Using incoherent scatter radar to investigate the neutral wind long-term trend over Arecibo

Authors


Abstract

[1] Thermospheric neutral winds can be the most important driver when modeling ionospheric densities and temperatures. Several papers in this special edition show interesting features of the neutral winds behavior during the last 30 years at the Arecibo Observatory (18.3°N, 66.75°W; ∼28.25° dip latitude) using Fabry-Perot Interferometer (FPI) data. A neutral wind vector that changes its direction, becoming more dominantly eastward over the years and a meridional neutral wind component that decreases in magnitude, were found. The main goal of this work is to look for similar evidence of long-term trends in the radar derived winds that might support these recent discoveries and explore the associated ionospheric parameter measurements to look for the effects of these changing winds on the ionosphere. With this purpose in mind, Incoherent Scatter Radar (ISR) measurements of the F region vector drifts are used to derive the corresponding meridional thermospheric neutral wind along the magnetic field over Arecibo during 20 years. Major findings include a decreasing long-term trend (lowering) of the height where the F2- layer peak density occurs (hmF2), which could be related with a more increased downward flux of the ionosphere. A slight decrease in the peak density of the F2- layer (NmF2) after local midnight also was found during the period studied. The meridional wind along the magnetic field derived from ISR data also revealed a long-term trend, becoming more northward during the period studied, with a maximum variation between 02:30UT and 05:30UT.

1. Introduction

[2] For the last decades scientific groups all over the world have been studying the long-term trends of ionospheric parameters in an attempt to better understand the global change in properties of the upper atmosphere. Initially, theses investigations were mainly addressed to verify a possible connection between the ionospheric long-term behavior and the increasing thermospheric-mesospheric cooling due to the enhancement of greenhouse gas emissions [Roble and Dickinson, 1989; Rishbeth, 1990; Rishbeth and Roble, 1992]. Since then, the greenhouse hypothesis has received serious support from the results that revealed a steady decrease in the thermospheric density over the period of 2–3 solar cycles [Keating et al., 2000; Emmert et al., 2004; Marcos et al., 2005]. However, the global pattern of experimental NmF2 and foF2 (the F2-peak plasma frequency) of several worldwide stations over the last 40 years is highly complex and can hardly be reconciled only with that hypothesis [Upadhyay and Mahajan, 1998; Bremer, 2001; Mikhailov and Marin, 2001].

[3] One of the key elements used to understand the upper atmospheric global changes are the thermospheric neutral winds. They are coupled with the ion flow through ion-neutral collisions, playing a direct role in the transport of ionospheric plasma along the magnetic field lines and modifying the ion densities above about 200 km by affecting the rate at which the O+ ions diffuse downward. The most obvious effects of these winds are on NmF2 and hmF2 behavior [Richards, 1991]. In addition, the neutral winds also modulate ionospheric phenomena such as equatorial spread F, scintillation, and high-latitude convection [Emmert et al., 2006].

[4] At the Arecibo Observatory the direct optical observation of the neutral winds has been made using a FPI to measure the Doppler shift of the O (1D) 630nm line profile since the 1980s' [Burnside et al., 1981; Burnside and Tepley, 1989; Burnside et al., 1991a, 1991b; Emmert et al., 2006]. Recently, Tepley et al. [2011] and C. G. M. Brum et al. (Long-term changes in the thermospheric neutral winds over Arecibo: A climatology based on over three decades of Fabry Perot observations, submitted to Journal of Geophysical Research, 2011) analyzed 30 years (approximately three complete solar cycles) of thermospheric neutral winds at Arecibo using FPI data. The Tepley et al. [2011] study focused on the annual variability of the neutral winds, while Brum et al. (submitted manuscript, 2011), using the same database and a different statistical approach, focused on the time and seasonal climatology and variability of the Arecibo thermosphere. Their results agree on a long-term trend of the thermospheric neutral wind velocities during the last 30 years. Additionally, Tepley et al. [2011] found a gradual counter-clockwise rotation of 26° of the neutral wind vector, becoming more dominantly eastward over the years.

[5] Within this framework, we developed a study using ISR data obtained between 1985 and 2004 at the Arecibo Observatory as a tool to investigate how the ionospheric parameters had been related with the long-term behavior of the neutral winds over Arecibo in the last 30 years. Section 2 of this work describes the data and the methodology used in this study and also includes the approach used to develop an empirical model of the neutral wind parameters observed by the ISR. We follow with a discussion on the results obtained (Section 3) and provide our final conclusions at Section 4.

