The electromagnetic ion cyclotron (EMIC) waves fall into the ultralow frequency range (0.1–5 Hz) commonly referred to as Pc1–Pc2 pulsations. The EMIC waves are known for precipitating the relativistic electrons and ring current ions through pitch angle scattering (Thorne and Kennel, 1971; Summers and Thorne, 2003; Sandanger et al., 2007) and are responsible for ion heating (Horne and Thorne, 1997; Yuan et al., 2014) and proton auroras (Sakaguchi et al., 2015; Ozaki et al., 2021). These waves are generated in the equatorial regions over a wide range of L-shell due to the anisotropic distribution of ring current ions (Cornwall, 1965; Anderson et al., 1996). Different aspects of EMIC waves have been examined using both ground and satellite data to gather more information about their occurrence, characteristics, MLT-azimuthal-radial extent, etc. (Usanova et al., 2008; Posch et al., 2010; Usanova et al., 2013; Mann et al., 2014; Allen et al., 2015; Engebretson et al., 2015; Blum et al., 2017; Yu et al., 2017; Wang et al., 2021). The gyrofrequency of the ion responsible for the generation of the wave determines the frequency band (H ⁺ or He ⁺) of these waves. The global distribution of EMIC wave occurrence has been widely explored by several past satellite missions. The DE-1 spacecraft surveyed the inner magnetosphere (3 > L-shell > 5 ) and reported the dusk sector is the preferential zone for the generation of the EMIC waves (Erlandson and Ukhorskiy, 2001). L-shell is a distance of magnetic field line in the magnetic equatorial plane in units of Earth radii. By using AMPTE satellite data, Anderson et al. (1992) have shown that the EMIC waves are primarily observed at L > 7 in the afternoon sector, which is consistent with the recent results by Keika et al. (2013) and GOES spacecraft observations (Clausen et al., 2011). The CRESS spacecraft gave broader time coverage, and these observations indicated that 80% of the EMIC waves occur during 13–20 magnetic local time for L > 3 (Meredith et al., 2003). A recent study by Min et al. (2012) has used THEMIS spacecraft data and reported two peaks in the EMIC wave occurrence probability: one in the dawn sector (dominated by H band waves) and the other in the dusk (dominated by He band waves). All these scenarios are based on satellite observations.

After the generation, the EMIC waves propagate along geomagnetic field lines from their source in the equatorial region to the mid- and high-latitude ionosphere. A part of EMIC wave energy can get trapped in the F-region ionospheric duct and propagate horizontally to lower or higher latitudes. Thus, EMIC wave signatures can be observed on the ground as well. However, due to ducted propagation, we do not have any information about L-shell associated with their source locations. Unlike satellite observations, it is difficult to associate the EMIC wave signatures observed on the ground to a particular band (H ⁺ or He ⁺) due to its propagation through ionospheric ducts. However, the EMIC wave event observed on the ground is the manifestation of the EMIC wave generation in the magnetosphere at a particular L-shell. Thus, the ground EMIC wave observations are also widely used to examine the characteristics of these waves. The diurnal distribution has been studied using the ground-based EMIC wave observations from Sodankylāics geophysical observatory (L = 5.2) for the maxima and minima of the 21st solar cycle (Mursula et al., 1991, 1996). These authors also observed two peak structures in the occurrence pattern, but the morning peak was shown by structured and the afternoon–dusk peak by unstructured pulsations. A similar study from an observatory (Davis station, geomagnetic 74.7 ^o S  100.8 ^o E) located in the polar cusp showed a single peak around local noon in the occurrence of the EMIC waves (Morris and Cole, 1991). A recent study by Upadhyay et al. (2020) indicated that the occurrence of EMIC wave peaks around the afternoon–dusk sector. In all the above ground-based studies, not much attention was given to the frequencies of the EMIC waves. Particularly, the local time distribution of EMIC waves in different frequency ranges has not been explored well with the ground observations.

