–19:57:47 UT, the 4-km integrated emission intensity was ∼3.7 × 104 counts. The ∼3.7 × 104 counts corresponds to the 4-km integrated sodium density of ∼8.5 × 1012 m−2, which has been calculated from 30-min integrated data with the two frequency method. After that, we can see smooth increases at ∼19:59–20:00 UT and at ∼20:01–20:02 UT in the data taken by the −151.4-MHz laser frequency. The 4-km integrated emission intensities were ∼5.1 × 104 counts at 19:58:57 UT, and ∼7.2 × 104 counts at 19:59:47 UT. This increasing rate (∼2.1 × 104 counts per 50 sec) corresponds to ∼9.6 × 1010 m−2 s−1 in the height integrated sodium density. In the same manner, the increasing rate for 20:00:57–20:01:47 UT was ∼6.4 × 1010 m−2 s−1. The rates of 6.4–9.6 × 1010 m−2 s−1 are significantly large compared with the modeled and observed results (e.g., ∼2.0 × 1013 m−2 per 30 min) by Cox and Plane [1998], Heinselman et al. [1998], and Heinselman [2000]. This large increase of the sodium density actually occurred within ∼1 km height range (see Figure 2a).

Figure 2.

(a) The sodium lidar received emission intensity data at 92.5–99.5 km for 19:55–20:20 UT on 11 January 2011, enlarged view of Figure 1b. (b) The emission intensities integrated for 94–98 km in height. Solid line indicates the −151.4-MHz laser frequency data, which is fairly insensitive to the temperature and is therefore a representative of the sodium content.

[11] To investigate the short-period wavelike structures in the SSLs, detected with the high resolution data of 5-sec and 96-m (see Figures 1b and 2a), we have calculated the normalized power spectral density (PSD) of the peak height in the emission intensity of the SSL (the peak height series of Figure 1c) based on the Lomb-Scargle method [Press and Rybicki, 1989; Hocke, 1998]. As shown in Figure 3, it is found that predominant periods of the wavelike structures were 7–11 min (0.0015–0.0024 Hz) at 95–98 km and 3 min (0.0056 Hz) at 92–95 km. The 7–11-min wavelike structure would be signatures of atmospheric gravity waves. On the other hand, for the 3-min wavelike structure may be signatures of an atmospheric gravity wave or acoustic wave. Concerning to variations at periods of 10-sec to 2-min, there were no remarkable variations with the significance level of more than 95% in this event. Another point of interest is the relationship between the PSD and the frequency, which has interpretations for waves and turbulence. The shape of the PSD at 95–98 km is well represented by a straight line with a power law index of −1.6. The index of −1.6 was larger than that at 92–95 km (−1.1). In both cases, no white noise was seen, implying significant waves and turbulence in frequency range of 0.0001–0.1 Hz. For Kolmogorov turbulence, the one-dimensional spectrum is a power law with an index of −5/3. The observed index (−1.6) at 95–98 km is very close to −5/3 and similar to an observed result (−1.9) by Pfrommer et al. [2009], while it should be noted that the calculated PSDs are not necessarily equivalent to the statistical features reported previously.

Figure 3.

Normalized power spectral density (PSD) of the emission intensity peak height in the sporadic sodium layers (a) at 95–98 km (from the peak height series shown by red symbols in Figure 1c) and (b) at 92–95 km (from the peak height series shown by blue symbols in Figure 1c). The red lines correspond to power law indices of −1.6 at 95–98 km and −1.1 at 92–95 km. 1-sigma uncertainties for the indices of −1.6 at 95–98 km and −1.1 at 92–95 km are 0.005 and 0.118 respectively.

4. Summary

[12] The high time-resolution measurement of 5-sec with the Tromsø sodium lidar reveals shorter time-scale atmospheric features, compared with previously published results, at a high-latitude region. The present paper is just the initial report of the fine structure of the SSL observed on 11 January 2011, and thus we need more extended study of the fine structures of the atmospheric features. For example, the observed dramatic sodium density increases at rates of 6.4–9.6 × 1010 m−2 s−1 might involve not only temporal variation but also spatial variation. The periods of 7–11-min and 3-min in the wavelike structure were not identical to the intrinsic periods of the waves. To distinguish between temporal and spatial variations, it is important to perform multi-point measurements using by the sodium lidar and/or simultaneous observations with some optical observations (such as all-sky imager observations). Furthermore, more comprehensive observations, by using the sodium lidar and also radars (such as the MF and meteor radars), will allow us to investigate background temperature and wind velocity, obtaining better understanding of short-period waves and also atmospheric instabilities.


[13] The Tromsø sodium lidar project is mainly supported by Special Funds for Education and Research (Energy Transport Processes in Geospace) from MEXT, Japan, in collaboration with Nagoya University, Shinshu University, RIKEN, University of Tromsø, and EISCAT Scientific Association. We wish to appreciate K. Shiokawa and T. Motoba for their valuable comments. We thank K. Hocke for letting us use his Lomb-Scargle periodogram method routines. Efforts of A. H. Manson and C. E. Meek allow us to use the MF radar data. The ionosonde is operated by University of Tromsø. This research is partly supported by the Grant-in-Aid for Nagoya University Global COE Program, “Quest for Fundamental Principles in the Universe: from Particles to the Solar System and the Cosmos” from MEXT, Japan, by a Grant-in-Aid for Scientific Research B (18403010, 22403010) from MEXT, Japan, and by Norwegian Research Council Project 184750/V30.

[14] The Editor thanks Mark Conde and an anonymous reviewer for their assistance in evaluating this paper.