2. ISR Data Distribution and Methodology

[6] The database used in this work was obtained from 1985 to 2004 during World Days experiments with the “beam swinging” technique developed by Hagfors and Behnke [1974]. This method takes advantage of the azimuth steering capabilities of the Arecibo ISR, so that the spectra is measured while continuously rotating the antenna beam back and forth 360° in azimuth at 15° off zenith. The line-of-sight velocities obtained from the ISR spectra are then converted into vector velocities by assuming a constant vector field in one rotation and neglecting horizontal gradients.

[7] Although the Arecibo ISR database with vector components is available since mid-1970s our study is restricted to the period cited above in order to keep the same pattern of data, i.e., the database using the “single beam” experiment and the multiple-radar autocorrelation function (MRACF) technique developed by Sulzer [1986]. Even with this restriction leading to a marginal long-term trend when compared with the entire database available, the methodology based on the seasonal and solar dependencies developed in this work allows for different statistical conditions to be considered leading to reliable results.

[8] Given the characteristics of the Arecibo ISR system (carrier frequency of 430 MHz; peak radiated power 1.5–2.5 MW and a 305m diameter dish) and the multiradar technique it is possible to measure scalar parameters of the F region ionosphere like electron density, electron and ion temperatures, and ion line-of-sight velocity in just 10–20 s [Aponte et al., 2005].

[9] The determination of the meridional neutral wind velocity in the direction of the geomagnetic meridian from ISR data is a well-known technique [see, e.g., Behnke and Harper, 1973; Wickwar et al., 1984; Salah and Holt, 1974; Burnside et al., 1991b, 1983]. In this work we follow the approach reviewed by Aponte et al. [2005] where the equatorward winds along the magnetic meridian can be expressed as:

equation image

where I is the magnetic dip angle (which decreased approximately 2.5° during the period analyzed in this work), vap and vd are the anti - parallel plasma drift obtained from the observed ion drift vector measurements and the anti-parallel diffusion velocity of O+ ions through the neutral gas along the magnetic field line, respectively. vd is derived by the relation:

equation image

where Tp = (Ti + Te)/2 and Tr = (Ti + Tn)/2 with Ti, Te and Tn being ion, electron, and neutral temperatures, respectively; ne is the electron density; Hp is the plasma scale height (Hp = 2kbTp/mig; where kb is the Boltzman constant, mi is the ion mass, and g is the gravitational acceleration); and Da is the ion-neutral diffusion coefficient given by

equation image

where υin is the collision frequency between the dominant ion O+ and the major neutral constituents O, N2, and O2. For image we use the formula of Pesnell et al. [1993], and for the other collision frequencies image + image from Banks [1966]. The neutral densities and Tn are obtained from MSIS-90 [Hedin, 1991]. Ti and vap are measured by the ISR.

[10] The calculations were developed for the nighttime period where there is little or no energy input into the F region and the ion, electron and neutral temperatures at a given height should be equal [Cogger et al., 1970]. Also, the calculations follow the ionospheric peak hmF2, where the vertical gradients of electron density, ion, electron, and neutral temperatures can be neglected, i.e., equation imageequation imageequation image ≈ 0 [Vasseur, 1969; Behnke and Kohl, 1974].

[11] In this study, 218 nights between 1985 and 2004 are used. The data points were averaged each 15 min given a total of 5,779 points, of which 50% were from approximately F107 index <120 and a mean Kp value around 2+. During enhanced geomagnetic activity the heating and expansion of the polar atmosphere produce large winds blowing away from the poles, in the E and F regions [Titheridge, 1995]. In order to avoid these effects, the data with Kp index 4 were discarded in our analysis.

[12] The database distribution according to different levels of solar and geomagnetic activity based on the F10.7cm Solar Flux and Kp index, respectively, is shown in Figure 1 for three different seasons: June Solstice, Equinoxes, and December Solstice. The data points are distributed in four intervals of time (0.5 ≤ UT ≤ 2.5, 2.5 ≤ UT ≤ 4.5, 4.5 ≤ UT ≤ 6.5, and 6.5 ≤ UT ≤ 8.5) to make easier the visualization of the sample.

Figure 1.

The distribution of Arecibo ISR data as a function of solar and geomagnetic conditions. (left) The number of observations as a function of F10.7 level; (right) illustration of the number of observations as a function of Kp index for four intervals of time. The different shades of gray describe the seasons.

