Locations of chorus emissions observed by the Polar Plasma Wave Instrument

Transcription

1 Click Here for Full Article JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115,, doi: /2009ja014579, 2010 Locations of chorus emissions observed by the Polar Plasma Wave Instrument K. Sigsbee, 1 J. D. Menietti, 1 O. Santolík, 2,3 and J. S. Pickett 1 Received 18 June 2009; revised 20 November 2009; accepted 17 December 2009; published 8 June [1] We performed a statistical study of the locations of chorus emissions observed by the Polar spacecraft s Plasma Wave Instrument (PWI) from March 1996 to September 1997, near the minimum of solar cycles 22/23. We examined how the occurrence of chorus emissions in the Polar PWI data set depends upon magnetic local time, magnetic latitude, L shell, and L*. The Polar PWI observed chorus most often over a range of magnetic local times extending from about 2100 MLT around to the dawn flank and into the dayside magnetosphere near 1500 MLT. Chorus was least likely to be observed near the dusk flank. On the dayside, near noon, the region in which Polar observed chorus extended to larger radial distances and higher latitudes than at other local times. Away from noon, the regions in which chorus occurred were more restricted in both radial and latitudinal extent. We found that for high latitude chorus near local noon, L* provides a more reasonable mapping to the equatorial plane than the standard L shell. Chorus was observed slightly more often when the magnitude of the solar wind magnetic field B SW was greater than 5 nt than it was for smaller interplanetary magnetic field strengths. We also found that near solar minimum, chorus is twice as likely to be observed when the solar wind speed is greater than 450 km/s than it is when the solar wind speed is less than 450 km/s. Citation: Sigsbee, K., J. D. Menietti, O. Santolík, and J. S. Pickett (2010), Locations of chorus emissions observed by the Polar Plasma Wave Instrument, J. Geophys. Res., 115,, doi: /2009ja Introduction [2] Whistler mode chorus emissions are electromagnetic waves at frequencies between about one tenth of the equatorial cyclotron frequency, f ce /10, up to near f ce, which typically covers a frequency range from a few hundred Hz to several khz [see Burtis and Helliwell, 1969; Tsurutani and Smith, 1974; Burtis and Helliwell, 1976]. When viewed on timescales of minutes or hours, chorus typically appears in time frequency spectrograms as narrow band emissions that sometimes have a gap near f ce /2 that divides these waves into a lower and an upper band. When viewed on timescales of a few seconds, a time frequency spectrogram of chorus emissions reveals discrete chorus elements, which correspond to individual whistler mode wave packets. Chorus elements typically consist of rising tones, but falling tones and elements with more complicated structures have also been observed [Helliwell, 1965; Burtis and Helliwell, 1976; Lauben et al., 2002]. [3] Substorm electron injections can excite intense whistler mode chorus emissions near the geomagnetic equator, outside the plasmapause [Tsurutani and Smith, 1974, 1977; 1 Department of Physics and Astronomy, University of Iowa, Iowa City, Iowa, USA. 2 Institute of Atmospheric Physics, Academy of Sciences of the Czech Republic, Prague, Czech Republic. 3 Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic. Copyright 2010 by the American Geophysical Union /10/2009JA Meredith et al., 2001, 2002; Summers et al., 2004]. Chorus emissions are thought to be generated through nonlinear processes involving an electron cyclotron resonance between whistler mode waves and energetic electrons in the outer radiation belt [Helliwell, 1967; Nunn et al., 1997]. These waves have also been associated with solar wind pressure pulses [Lauben et al., 1998] and they may play an important role in geomagnetic storms [Santolík et al., 2004; Li et al., 2007]. Chorus emissions may be able to accelerate electrons in the outer radiation belts to energies greater than 1 MeV through a stochastic process occurring in the low density region outside the plasmapause [Horne and Thorne, 1998; Summers et al., 2002; Horne et al., 2003, 2005a, 2005b]. Recent simulation results have suggested that relativistic turning acceleration due to nonlinear particle dynamics could also accelerate electrons from a few hundred kev up to a few MeV through a resonant wave trapping process [Omura et al., 2007]. High energy electron microbursts are thought to be associated with chorus [e.g., Oliven and Gurnett, 1968; O Brien et al., 2003] and may be related to auroral X rays observed on balloon experiments [Anderson and Milton, 1964]. Other research has suggested that the scattering of electrons by chorus emissions may contribute to particle precipitation in the diffuse aurora [Inan et al., 1992; Ni et al., 2008; Meredith et al., 2009]. [4] Although chorus emissions are thought to be generated in the equatorial plane, chorus like electromagnetic emissions can be found at higher latitudes. Some studies have suggested that the whistler mode waves observed at 1of17

2 high latitudes are chorus generated near the magnetic equator that has propagated to high latitudes [e.g., Bortnik et al., 2007]. Under certain conditions, it appears that these waves can be reflected back toward the equator after propagating to high latitudes [Chum and Santolík, 2005; Santolík et al., 2006]. There has also been evidence that some of the off equatorial whistler mode emissions may be generated in minimum B pockets at high latitudes on the dayside, near the cusp [Tsurutani and Smith, 1977; Pickett et al., 2001; Vaivads et al., 2007; Tsurutani et al., 2009]. [5] In this paper, we will explore how the occurrence probability of chorus emissions in the Polar Plasma Wave Instrument (PWI) data set depends upon magnetic local time, magnetic latitude, L shell, and L*. We will also discuss how the chorus occurrence probability depends upon the Kp and Dst geomagnetic indices, as well as upstream solar wind conditions. [6] The Polar spacecraft was launched on 24 February 1996 into a highly elliptical orbit, initially with apogee at 9 Earth radii (R E ) in the Northern Hemsiphere, perigee at 1.8 R E in the Southern Hemisphere, and an approximately 90 inclination [Acuña et al., 1995]. The Polar PWI made observations of plasma waves in the Earth s magnetosphere between frequencies of 0.1 Hz and 800 khz from March 1996 to September During 1996 and 1997, as the Polar spacecraft traveled from high latitudes in the Northern Hemisphere down to the equatorial plane, it often skimmed along the boundary of the plasmapause, cutting through the Van Allen radiation belts and entering the plasmasphere at low altitudes near the equator. After reaching perigee in the Southern Hemisphere, Polar encountered similar regions as it traveled back to high latitudes in the Northern Hemisphere. The 18 h period of Polar s orbit provided roughly two passes through the radiation belts every day. [7] The extended coverage at high latitudes