The EMIC waves are typically generated by the temperature anisotropy (T _⊥ > T _‖) of medium energy (1–100 keV) ring current protons (Kennel and Petschek, 1966). There can be multiple reasons for the enhancement in the anisotropy of the ions; one of the recognized sources is the energetic particle injections into the inner magnetosphere during storms and substorms (Liu et al., 2019). The geomagnetic storms and substorms enhance the earthward convection of ring current ions in the inner magnetosphere, which leads to an increase in the anisotropic distribution of ions and eventually assist in the excitation of the EMIC waves (Criswell, 1969). Another possible source is the solar wind pressure. The compression of the magnetopause due to an increase in solar wind pressure leads to an increase in the anisotropy of the ions by the process of betatron acceleration (McCollough et al., 2010; Fu et al., 2012a,b; Park et al., 2016; Xue et al., 2021). As multiple factors control the generation of EMIC waves, many statistical studies have been conducted using satellite data to understand the role of different interplanetary parameters and magnetospheric conditions in the generation of the EMIC waves. For example, solar wind dynamic pressure pulses contribute positively to the growth of EMIC waves observed on the dayside (Usanova et al., 2012; Remya et al., 2015; Liu et al., 2019). The association of solar wind pressure and EMIC wave generation is reflected in the ground observations as well. A survey of ground EMIC waves has revealed that most of the EMIC wave events were linked with an increase in the solar wind dynamic pressure (Regi et al., 2017). On the contrary, the ion injections from the magnetotail, enhanced convection, and substorms are known to trigger the generation of EMIC waves observed in the dusk side (Jordanova et al., 2001; Kim et al., 2017; Remya et al., 2018). The auroral electrojet (AE) index is one of the proxies, which is used to identify the substorm (Davis and Sugiura, 1966). Meredith et al. (2014) studied the effect of the AE index on the occurrence of H and He band EMIC waves. Keika et al. (2013) have studied the global occurrence of EMIC waves using satellite data and attributed their excitation to two independent processes: 1) externally triggered waves on the dayside that are driven by solar wind pressure and 2) the internally excited waves in the dusk side caused by the energetic ion injections from the midnight sector. In addition, these authors reported that during the quiet time, the occurrence of EMIC waves is more likely in the morning sector. Above all studies hint that overall, the EMIC wave generation is likely in the afternoon, evening, and morning sectors, which are associated with different triggering source mechanisms. Thus, a detailed investigation of the local time distribution of EMIC waves in different frequency ranges to understand underlined physical processes is worthwhile.

In this context, we present the local time distribution of EMIC waves in different frequency ranges and their dependence on the solar wind pressure and auroral electrojet indices using long-term ground observations from the Indian Antarctic station, Maitri. The article is organized as follows: The information about the experimental setup and identification of the EMIC waves are detailed in Section 2. The results are presented in Section 3 and discussed in Section 4. The present work is summarized and concluded in Section 5.