[13] Figure 2 shows the variation of altitude where the neutral winds were calculated (hmF2), the parameters used to calculate vd (ν(O+,O) and Ti), as well as the observed vap projected in the horizontal direction (−vap.secI) of the entire data set available to this study as a function of Universal Time (UT). The solid lines with white circles in each panel represent the 30-min averages and the error bars their respective standard deviations. From the panels of Figure 2, general features of the four parameters can be visualized: (−vap.secI) intensifies after local midnight (4UT) and reaches its maximum at the end of the night; Ti has higher values before local midnight reaching its minimum at the end of the night; v(O+,O) reaches its minimum around local midnight while hmF2 drops faster after local midnight. In addition, from the hmF2 profiles it is possible to see that all the Unm data were derived between approximately 250 and 450 km.

Figure 2.

Profiles of v(O+,O) values and −vap.secI, Ti, and hmF2 measured by the Arecibo ISR between 1985 and 2004 (UT = LT + 4 h). The solid lines with white circles in each panel represent the averages and the error bars the respective standard deviations.

[14] Figure 3 shows both −vd.secI and the Unm computed values of the entire data set available for this study as a function of the Universal Time. As in Figure 2, the solid lines with white circles in each panel represent the 30-min averages and the error bars their respective standard deviations of the data points. The similarity between −vd.secI and Unm behavior is explained by the major contribution of the term −vd.secI when compared with −vap.secI (shown in Figure 2), to the derivation of Unm.

Figure 3.

Profiles of (a) −vd secI and (b) Unm calculated using ISR data during 1985 and 2004 as a function of Universal Time. The solid lines with white circles represent the average values and their respective standard deviations (±1σ).

[15] Before proceeding with our analysis, the Unm values shown in Figure 3b need to be validated once the vertical gradients of electron density and temperature were neglected in the calculations. In order to facilitate the following comparison between the neutral winds derived from ISR and those measured by the FPI, we will use the meridional component calculated in the northward direction along to the magnetic meridian (Unm), where: Unm = −Us.

[16] At F region the horizontal winds at different heights are coupled through frictional drag which is proportional to μ/ρ, where μ is the coefficient of viscosity (approximately constant). The mean gas density ρ decreases exponentially with height, given an increase in the frictional drag such that, at heights above about 250km, the horizontal velocity becomes independent of height [Titheridge, 1995]. At Arecibo, despite the fact that Burnside et al. [1983] observed some degree of vertical shear on the meridional winds prior to midnight and between 02:00 and 05:00 AST in spring and summer months, in general they found that the meridional wind was fairly constant with altitude. Therefore, for the subsequent calculations, the horizontal components of the neutral winds were then assumed to be constant altitudes above 250km.

[17] Figure 4 shows a comparison between the Unm derived from ISR and the winds simultaneously measured using the FPI installed at the Arecibo Observatory for 16 days under different solar, seasonal and geomagnetic conditions. For each day, two panels are shown. The top panels present the altitude hmF2 where the ISR data were measured and, for the bottom panels, Unm derivates from the ISR (lines) and those measured by the FPI (circles). The bars are the standard deviation error of the FPI data. The occurrence of some big error bars for Unm measured with the FPI can be explained by the fact that it is measured through the O (1D) 630nm emission and is subject to fluctuations in the local weather conditions.

Figure 4.

Simultaneous experiments of Unm derived from ISR (lines) and those measured by the FPI (circles) under different conditions of F107 cm flux and Kp index activity solar activity.

[18] In a general overview, we can conclude that the Unm derived from ISR data is reliable and can be used to examine how the ionosphere over Arecibo is playing a role in influencing the neutral winds of the low-latitude thermosphere.

3. Unm Empirical Model Development

[19] The development of an empirical model of Unm based on solar and seasonal dependencies is an important mathematical tool to predict the behavior of this parameter when looking for a long-term trend. As mentioned before, the methodology used in this work consist in sort the 5779 data points into groups based on season (represented by the Day of the Year - Doy), time and solar activity (we are using F107cm index as solar activity proxy).