provided by Polar is especially important near local noon, where structured, chorus like emissions can often extend into the vicinity of the cusp. These high latitude dayside emissions can feature the classic rising elements of equatorial chorus, but we have also found falling tones, elements with hooklike structures, combinations of different spectral forms, and possible triggered emissions [see, e.g., Helliwell, 1967, p. 206] in the Polar PWI data set at high latitudes. Regardless of their original source, these waves propagate in the whistler mode and may interact with electrons in a similar manner to chorus generated near the equator. Our study of chorus using the Polar PWI data set provides a different perspective on chorus emissions from earlier studies using satellites such as CRRES [Meredith et al., 2002], which were not polar orbiting and unable to observe high latitude chorus events near noon. The studies of chorus using CRRES data [e.g., Meredith et al., 2002, 2003] were also performed using data taken near the maximum of solar cycle 22, while the Polar PWI data were available near the minimum of solar cycle 22/23. Our study is therefore complementary to past statistical studies using data from CRRES and other satellites. 2. Data Sets and Models 2.1. Polar Plasma Wave Instrument [8] The suite of 12 scientific instruments on board the Polar satellite included the Polar Plasma Wave Instrument (PWI) [Gurnett et al., 1995], which used three orthogonal electric dipole antennas provided by the Electric Field Instrument (EFI) [Harvey et al., 1995], two in the spin plane and one aligned along the spacecraft spin axis. Magnetic fields were measured using the PWI s magnetic loop antenna and triaxial magnetic search coil antenna. The Polar PWI used five receiver systems to process signals from these antennas: a digital wideband receiver (WBR), a highfrequency waveform receiver (HFWR), a low frequency waveform receiver (LFWR), two multichannel analyzers (MCA), and a pair of sweep frequency receivers (SFR). A key feature of the Polar plasma wave instrument was the ability to make simultaneous measurements from six orthogonal electric and magnetic field sensors. This feature permitted the analysis of the chorus wave vectors [LeDocq et al., 1998; Sigsbee et al., 2008]. The WBR was capable of using up to 8 bit resolution and sample rates as high as 249 k samples/s, providing sufficient time and frequency resolution for viewing individual chorus elements. These features of the Polar PWI are essential for identifying chorus events and analyzing the properties of the chorus emissions. The Polar PWI provided plasma wave data from 26 March 1996 to 16 September 1997 (or 540 days). [9] In our data analysis, we were careful to avoid known difficulties with the PWI data set. During intervals of roughly 1 h or less centered on the lowest L shell values of the Polar orbit (twice per orbit), almost all PWI data taken with the electric and loop antennas are contaminated with interference and crosstalk. This contamination resulted from electric field antenna preamplifier oscillations in regions of high plasma density and interference from the EFI. Polar was generally inside the plasmasphere, where chorus is not likely to be observed, when these oscillations occurred so eliminating these intervals should not adversely affect our statistics. These oscillations can be clearly seen in the SFR electric field spectrograms at, or slightly below, the upper hybrid resonance frequency and its harmonics. The PWI electric field measurements taken during the approximate 5 10 min periods near the end of lengthy eclipses are also considered uninterpretable due to abnormal behavior of the EFI electric antennas, so we did not include data from periods when Polar was eclipsed. The behavior during eclipses may have been caused by a combination of the preamplifier oscillations and the effects of spacecraft charging [e.g., Laakso, 2002] on the antennas when they are in shadow. We also avoided intervals when the Polar Plasma Source Instrument (PSI) was operating, as interactions with PSI caused enhanced background noise levels and interference below 40 khz. [10] To insure that only chorus emissions were included in our study, we identified intervals when chorus emissions were observed by visually inspecting the PWI SFR spectrograms from 26 March 1996 to 16 September The PWI SFR experiment obtained both electric and magnetic field measurements over the frequency range 24 Hz to 808 khz. For logarithmically spaced frequencies, the PWI SFR took about 33 s to sweep through the frequency range in which chorus is typically observed (200 Hz to 12.5 khz). [11] Many previous studies of chorus and electron acceleration have used fixed frequency ranges for the lower ( f ce /10 < f < f ce /2) and upper chorus emission bands ( f ce /2 < f < f ce ) with amplitude thresholds to automatically distin- 2of17

3 guish chorus from other waves [e.g., Meredith et al., 2002; Li et al., 2009; Pokhotelov et al., 2008]. Chorus is believed to be generated near the equatorial plane, so at the high latitudes reached by Polar it may not be appropriate to use the local electron cyclotron frequency to distinguish these waves from other emissions. While it is possible to project the local cyclotron frequency down to the equator using a dipole field or other magnetic field model, we preferred to visually screen the PWI SFR data for evidence of chorus as there are uncertainties in the field line mapping. When chorus was observed by the PWI SFR, we recorded the lower frequency limit of these emissions. We will discuss the behavior of the lower chorus frequency limit and uncertainties in determining the equatorial electron cyclotron frequency further in section 3. [12] When available, data from the PWI WBR were also used to help distinguish chorus emissions from other types of waves. Polar data do not always show a clear, distinct plasmapause, particularly during times when the spacecraft is skimming along this boundary. This can make it difficult to determine whether or not the waves observed by the SFR are chorus or plasmaspheric hiss. Like chorus, plasmaspheric hiss is also a whistler mode emission, but hiss tends to be more broadband, has less fine structure than chorus, and may be involved in the loss of radiation belt electrons [e.g., Li et al., 2007; Santolík et al., 2001; Meredith et al., 2006]. Recent studies suggest that in some regions, chorus may leak into the plasmasphere and evolve into hiss [Chum and Santolík, 2005; Bortnik et al., 2008, 2009]. In addition, structureless, hiss like emissions within the typical chorus frequency range are often observed outside of the plasmasphere. Close to midnight, the highly elliptical nature of Polar s orbit often caused the spacecraft to leave the plasmasphere at high latitudes and enter directly into the auroral zone. While chorus is expected to be observed just outside the plasmasphere near the equator, the waves observed by Polar near the plasmapause on the nightside