In the present study, we have utilized ground induction coil magnetometer (ICM) data recorded at the Indian Antarctic station, Maitri (geographic lat. 70.7 ^o S, geographic long. 11.8 ^o E, L≈5) for the period of 2011–2017. The ICM consists of three magnetic coil sensors, which record the magnetic field fluctuations in N-S, E-W, and vertical directions. The data are sampled at either 64 Hz or 256 Hz. For each day, the variations in three components of the magnetic field, namely, B _x , B _y , and B _z are used to get the Fourier spectrogram. The power spectral density in the spectrogram of total magnetic field variation is obtained from B(f, t) = P _x (f, t) + P _y (f, t) + P _z (f, t). Here, P _x (f, t), P _y (f, t), and P _z (f, t) are the power spectral densities of individual magnetic field components B _x , B _y , and B _z , respectively. The spectrogram represents the magnetic field power B(f, t) in the frequency–time domain with the resolution of 0.011 Hz and 18 s, and it is in the units of nT²/Hz. We noticed that at Maitri, the observed EMIC wave frequencies predominantly lie below 2 Hz and occasionally exceed 2 Hz. Therefore, EMIC wave occurrence is noted in four frequency bins as 1) f ₁ = 0.12–0.5 Hz, 2) f ₂ = 0.5–1 Hz, 3) f ₃ = 1–1.5 Hz, and 4) f ₄ = 1.5– 2 Hz. These frequency limits are not associated with the EMIC wavebands; it is simply to separate EMIC waves into four categories based on the frequencies associated with the EMIC wave observed at ground station Maitri. For each day, we checked the average power in the spectrogram (i.e., ⟨B(f, t)⟩) corresponding to each frequency range. Using the Fourier spectrogram, one can identify the presence/absence of the wave in the Pc1–Pc2 range, and time variations of average power, ⟨B(f, t)⟩ in different frequency ranges allow us to verify whether or not the EMIC wave power is above the noise level. This way, we identified the presence/absence of EMIC waves and cataloged their details like frequency range and start and end time. In our study, we have also used hourly values of solar wind dynamic pressure, and geomagnetic activity indices like AE, Dst, and Kp from OMNIWeb. We would like to highlight the fact that, unlike the satellite observations, the frequency bins used here do not give any direct information about the source ions and generation region of EMIC waves. This is due to the fact that the propagation of EMIC waves along magnetic field lines and their ducting through the ionosphere attenuate wave frequency and amplitude (Johnson and Cheng, 1999; Kim and Johnson, 2016). All ICM data on magnetically quiet days are considered in the present study. A day is considered magnetically quiet if the Ap value for that day or the previous day is less than 20.

As an illustration, we have shown one example of EMIC waves that occurred on 30 April 2017 in Figure 1. Figure 1A shows the Fourier spectrogram of variation in total magnetic field recorded by ICM for the entire day, wherein two distinct EMIC wave events are clearly visible. We have highlighted one EMIC wave event with a dotted rectangle, which has an extended frequency range. In Figure 1B, we have shown an enlarged view of the highlighted event. For this highlighted event, Figure 1C shows the variation of the average power of the Fourier spectrogram (i.e., ⟨B(f, t)⟩) for four different frequency ranges f ₁ (black), f ₂ (red), f ₃ (blue), and f ₄ (green). The horizontal line at -100 dB represents the cutoff used to select the EMIC wave events with significant power. We have observed that in the Fourier spectrogram, the average power generally falls below -100 dB in the absence of EMIC wave activity. Therefore, we chose this limit. The horizontal dotted line indicates the EMIC wave power above -100 dB (i.e., > 1 0 − 5 nT/Hz) in respective frequency ranges, and their durations are 2.4–3.97, 1.2–6.5, 3.7–5.0, and 4.4–4.6 UT hours for f ₁, f ₂, f ₃, and f ₄ frequency ranges, respectively. Here, UT = LT-0.77 h. This way, we have cataloged all EMIC wave events, and the total occurrence of EMIC waves in different frequency ranges is summarized in Table 1. It is found that the EMIC wave occurrence is maximum in the f ₁ frequency range, which gradually decreases for f ₂ and f ₃ frequencies, and it is minimum in the f ₄ frequency range. The total occurrence durations of EMIC waves are approximately 1969 h, 1083 h, 125 h, and 17 h for f ₁, f ₂, f ₃, and f ₄ frequency ranges, respectively. The percentage of the duration of EMIC wave occurrence in f ₁, f ₂, f ₃, and f ₄ frequency ranges are 61.7%, 33.9%, 3.9, and 0.5%, respectively.