[20] The panels in Figure 5 illustrate the steps of this methodology applied for nine intervals of time. At the top block of panels, the dots represent the residual average of Unm (ms−1) calculated for one hour time interval and ±2.5 SFU (1 SFU = 10−22 W/m2/Hz). These residuals were obtained excluding any seasonal dependence. In addition, for a better visualization of the Unm behavior by time due to solar activity variation, we normalized the profiles to be zero in a condition of F107 = 120SFU. The solid lines of these panels are the best sigmoid function representation of the data. We choose this function because it has asymptotic characteristics, similar to the responses of the Unm with solar activity. The sigmoid function is often used when a detailed description is lacking. It is produced by a mathematical function having an “S” shape. Many natural processes, including those of complex system learning curves, exhibit a progression from small beginnings that accelerates and approaches a climax over time. For this work we are just using the second half of the sigmoid function to describe the thermospheric neutral wind responses to the solar activity based on the F107cm index. At the sigmoid function the center and the growth rate of the slope can be define. In the middle block of panels we are presenting the residual seasonal variation of the Unm for each interval of time (black dots - any solar dependence was excluded) and the solid black lines represent their reconstruction by Fourier transform. Finally, at the bottom block of panels the dispersion diagrams between the developed empirical model predictions (which is the sum of the dependences presented in the above panels by time) and the Unm data for the entire period and their best linear fit are shown. The overall prediction of our model is very consistent since the slope (SLP) and the correlation coefficient (R) obtained are very close to unity.

Figure 5.

Dispersion diagrams of the methodology applied for nine intervals of time. (top) The residual average of Unm (ms−1) (dots) calculated for one hour interval time and ±2.5 SFU (solar flux) are presented; the solid lines are the best sigmoidal function representation of the data. (middle) The residual seasonal variation of the Unm for each interval of time (dots); the solid lines represent their reconstruction by Fourier transform. (bottom) The dispersion diagrams between the developed empirical model predictions and the Unm data for the entire period are shown as well as their best linear fit.

4. Results and Discussion

[21] As mentioned above, Tepley et al. [2011] and Brum et al. (submitted manuscript, 2011), using different methodologies, found a long-term trend in the thermospheric neutral wind field over Arecibo when analyzing three decades of FPI data. The rates found for both authors differ in magnitude, which can be explained by the different statistical approaches followed by each method. A particularly feature found by both studies is that the zonal and meridional wind components presented different yearly velocity rates during the last 30 years. Tepley et al. [2011] associate this result with the different influence that the ionosphere has on each wind component as a possible cause of the rotation of the neutral wind vector observed. Using ISR data to investigate the behavior of the meridional neutral winds over Arecibo has the advantage to look for similar indications of these long-term trends in the ionospheric components that might correlate with, and affect the neutral flow.

[22] The term −vapsecI and the corresponding diffusion velocity (−vd.secI) are linked to the magnetic northward meridional neutral winds, Unm, through the transfer of momentum via ion-neutral coupling. Therefore, a change in −vapsecI would affect the Unm as estimated from ISR measurements. We computed the variations of Unm due to the magnetic declination changes during the period in study (the biggest variation computed was around 4ms−1 southward for December solstice) and this variation was taken into account in our calculation.

[23] Figure 6 shows the long-term behavior of Unm, −vapsecI and −vdsecI from ISR measurements as a function of year for different intervals of time based on the difference between the data and their prediction values is shown is this figure in m.s−1. From the linear approximation of these results, the long-term trend of each parameter is found for each time interval.

Figure 6.

The long-term behavior of (from top to bottom) Unm, −vapsecI and −vdsecI as a function of year for nine different intervals of time.

[24] Examining the term −vapsecI shown in Figure 6, we note that this velocity is becoming more northward during the years for the period between 00:00UT and 06:00UT. The maximum variation is found before 04:00UT. Around 08:00UT −vapsecI reverses slightly its direction, becoming more southward or with no variation. The meridional wind along the magnetic field also becomes more northward during the years with a maximum variation between 02:00UT and 05:00UT. The only exception is for the time interval between 07:00UT ≤ 08:00UT where Unm presents almost no variation. These results agree with that presented in the two companion papers of Brum et al. (submitted manuscript, 2011) and Tepley et al. [2011].

[25] A direct consequence of −vapsecI increasing over of the years would be intensified downward flux of the ionosphere, that would push the height of hmF2 to lower altitudes at Arecibo. This is illustrated in Figure 7, where hmF2 measured by the Arecibo ISR during the period studied is shown. The data points show the difference in (km) between hmF2 data and their empirical values derived from the Brum et al. [2011] model. A negative trend of hmF2 is observed during the entire night, but mainly after 05:30 UT where a decrease in hmF2 is more emphasized. Before 05:30UT hmF2 decreased approximately 4 km during the years and after this time approximately 10 km.