often appeared to be generated locally in the auroral zone. Close to noon, chorus emissions can propagate to high latitudes and become confused with whistler mode and magnetosonic waves generated locally in the cusp region. These features of the waves observed in the inner magnetosphere sometimes make it difficult to identify chorus from the SFR alone, so we consulted the high resolution WBR data to confirm the presence of discrete chorus elements whenever possible. As the actual frequencies of chorus emissions may vary somewhat and the lower frequency of chorus is not always exactly f ce /10, our approach has the advantage that it separates chorus from other types of waves found in the inner magnetosphere Magnetic Field Mapping and L* Calculation [13] The standard dipole L shell L ¼ R cos 2 where L is the dipole magnetic latitude and R is the radial distance measured from the center of the Earth, is often used to map spacecraft locations along the magnetic field to determine the radial distance away from the Earth at which a given field line crosses the equatorial plane. However, L ð1þ may not be the best parameter to use in this study due to the high latitudes reached by Polar and the compression of the Earth s magnetic field on the dayside. A parameter that may more accurately reflect the structure of the magnetospheric magnetic fields and behavior of the trapped particles in the radiation belts is L* [Roederer, 1970]. The definition of this parameter is L* ¼ 2k 0 R E where k 0 is the Earth s dipole moment, R E is the radius of the Earth, and F is the magnetic flux enclosed by a drift shell [Roederer, 1970]. In a dipole magnetic field L* is the distance from the center of the Earth to the equatorial point of a given field line, in units of Earth radii. All pitch angles have the same L* for a given point in space. To determine L*, we used the ONERA DESP library version 4.2 (available at and the Tsyganenko 1989c magnetic field model [Tsyganenko, 1989] with real Kp values downloaded from the World Data Center for Geomagnetism in Kyoto, Japan. The Tsyganenko 1989c magnetic field model was selected because the only geomagnetic parameter required as an input to this model was the Kp index, which is available throughout the entire time period of interest. Later magnetic field models require additional inputs which were often unavailable during 1996 and Geomagnetic Indices and Solar Wind Parameters [14] The auroral electrojet (AE) index may be relevant to the physical processes studied in this paper and has been used in earlier chorus studies [e.g., Meredith et al., 2001, 2002; Li et al., 2009]. Unfortunately, the AE indices have not been calculated for 1996 by the World Data Center for Geomagnetism and only the quick look AE indices, which are intended just for space weather monitoring purposes, are available in Because AE was not available throughout the entire time period of interest, we instead attempted to look for correlations between chorus occurrence rates and the Kp and Dst indices. The planetary Kp index is an activity level rating from 0 to 9 for 3 h intervals based upon ground magnetometer data. The Dst indices are derived from a network of near equatorial geomagnetic observatories that measures the low latitude horizontal magnetic variations due to the globally symmetrical equatorial electrojet, also known as the ring current. The equatorial ring current causes a global depression in the H component of the magnetic field during the main phase of geomagnetic storms. The final Dst values, in units of nt, are available for all of the time intervals of interest in this paper. [15] We also attempted to explore possible relations between chorus and upstream solar wind conditions using the OMNI 2 data set [see King and Papitashvili, 2005] provided by the NASA Goddard Space Flight Center Space Physics Data Facility (SPDF). The OMNI 2 data set contains multispacecraft solar wind magnetic field and plasma data. It includes data from the ISEE 3, Wind, and ACE spacecraft, which are often located about an hour upstream of the magnetosphere, as the solar wind flows. OMNI 2 also includes data from 15 geocentric spacecraft located closer to Earth, such as IMP 8. In constructing the OMNI 2 data set, ð2þ 3of17

4 the SPDF took into consideration studies of solar wind structures [e.g., Richardson and Paularena, 1998; Weimer et al., 2002] and time shifted the data to Earth, when appropriate. To merge upstream ISEE 3, Wind, and ACE data into OMNI 2, it was assumed that solar wind variation phase fronts are planar, normal to the ecliptic plane, and intersect the ecliptic plane along a line exactly halfway between the ideal Parker spiral and a normal to the Earth Sun line (the Y GSE axis). The spacecraft locations and the measured solar wind flow speeds in the data sets being shifted were used with the above assumptions to compute the appropriate time delays. Data from the three upstream spacecraft were time shifted to Earth at 1 5 min resolution and then average values were computed using the data points whose shifted time tags fell within a given hour. For example, the first OMNI 2 value of the day is the average of the data points with shifted time tags between 0000 and 0100 UT. During the time periods of interest in , the OMNI 2 solar wind parameters consisted mainly of data from either Wind or IMP 8. When multiple data sources are available, OMNI 2 gives inclusion priority to the more timecontinuous data source. Although IMP 8 was closer to Earth, Wind data were given priority for inclusion in OMNI 2 when they were available because Wind data had fewer gaps than IMP 8 data. The merged OMNI 2 data set gives the computed hourly averages at Earth of parameters such as the GSM B Z component of the solar wind magnetic field, solar wind dynamic pressure, and solar wind bulk speed, all of which are useful in assessing coupling between the solar wind and magnetosphere. 3. Chorus Location Statistics [16] Nearly 13,000 h of PWI data were recorded between 26 March 1996 and 16 September Out of the nearly 8602 h of data recorded between magnetic latitudes of 70 and 70, about 407 h of data had chorus emissions. The highest southern magnetic latitude at which chorus was observed was 53, but there were very few cases of chorus (about 50 min) in the Southern Hemisphere at latitudes higher than 30. These cases represent only about 0.2% of the chorus emissions observed by the Polar PWI during During , Polar s perigee was located in the Southern Hemisphere, so Polar generally entered the plasmasphere before it crossed the magnetic equator as it traveled from the Northern Hemisphere to Southern Hemisphere. The maximum magnetic latitude at which chorus was observed in the Northern Hemisphere was 63. [17] The manner in which we selected chorus events allowed us to examine the ratio f/f ce using the observed lower frequency limit f of the chorus emissions obtained from visual inspection of the SFR spectrograms and the electron cyclotron frequency f ce calculated using the