In the present study, we have shown that the EMIC waves are mainly seen during the early morning, afternoon, and evening time sectors. The afternoon and evening time EMIC waves are attributed to magnetospheric compression due to solar wind pressure and nightside particle injection, respectively. Also, these afternoon–evening time EMIC waves are found to be associated with lower frequencies ( < 1 Hz), whereas, morning time EMIC waves have some link with the magnetic activity that occurred on preceding days, and mostly these waves have high frequencies ( > 1 Hz). The EMIC wave occurrence during the early morning, afternoon–evening sectors has been reported by various satellite observations of EMIC waves. These tendencies were attributed to two independent processes that cause the EMIC wave generation, that is, magnetospheric compression by the solar wind pressure and the energetic ion injections from the midnight sector (Keika et al., 2013; Kim et al., 2017; Remya et al., 2018). Min et al. (2012) also reported two peaks in the EMIC wave occurrence, one is in the dawn, which is dominated by the hydrogen band and another is in the dusk sector, which is dominated by the helium band. Although we are unable to relate the EMIC wave frequency observed on the ground to the gyro-frequency of the source ions (like H + or He +) responsible for the generation of these waves, the local time patterns observed on the ground are mere reflections of the local time occurrence of EMIC waves in space reported from the satellite observations.

Though we have some clue about the possible mechanism responsible for these preferred local times, however, one has to understand the difference in the frequencies for the early morning and afternoon–evening time EMIC waves. Using the AMPTE satellite Anderson et al. (1996), have observed diurnal patterns and reported differences in the characteristics of EMIC waves in the morning and afternoon time. These authors examined the generation conditions and found that actual ion temperature is low and the temperature anisotropy A _p (=T _⊥/T _‖) is higher during the morning hours in comparison with the afternoon. The relation of wave frequency to gyro frequency is given by Ω = Ω _i [A _p /(A _p + 1)]. The value of A _p /(A _p + 1) was 0.37 for the noon event and 0.46 for the morning event. This difference in the temperature anisotropy could be a possible reason to see higher frequency waves during the morning. Moreover, during the afternoon–evening time, the cold plasma density is higher than that to dawn because of the presence of the plasmaspheric plume. High cold plasma density leads to a lower frequency of the waves during the afternoon–evening time (Horne and Thorne, 1994). There could also be a difference in the composition of the source ion population. Satellite studies (Van Allen probe) have also reported more H ⁺ band waves at dawn and He ⁺ band waves at dusk (Gamayunov et al., 2018).

In the present study, we have shown the diurnal distribution pattern of EMIC wave occurrence using seven years of ground observations from the Indian Antarctic station, Maitri. The EMIC waves were identified and their occurrence is noted in four different frequency ranges. Overall occurrence in different frequency ranges is summarized in Table 1. On the ground, we found two preferred local time sectors, namely, morning and afternoon–evening for the occurrence of EMIC waves. The high-frequency ( > 1 H z ) EMIC waves are mainly seen in the morning sector, whereas the low frequency ( < 1 H z ) EMIC waves are prevalent in the afternoon–evening sector. These local time distributions of EMIC wave occurrence are examined in light of possible controlling parameters like solar wind dynamic pressure and AE index (a proxy for the substorm activity). The main outcomes of the present study are listed as follows:

1. The majority of the EMIC wave events observed at ground station Maitri are having a frequency below 1 Hz.

Recently, Chen et al. (2020) have investigated the role of AE and solar wind pressure separately using satellite observations and found that there exist two local time zones (i.e., afternoon 9–15 MLT and evening 16–20 MLT) for the generation of EMIC waves. The present study clearly indicates that the preferential local time zones reported by satellite for the EMIC wave occurrence are also evident in the ground observations. The study by Mann et al. (2014) has reported that although from the ground magnetometers the wave was 18 + hours long, the satellite coverage showed the wave lasted only for 10−15 min. Satellite observations are susceptible to bias based on their locations and coverage, which can be overcome by long-term ground observations. For a comprehensive picture of EMIC wave distribution and the particle dynamics in the Earth’s magnetosphere, along with satellite studies, such a statistical study of EMIC wave observations using multi-ground stations will be useful, which is left as a future assignment.