Figure 7.

hmF2 long-term trends during the period studied. The data points show the difference in km between the hmF2 data and their empirical values.

[26] Figure 8a shows how the yearly rate (ms−1year−1) assuming a linear variation with the year of the three correlated terms: Unm, −vap.secI, −vd.secI vary during the period studied, as a function of Universal Time. Unm is increasing for almost the entire period, presenting the highest rates around local midnight (4UT); −vap.secI also is also increasing for the most part of the time, with the highest rates observed around 3UT; −vd.secI is closely modulated by Unm.

Figure 8.

(a) Yearly rates of Unm, −vd.secI and −vap.secI; (b) hmF2; and (c) foF2 as a function of Universal Time for the period studied.

[27] Figures 8b and 8c show the yearly rate variation of hmF2 (km−1year−1) and foF2 (MHz−1year−1) for the period studied. foF2 is related to NmF2 by:, where NmF2 is given in elec.cm−3 and foF2 is in MHz. When comparing these figures, in the beginning of the night (before 03:30UT) −vap.secI rate is higher, increasing the downward flux and pushing the ionosphere down faster (hmF2). Unm component is also increasing but at a higher rate than −vap.secI, due to the term −vd.secI that, at this time, is increasing too. NmF2 (represented by Figure 8c) has its maximum at this period, due to the residual ionization from the daytime ionosphere. Approximately at 04:00 UT (local midnight), NmF2 starts to decrease and this behavior increases the relative effectiveness of −vd.secI, permitting the wind to flow more northward during the period [Tepley et al., 2011]. Around this time, hmF2 illustrates the minimum rate of variation. At the end of the night, Unm, −vap.secI and −vd.secI all reach their minimum rates as well hmF2 and foF2.

5. Summary

[28] In this work, we used ISR measurements of the F region vector drifts to derive the corresponding thermospheric neutral wind in the northward direction along the magnetic field over Arecibo. The purpose was to look for indications of long-term trends in the ionospheric components that might correlate with the recent discoveries of Brum et al. (submitted manuscript, 2011) and Tepley et al. [2011] regarding the behavior of the thermospheric neutral winds. The ISR data were all obtained during World Day experiments at Arecibo during nighttime periods between 1985 and 2004.

[29] The magnetic meridional winds, Unm, derived from ISR measurements were compared with those measured using a FPI installed at the Arecibo Observatory during different solar and seasonal conditions. The solar and seasonal dependencies of Unm were found using an empirical model based on the ISR data.

[30] The meridional wind along the magnetic field derived from ISR shows a long-term trend, becoming more northward during the period studied, with a maximum variation between 02:30UT and 05:30UT. These results agree with, and support the conclusions of the two companion papers by Brum et al. (submitted manuscript, 2011) and Tepley et al. [2011].

[31] The term −vapsecI also was found to become more northward during the period studied. The maximum variation was found before 03:30UT. A direct consequence of the −vapsecI increasing over of the years was intensified downward flux of the ionosphere, which pushed the height of hmF2 at Arecibo to lower altitudes. ISR measurements confirmed a decreasing trend of hmF2 mainly after 05:30UT during the period studied. Before 05:30UT hmF2 decreased approximately 4 km during the years and after this time the variation reached approximately 10 km.

[32] A negative long-term trend of NmF2 (and consequently, of foF2) also was observed by ISR measurements after approximately 04:00 UT (local midnight). This helped increase the effectiveness of −vd.secI and permitted the wind to flow more northward during the period, in agreement with the qualitatively test done by Tepley et al. [2011].

[33] As summarized above, analyses of the associated ionospheric parameter measurements from ISR demonstrates a consistency between the ion and neutral behaviors at Arecibo, i. e., the ion drifts are responding similarly with the neutral winds during the last years. Although a negative long-term trend of NmF2 was detected in the Arecibo's ionosphere during the period studied, the magnitude of the changes was not as that outlined by Roble and Dickinson [1989] and further studies need to be developed to quantify how this drop in neutral density is really related the greenhouse hypothesis.

Acknowledgments

[34] The Arecibo Observatory is operated by SRI International under a cooperative agreement with the National Science Foundation (AST-1100968), and in alliance with Ana G. Méndez-Universidad Metropolitana, and the Universities Space Research Association.

[35] Robert Lysak thanks the reviewers for their assistance in evaluating this paper.