equatorial magnetic field from the T89 magnetic field model. For 93% of our chorus observations (378 h), the ratio of the lower chorus frequency limit to the equatorial electron cyclotron frequency fell within the typical lower band chorus range 0.1 f/f ce 0.5, as expected. For about 3% of our chorus observations (12 h), the ratio of the lower chorus frequency limit to the equatorial cyclotron frequency fell within the typical upper band chorus range 0.5 f/f ce 1.0. None of the chorus observations made between magnetic latitudes of 5 to 5 degrees had f/f ce ratios within the upper band range. Most of the chorus observations which had 0.5 f/f ce 1.0 occurred at latitudes above 20, and many of these observations occurred near noon and midnight. These observations could indicate times when only upper band chorus was observed. However, this appears unlikely, as inspection of the Polar WBR measurements showed that the upper band tends to be bursty, often appearing and disappearing intermittently throughout the same radiation belt pass, frequently embedded in structureless, hiss like emissions. Another possibility is that there could have been an off equatorial source for some of these waves [e.g., Tsurutani et al., 2009], but there may also have been problems with the field line mappings to the equator at high latitudes near noon and midnight due to the structure of the magnetosphere in these regions. [18] We suspect that there were errors in the equatorial cyclotron frequency calculated from the model for the remaining 4% of our chorus observations (17 h). For about 15 h of data, spread over the entire range of magnetic latitudes and local times covered by Polar, the estimated f/f ce ratio was less than 0.1. However, the average f/f ce value for these data points was 0.08, which is quite close to the typical lower limit of the chorus frequency range. The calculation of the equatorial cyclotron frequency may have been slightly inaccurate for these times, but it is also possible that these values simply represent occasional outliers of the typical chorus frequency range. For about 2 h of data, the calculation of the equatorial cyclotron frequency failed because the estimated f/f ce ratio was greater than 1. These data points were located at magnetic latitudes greater than 24, mainly near noon and midnight. Problems with the field line mapping to the equatorial plane might be expected in these regions, owing to the dayside compression of the magnetosphere and the stretching of the magnetotail on the nightside. [19] On the basis of our event selection using the observed lower frequency limit f of the chorus emissions obtained from visual inspection of the SFR spectrograms and the above examination of f/f ce, the chorus location statistics we will present in this paper represent the occurrence probability of lower band chorus, although the upper band may have been present in some cases. To establish the character of Polar s orbit and the regions it samples, we will first examine the orbit of the Polar spacecraft and the occurrence rate of chorus observed by the Polar PWI as a function of Z SM and radial distance in the plane of the magnetic equator. In solar magnetospheric (SM) coordinates, the Z axis is aligned along the Earth s magnetic dipole axis and gives the distance above the plane of the magnetic equator. In later sections, we will explore the regions in which Polar observed chorus using polar coordinate maps of R versus MLT, L versus MLT, and L* versus MLT to gain a better understanding of the field line mappings in the regions sampled by Polar along its orbit and how chorus may be related to particle motions within the large scale magnetic field structure of the regions sampled by Polar Z SM Position and Radial Distance [20] Figure 1 shows the percentage of time that chorus was observed by the Polar spacecraft from March 1996 to September 1997 and the number of minutes spent by Polar between ±70 magnetic latitude in four local time sectors 4of17

5 Figure 1 5of17

6 using SM coordinates in units of Earth radii (R E ). Following the example of Meredith et al. [2002], we plotted the chorus occurrence rate using Polar s Z SM position and radial distance R in the plane of the magnetic equator. In SM coordinates, the Z axis is aligned along the Earth s magnetic dipole axis, the X and Y axes lie in the plane of the magnetic equator, and the Y axis is oriented perpendicular to the Earth Sun line with the positive direction pointing toward dusk. [21] Figure 1a (top) shows the percentage of time chorus was observed in the midnight sector (2100 to 0300 MLT) for 0.5 R E bins of Z SM and R. Figure 1a (bottom) shows the number of minutes spent by the Polar spacecraft in the same 0.5 R E bins of Z SM and R in the midnight sector. The white areas in Figure 1 and all subsequent figures indicate regions where no data were available. To facilitate comparisons between figures, the color scales in Figure 1 and all subsequent figures start from 0% or 0 min of data. The black and purple colored areas therefore represent very small, but nonzero numbers of minutes or very small, but nonzero percentages of time when chorus was observed. Following the example of Meredith et al. [2003], we have overplotted black lines to indicate dipole field lines for L shells (equation (1)) L =2,4,6,and8R E. The diagonal lines indicate lines of constant magnetic latitude for ±10, 20, 30, 40, 50, 60, 70, and 80. [22] Our event database included 73 h of chorus observations between 2100 and 0300 MLT, which represents about 18% of our chorus observations. Polar spent about 2242 h between 2100 and 0300 MLT, so chorus was observed about 3% of the time Polar spent in this sector. Although the Polar spacecraft spends a great deal of the time in the midnight sector more than 4 R E above the plane of the magnetic equator, chorus was most often observed less than 2R E above the equatorial plane in this local time sector. As Figure 1a illustrates, the orbit of Polar during was such that the spacecraft was located at small radial distances from Earth in the Southern Hemisphere. Because Polar was located so close to Earth, it was often deep inside the plasmasphere during perigee passes through the Southern Hemisphere. As a result, we do not have many chorus observations for Z SM < 0 in any of the four local time sectors. [23] Figure 1b shows the percentage of time chorus was observed in the dawn sector (0300 to 0900 MLT) and number of minutes spent by Polar in 0.5 R E bins of Z SM and R in the same format as Figure 1a. Polar made about 95 h of chorus observations between 0300 and 0900 MLT, which represents about 23% of our chorus observations. Polar spent about 1981 h between 0300 and 0900 MLT, so chorus was observed about 5% of the time Polar spent in this sector. As in the midnight sector, Polar spends a great deal of its time in this sector more than 4 R E above the equatorial plane, but chorus was mainly observed close to the equatorial plane. However, in the dawn sector, some chorus was observed up to 5 R E above the equatorial plane, even though the distribution was peaked around 1.5 to 2.5 R E above the equatorial plane. [24] Figure 1c shows the percentage of time chorus was observed in the noon sector (0900 to 1500 MLT) and number of minutes spent by Polar in the same 0.5 R E bins of Z SM and R used in Figure 1a. We have about 200 h of chorus observations between 0900 and 1500 MLT, which represents about 49% of our chorus observations. Polar spent 2173 h between 0900 and 1500 MLT, so chorus was observed about 9% of the time Polar spent in this sector. Chorus was observed most often in this local time sector. As in the midnight and dawn sectors, Polar spends a great deal of time more than 4 R E above the equatorial plane. However, in the noon sector, chorus was observed up to 7 R E above the equatorial plane. This is consistent with past studies showing that the extent of the region in which chorus is observed on the dayside is much larger [Bortnik et al., 2007; Sigsbee et al., 2008]. Although chorus was observed for Z SM locations up to 7 R E in this local time sector, the distribution of chorus observations still had a peak for Z SM < 4R E, close to the equatorial plane. [25] Figure 1d shows the percentage of time chorus was observed in the dusk sector (1500 to 2100 MLT) and number of minutes spent by Polar in the same 0.5 R E bins of Z SM and R as before. We have about 39 h of chorus observations between 1500 and 2100 MLT, which represents 10% of our chorus observations. Polar spent about 2206 h in between 1500 and 2100 MLT, so chorus was only observed about 2% of the time Polar spent in this sector. Chorus was observed least often in this local time sector. As in the midnight, dawn, and noon sectors, Polar spends a great deal of the time more than 4 R E above the equatorial plane. The distribution of Z SM locations where chorus was observed in the dusk sector is surprisingly similar to the distribution of the locations where chorus was observed near dawn. [26] The local time distribution of chorus presented in this paper may be different from what many readers may believe the expected distribution of chorus occurrence to be. In an early paper on ground based whistler mode wave observations [Storey, 1953], the term dawn chorus was used to describe the type of discrete emissions discussed in our paper. Although Storey [1953] reported that chorus emissions varied in strength daily, with a maximum near 0600, later satellite data studies [e.g., Russell et al., 1969; Dunckel and Helliwell, 1969; Tsurutani and Smith, 1977; Meredith et al., 2009] have shown that chorus can be observed at all magnetic local times, including near dusk. Although we found a similar spatial distribution of chorus observations near dusk and dawn, we can see from Figure 1d that the percentage of time when chorus was observed in the dusk sector is much less than in the dawn sector. We will discuss the local time variation in chorus occurrence further in sections 3.2 and 3.3. Figure 1. The percentage of time that chorus was observed by the Polar Plasma Wave Instrument (PWI) and the number of minutes spent by the Polar spacecraft from March 1996 to September 1997 in four local time sectors. (a) At the top is shown the percentage of time chorus was observed in the midnight sector for 0.5 R E bins of the Z SM location and the radial distance R in the plane of the magnetic equator. At the bottom is shown the number of minutes spent by the spacecraft in the same bins of Z SM and R. Black lines indicate constant magnetic latitude and L shells. (b) Same format as Figure 1a for the dawn sector. (c) Same format as Figure 1a for the noon sector. (c) Same format as Figure 1a for the dusk sector. 6of17

7 dayside outer zone chorus in minimum B pockets located at above the magnetic equator. These pockets are created by the solar wind compression of the dayside magnetosphere and the field magnitudes inside them may be less than at the equator. The character of the waves observed by Polar in this region is often slightly different from those observed near the equator. The elements often do not have the typical rising tone structure of chorus. Both rising and falling tones, as well as elements with a hook like structure can be found here, often embedded in a background of structureless, hiss like emissions. However, these waves do appear to be whistler mode [Santolík et al., 2006] so they may still be of dynamical importance to the behavior of high energy electrons in this region. Figure 2. (top) Percentage of time chorus was observed by Polar PWI as a function of radial distance R in Earth radii (R E ) and magnetic local time in bins of 0.5 R E in radial distance and 1 h in magnetic local time for magnetic latitudes between 70 and 70. (bottom) Orbital coverage by Polar during the entire time interval when Polar PWI data were available. [27] The difference in the extent of the regions above the equatorial plane where chorus was observed in the four local time sectors is due to the differences in the structure of the magnetosphere in the midnight, dawn, and noon local time sectors. On the nightside, near midnight, the Polar spacecraft often encountered the auroral zone shortly after leaving the plasmasphere. The waves observed by Polar on the nightside tend to be dominated by a variety of waves from the auroral zone, but chorus is still observed occasionally on the nightside. The small area in which chorus was observed by Polar on the nightside is due both to the structure of the magnetosphere in this local time sector and the behavior of Polar s orbit. [28] On the dayside, waves showing chorus like element structures are often observed heading into the region near the cusp at high latitudes. It is not always clear whether or not these waves are chorus emissions that have propagated from an equatorial source or if the source region is located at higher latitudes. Some of the chorus like emissions observed by the Polar spacecraft on the dayside at high latitudes may have a source near the magnetic equator [Menietti et al., 2009]. Other studies [e.g., Tsurutani and Smith, 1977; Pickett et al., 2001; Vaivads et al., 2007; Tsurutani et al., 2009] indicate there may be a high latitude source for 3.2. Variation With R, Magnetic Local Time, and Magnetic Latitude [29] Figure 2 (top) shows a polar coordinate plot of the overall percentage of time chorus was observed by Polar PWI as a function of radial distance R in Earth radii (R E ) and magnetic local time. Figure 2 (bottom) shows the orbital coverage by Polar during the entire time interval when Polar PWI data were available (about 540 days) in polar coordinates as a function of radial distance R in R E and magnetic local time. The data have been organized into bins of 0.5 R E in radial distance and 1 h in magnetic local time for magnetic latitudes between 70 and 70. Note that in Figure 2, R is the total radial distance of Polar from the center of the Earth, not the radial distance in the plane of the magnetic equator. The minimum radial distance at which chorus was observed was about 2.2 R E, and the maximum radial distance at which chorus was observed was about 7.8 R E. [30] The plane of the Polar spacecraft s orbit underwent a virtual rotation through approximately 24 h of MLT in 1 year, owing to the orbit of the Earth around the Sun. Polar PWI data were only available for a little over 17 months (1.417 years), so the spacecraft covered a region of about 10 h of MLT twice and the other 14 h of MLT only once during the time interval we considered. This partially explains the slight bias in orbital coverage toward the duskside of the magnetosphere from 12 noon to midnight visible in Figure 2. However, the apsidal precession rate of Polar s orbit also was a factor. The bias in orbital coverage toward the duskside of the magnetosphere is much more apparent at radial distances greater than 6 R E because the apogee of the Polar spacecraft s orbit moved toward the equator at a rate of about 15 per year [Acuña et al., 1995]. Because Polar s apogee was no longer located directly over the north pole, the spacecraft spent noticeably more time at larger radial distances (>6 R E ) on the duskside than on the dawnside. [31] In spite of the slight orbital bias of the Polar spacecraft toward the duskside of the magnetosphere, Figure 2 shows a typical distribution of chorus observations [e.g., Tsurutani and Smith, 1977; Meredith et al., 2003]. The distribution of chorus observed by the Polar spacecraft is peaked in a region starting from about 21 h magnetic local time, extending around the dawn flank in the direction of increasing local times to 15 h magnetic local time on the dayside. Chorus was observed most often on the day side from about 9 h magnetic local time to 13 h magnetic local time. In the magnetic local time region between 9 and 13 h, chorus was also observed at greater radial distances than it 7of17

8 was at other magnetic local times. This is in part due to the large number of high latitude dayside chorus events observed by Polar. [32] Figure 3 shows polar coordinate plots of the distribution of chorus observed by Polar PWI in four different magnetic latitude ranges as a function of 0.5 R E bins in radial distanceand1hbinsinmagneticlocaltime.figures3a and 3b show that the magnetic local time distribution of chorus from 15 to 15 magnetic latitude and 15 to 30 magnetic latitude reflects the overall distribution of chorus shown in Figure 2. However, in Figure 3c, we see that between 30 and 45 magnetic latitude, the distribution of chorus is beginning to shift more toward the dayside magnetosphere. In Figure 3d, we see that between 45 and 70 magnetic latitude, chorus is only observed in a narrow region between 6 h magnetic local time to 15 h magnetic local time. This appears to be due to both the nature of Polar s orbit and the structure of the magnetosphere, as discussed in the previous section. These high latitude dayside chorus events represent only a small percentage of the chorus emissions observed by Polar. However, as Figure 3 illustrates, chorus is not observed exclusively in the region immediately surrounding the equatorial plane. The waves observed by Polar away from the equatorial plane may simply be chorus generated near the magnetic equator that has propagated to higher magnetic latitudes. However, these waves could also have been generated by other means, such as anisotropic electrons that have drifted from the midnight sector to dayside minimum B pockets [Tsurutani et al., 2009] or by the effects of solar wind pressure fluctuations on an existing population of energetic electrons in the dayside outer zone [Lauben et al., 1998; Tsurutani et al., 2009] Dependence Upon L, L*, Magnetic Local Time, and Magnetic Latitude [33] Several past studies of chorus have examined the chorus occurrence rate as a function of L and MLT [Russell et al., 1969; Burtis and Helliwell, 1976; Tsurutani and Smith, 1977; Li et al., 2009]. Figure 4a (top) shows a polar coordinate plot of the overall percentage of time chorus was observed by Polar PWI as a function of L and MLT. Figure 4a (bottom) shows the orbital coverage by Polar during the entire time interval when Polar PWI data were available in polar coordinates as a function of L (equation (1)) and magnetic local time. The data have been organized into bins of 0.5 R E in L and 1 h in magnetic local time for magnetic latitudes between 70 and 70. As one can see from Figure 4a, the standard L shell mapping to the radial distance away from the Earth in the equatorial plane may not be accurate when Polar is located at high latitudes near noon. Because the standard L shell does not take into account the compression of the magnetic field in the dayside magnetosphere, when Polar is located at high latitudes near noon the L shell values are unrealistically high. However, in spite of the problems with the standard L shell mapping at high latitudes, the distribution of Polar PWI chorus observations as a function of L and MLT is similar to that of other studies. As Figure 4a shows, the chorus occurrence rate from Polar PWI data has a peak near noon on the dayside in agreement with earlier results [Burtis and Helliwell, 1976; Tsurutani and Smith, 1977; Li et al., 2009]. The region of the most frequent chorus occurrence in Figure 4 features an outward spiral from low L values on the nightside to higher L values on the dayside, similar to what was reported by Burtis and Helliwell [1976]. [34] While previous studies have used L to examine the chorus occurrence rate, a better parameter to use is L* (equation (2)), which attempts to model the drift shells of the trapped particles in the magnetosphere. The value of L* we used was computed using a magnetic field model that takes into account the compression of the magnetosphere on the day side, as we discussed in section 2.2. Figure 4b (top) shows a polar coordinate plot of the overall percentage of time chorus was observed by Polar PWI as a function of L* and MLT. Figure 4b (bottom) shows the orbital coverage by Polar during the entire time interval when Polar PWI data were available in polar coordinates as a function of L* and magnetic local time. As before, the data have been organized into bins of 0.5 R E in L* and 1 h in magnetic local time for magnetic latitudes between 70 and 70. The outward spiral of the region of most frequent chorus occurrence from the nightside to the dayside [Burtis and Helliwell, 1976] is more readily visible in Figure 4b than it was in Figure 4a, thanks to the improved mapping provided by L*. [35] Figure 5 shows polar coordinate plots of the percentage of time chorus was observed by Polar PWI and the orbital coverage by Polar as a function of L* and magnetic local time for the same magnetic latitude ranges used in Figure 3. As before, the data have been organized into bins of 0.5 R E in L* and 1 h in magnetic local time. The amount of data shown for magnetic latitudes between 45 and 70 in Figure 5d is greatly reduced from the amount of data shown in Figure 3d for the same magnetic latitude range. L* could not always be calculated in the region near noon at high latitudes because the drift shells were not closed. Although Polar spent 4255 h between 45 and 70 magnetic latitude, L* could only be calculated for about 36 h of data in this latitude range. Chorus was observed for about 21 h between 45 and 70 latitude, but L* could only be calculated for about 3.5 h. In spite of the greatly reduced amount of data shown for high latitudes, Figures 4b and 5 clearly illustrate that the L* parameter maps Polar s location on the dayside at high latitudes to locations in the equatorial plane at radial distances much closer to Earth than the standard L shell. [36] Although we used a different approach to selecting chorus events to that of Li et al. [2009] and other previously published studies, our results for the chorus occurrence probability were similar. Figure 5 shows that the high latitude chorus occurs mainly on the dayside, in agreement with other recent studies [Li et al., 2009; Tsurutani et al., 2009]. We also agree with the finding of Li et al. [2009] that nightside chorus is generally observed at lower latitudes than on the dayside. The distribution of chorus with L* and MLT observed by Polar in the midlatitude range from 15 to 30 magnetic latitude is similar to that found by Li et al. [2009] for the distribution of chorus observed by THEMIS with L and MLT for 10 to 25 magnetic latitude. 4. Dependence Upon Geomagnetic Indices and Upstream Parameters [37] The AE index is not available for 1996, so attempting to examine the dependence of chorus occurrence probabil- 8of17

9 Figure 3 9of17

10 Figure 4. (a) At the top is shown the percentage of time chorus was observed by Polar PWI as a function of 0.5 R E bins of L and 1 h bins in magnetic local time for magnetic latitudes between 70 and 70. At the bottom is shown the orbital coverage by Polar during the entire time interval when Polar PWI data were available using the same bins in L and magnetic local time. (b) The same format as Figure 4a, only now the data are plotted in 0.5 R E bins of L* and 1 h bins in magnetic local time. ities on AE would require us to abandon half of our data set and would leave us with too few events to obtain statistically significant results. We instead attempted to examine the chorus locations and occurrence probability as a function of the Kp and Dst indices. Unfortunately, we found that most of the chorus events in the Polar data set occurred during Kp < 3 so we were unable to obtain statistically meaningful correlations with this geomagnetic index for a variety of activity levels. As Kp is a 3 h index, this parameter may also be inappropriate for our data set. [38] We then attempted to look for correlations with the Dst index. The largest negative value of Dst reached during a time interval when chorus was observed by Polar was only 70 nt. Only about 1/3 of the Polar chorus observations occurred for Dst < 20 nt. About 2/3 of the Polar chorus observations occurred during times when Dst 20 nt and about 18% of these observations occurred for Dst > 0 nt. As a result of the very low activity and limited range of Dst during the times when chorus was observed by Polar, we were also unable to obtain statistically meaningful correlations with the Dst index. Like Li et al. [2009], we found that chorus does occur during relatively quiet times, even though many recent studies have focused upon chorus during geomagnetic storms. The Polar data set includes a high number of dayside chorus observations during quiet times at high L* values, which is consistent with the idea proposed by Li et al. [2009] that enhanced electron anisotropies in this region provide favorable conditions for chorus generation even during low geomagnetic activity. [39] We used the OMNI 2 data set to examine how the locations and occurrence rate of chorus emissions might depend upon upstream solar wind parameters. Figure 6a Figure 3. (a) At the top is shown the percentage of time chorus was observed by Polar PWI as a function of 0.5 R E bins of radial distance and 1 h bins in magnetic local time for magnetic latitudes between 15 and 15. At the bottom is shown the orbital coverage by Polar during the entire time interval when Polar PWI data were available for magnetic latitudes between 15 and 15 using the same bins in R and magnetic local time. (b) The same format as Figure 3a for magnetic latitudes between 15 and 30, (c) for magnetic latitudes between 30 and 45, and (d) for magnetic latitudes between 45 and of 17

11 Figure 5 11 of 17

12 Figure 6. (a) At the top is shown the percentage of time chorus was observed by Polar PWI as a function of radial distance R in Earth radii (R E ) and magnetic local time for bins of 0.5 R E in R and 1 h in magnetic local time for magnetic latitudes between 70 and 70 during intervals when 250 km/s < V SW < 450 km/s. At the bottom is shown the orbital coverage by Polar during the entire time interval when Polar PWI data were available and when 250 km/s < V SW < 450 km/s. (b) The same as Figure 6a for intervals when 450 km/s < V SW < 750 km/s. (top) shows the percentage of time chorus was observed for hourly averaged solar wind bulk speeds V SW between 250 and 450 km/s as a function of radial distance R in R E and magnetic local time. Figure 6a (bottom) shows the orbital coverage by Polar as a function of radial distance R in R E and magnetic local time for intervals when the hourly averaged solar wind speed was between 250 and 450 km/s. The data have been organized into bins of 0.5 R E in radial distance and 1 h in magnetic local time for magnetic latitudes between 70 and 70. Figure 6b shows the percentage of time chorus was observed and the orbital coverage for solar wind speeds between 450 and 750 km/s in the same format as Figure 6a. Although the overall distribution of chorus observations does not appear to differ much between Figures 6a and 6b, it does appear that chorus tends to be observed significantly (1.5 2 times) more often for higher solar wind speeds. [40] To investigate the possible association of chorus with electron microbursts, we compared Figure 6 with the results of O Brien et al. [2003]. O Brien et al. examined the occurrence of low altitude MeV electron microbursts observed by the SAMPEX satellite from 1996 to 2001 as a function of L, MLT, and solar wind speed. Analysis shown by O Brien et al. [2003] indicated that >1 MeV microbursts occur over a wider range of MLT for lower solar wind speeds than they do for higher solar wind speeds. For high solar wind speeds (500 < V SW < 600 km/s), the microburst occurrence frequency appeared to be strongly peaked near 30% between Figure 5. (a) At the top is shown the percentage of time chorus was observed by Polar PWI as a function of 0.5 R E bins of L* and 1 h bins in magnetic local time for magnetic latitudes between 15 and 15. At the bottom is shown the orbital coverage by Polar during the entire time interval when Polar PWI data were available for magnetic latitudes between 15 and 15 using the same bins in L* and magnetic local time. (b) The same format as Figure 5a for magnetic latitudes between 15 and 30, (c) for magnetic latitudes between 30 and 45, and (d) for magnetic latitudes between 45 and of 17

13 0600 and 1200 MLT, and 4 < L <7R E. For lower solar wind speeds (V SW < 400 km/s), microbursts appeared to be found between 4 < L <7R E at nearly all MLT, but had a peak occurrence of only about 3% on the nightside. O Brien et al. [2003] found that electron acceleration at low L shells was associated with ULF wave activity and MeV microbursts and assumed that it was therefore also associated with chorus activity. However, they did not have access to in situ plasma wave observations, so they could not confirm this. For low solar wind speeds, we found that chorus occurred between 20% and 30% of the time Polar spent on the dayside, which is an order of magnitude higher than the peak microburst occurrence frequency found by O Brien et al. [2003] for similar solar wind speeds. For high solar wind speeds, we found a peak occurrence frequency of around 40%, which is once again higher than the peak microburst occurrence frequency found by O Brien et al. [2003] for similar solar wind speeds. For both high and low solar wind speeds, the chorus occurrence rate was peaked on the dayside in the Polar data set. This result does not agree very well with the MLT distribution of MeV microbursts for high and low solar wind speeds found by O Brien et al. [2003]. However, this is consistent with an early study of chorus and > 40 kev electrons from the Injun 3 satellite [Oliven and Gurnett, 1968] which showed that microbursts are always accompanied by chorus, but not all chorus observations are associated with microbursts. Oliven and Gurnett showed an MLT distribution of chorus emissions from Injun 3 that appears quite similar to ours. They found that the distribution of microbursts fell within the region of maximum chorus occurrence but that the region of microburst occurrence was more restricted in local time. [41] We also attempted to examine how chorus occurrence rates and locations might depend upon the solar wind dynamic pressure. In past studies of Polar data [Lauben et al., 1998] there has been evidence for a correlation between solar wind pressure pulses and dayside chorus generation. However, we found little statistical variation with dynamic pressure in the Polar data set. It appears likely that while the solar wind speed may be more or less steady on the time scales of the hourly OMNI 2 data set, density perturbations associated with pressure pulses are short lived [see, e.g., Fox et al., 1998] and tend to be lost in the hourly averages. As a result, most of the Polar chorus events are associated with a range of average dynamic pressures in the OMNI 2 data set that has too little variation to obtain any meaningful statistical results. [42] The next upstream solar wind parameter we examined using the OMNI 2 data set was the direction of the solar wind magnetic field B Z GSM component. We had hoped that we might see some kind of correlation with the solar wind magnetic field direction and the occurrence of the high latitude events on the dayside. The distribution of chorus observation locations did appear to shift slightly toward the postnoon and dusk regions for B Z GSM > 0, but the number of Polar chorus observations in the dusk region is still quite low, even for these B Z GSM > 0 conditions. The shift toward dusk for B Z GSM > 0 is probably not a statistically significant result, so we do not show it here. Also, much of the geomagnetic activity during solar minimum appears to be associated with oscillating B Z GSM magnetic field directions in the solar wind [Tsurutani et al., 2006] which may not appear in an hourly averaged solar wind magnetic field data set. [43] We found a better correlation between chorus observations and the magnitude B SW of the solar wind magnetic field. Figure 7a (top) shows the percentage of time chorus was observed for hourly averaged solar wind B SW <5nT as a function of radial distance R in R E and magnetic local time. Figure 7a (bottom) shows the orbital coverage by Polar as a function of radial distance R in R E and magnetic local time for intervals when B SW < 5 nt. Figure 7b shows the percentage of time chorus was observed and the orbital coverage for solar wind B SW > 5 nt in the same format as Figure 7a. As shown in Figure 7, we were slightly more likely to observe chorus in the Polar PWI data set when B SW > 5 nt than for times when B SW < 5 nt. The effects of increasing solar wind magnetic field strength on the chorus occurrence probability seem to be more pronounced on the dayside between radial distances of 4 and 6 R E than at other magnetic local times. It is possible that conditions are more favorable for the creation of minimum B pockets in which waves may be generated for B SW > 5 nt. However, this may also indicate more efficient coupling in general between the processes responsible for dayside chorus and upstream conditions during times when the interplanetary magnetic field is more intense. 5. Discussion [44] Other recent studies have examined the average chorus amplitudes as a function of L and magnetic local time. While we cannot compare our results directly to these studies, we can compare the shape of the regions in which these studies found the largest chorus amplitudes to the shape of the regions in which we found the highest chorus occurrence probability. A survey of chorus observations from the Cluster STAFF SA experiment found that the largest lower band chorus amplitudes occurred on the dayside and that intense chorus was observed at larger radial distances on the dayside than on the nightside [Pokhotelov et al., 2008]. We found that dayside chorus occurs quite often and can reach larger L values than nightside chorus, which compares favorably with their result. We also found that the chorus occurrence probability is greatest near dawn and on the dayside and smallest near dusk. However, the distribution of the peak amplitude regions for lower band chorus from Pokhotelov et al. [2008] is much more symmetrical with magnetic local time than our chorus occurrence distribution. The data shown by Pokhotelov et al. [2008] indicate that the largest amplitude chorus is found in the postnoon region, while we generally found the largest chorus occurrence rates right around noon or in the prenoon sector. The smaller frequency range of the STAFF SA instrument, differences in the orbit of Cluster and Polar, the usage of L instead of L*, and the usage of fixed frequency ranges and amplitude thresholds to identify chorus by Pokhotelov et al. [2008] all are factors in the difference between our results and theirs. [45] On the other hand, the distributions of the average chorus amplitudes from THEMIS [Li et al., 2009] and CRRES [Meredith et al., 2001; Bortnik et al., 2007] show the same kind of dawn dusk asymmetry in the regions of large wave amplitudes that we see in our chorus occurrence 13 of 17