Thin layers of enhanced electron density compared to the background ionization in the ionosphere’s E region are referred to as sporadic E and abbreviated as Es (Whitehead, 1989; Mathews, 1998; Wu et al., 2005; Haldoupis, 2011). The Es layer is mainly known as a daytime and summer hemisphere phenomenon with higher occurrence rate at mid and low latitudes (Haldoupis et al., 2007; Christakis et al., 2009; Arras et al., 2010; Haldoupis, 2012).

Whitehead (1961) and Axford and Cunnold (1966) proposed that Es layers are formed by the wind shear mechanism. In subsequent studies, the wind shear theory has been confirmed as the physical mechanism responsible for the mid-latitude Es layer formation process (Whitehead, 1989; Haldoupis & Pancheva, 2002; Haldoupis, 2011; Yamazaki et al., 2022). According to the wind shear theory, metallic ions in the ionosphere dynamo region converge to thin layers of ionization by the vertical shears of neutral wind, mainly produced by atmospheric tides (Haldoupis, 2012; Chu et al., 2014; Shinagawa et al., 2017; Qiu et al., 2019; Sobhkhiz-Miandehi et al., 2022).

Es has been subject to many studies since the mid-twentieth century (Whitehead, 1961; 1970; 1989; Macleod, 1966) due to its potential disturbances on the radio signals in communication and navigation systems. Strong vertical electron density gradients of Es layers can cause severe radio wave propagation disruption and will directly influence the accuracy and reliability of satellite communication and navigation systems (Arras, 2010).

For many years, most of the observational studies of Es were performed using ionosonde recordings; there were also several investigations based on incoherent and coherent scatter radars and some through in situ measurements with rockets. These techniques have been documented in several review papers (Matsushita, 1962; Whitehead, 1989; Mathews, 1998; Haldoupis, 2011). However, with the development of the GPS radio occultation technique over the past two decades, satellite observations have become an increasingly popular tool for studying Es layers due to their ability to provide global data coverage (Igarashi et al., 2001; Wu et al., 2005; Arras et al., 2008).

First, Es occurrence global maps were introduced in the early 60s based on ionosonde measurements (Taguchi, 1961; Leighton et al., 1962). Reddy and Matsushita (1969) studied Es layers at mid and low latitudes utilizing the ionosonde to establish a more detailed understanding of temporal and latitudinal variations of them. Subsequently, until the 1990s, numerous studies typically focused on examining the diurnal and seasonal variations of Es by utilizing ionosonde observations (Axford & Cunnold, 1966; Harris and Taur, 1972; Chandra and Rastogi, 1975; Saksena, 1976; MacDougall, 1978; Baggaley, 1984).

In the 2000s, scientists started using GPS radio occultation measurements to study lower ionospheric irregularities such as Es. For the first time, Igarashi et al. (2001) conducted a study that used less than 6,000 occultation GPS/MET radio data from 1995 to examine the seasonal dependency of Es layers. Wickert et al. (2004) suggested that fluctuations in the CHAMP radio occultation amplitude data can be associated with ionospheric scintillations. Furthermore, Wu et al. (2005) presented a global climatology of the phase variances and signal-to-noise ratio (SNR) fluctuations using approximately 6000 GPS/CHAMP occultation in the E region.

Since the release of radio occultation measurements from FORMOSAT-3/COSMIC in the summer of 2006, our understanding of Es has significantly improved due to its global coverage. Arras et al. (2008) published the first Es occurrence rate global climatology using CHAMP, GRACE, and FORMOSAT-3/COSMIC radio occultation, which had a significantly better spatial and temporal resolution compared to previous studies. Subsequently, studies investigating Es on a global scale have been primarily based on radio occultation measurements (Wu et al., 2005; Chu et al., 2014; Qiu et al., 2019; Yu et al., 2019; Luo et al., 2021; Arras et al., 2022), while some literature has used ionosonde observations locally as an evaluation tool for their satellite-based investigations (Gooch et al., 2020; Carmona et al., 2022; Gan et al., 2022; Yu et al., 2022). Recently, Gan et al. (2022) used radio occultation data from the new China Seismo-Electromagnetic Satellite mission to derive Es events and used Wuhan ionosonde observations to assess the reliability of their research. Carmona et al. (2022) compared different GPS radio occultation techniques with ionosonde measurements over an 8-year period in order to identify the most accurate method for detecting Es occurrences using radio occultation data. Gooch et al. (2020) presented a global comparison of the intensity and height of mid-latitude and equatorial Es layers derived from FORMOSAT-3/COSMIC radio occultation and digital ionosonde measurements. All of these studies found a qualitative agreement between their satellite dataset and ionosonde measurements of the Es layer. However, they were limited by observations at a single ionosonde location during a restricted time frame and did not account for the diurnal, seasonal, and spatial dependence of Es layer detectability with satellite and ground-based techniques. In this paper, we have conducted a multi-year comparison of the mid- and low-latitude Es events recorded by an ionosonde ground-based technique to those observed using the satellite radio occultation method. Moreover, the diurnal, seasonal, and local dependence of the agreement between the two methods has been examined.

Through a comparison of ionosonde recordings and GNSS radio occultation profiles, the results demonstrate that the agreement between the two methods in detecting Es layers ranged from 60% to 80% at different stations. We identified four possible outcomes when comparing the detection of Es layers using ionograms and GPS SNR profiles: 1) both methods detected the Es layer, 2) the Es layer was identified in ionograms but not in GPS SNR profiles, 3) GPS SNR profiles indicated the presence of an Es layer, but no signal reflection was recorded via the ionosonde in the corresponding epoch in the E region, and 4) neither method recorded the occurrence of an Es layer. To gain a more comprehensive understanding of the distribution of these four types across all conjunction data points, we provided a confusion matrix of approximately 9,000 mid-latitude observations, presented in Table 2. It is important to note that in this paper, the notation Es = 1 signifies the presence of the Es layer, while Es = 0 denotes its absence. Furthermore, the detection of the Es layer using a specific technique is shown on the index. For instance, Es_ionogram = 1 indicates that the Es layer has been detected using an ionosonde.

Table 2 illustrates that the detection of Es layers is in agreement between ionograms and radio occultation observations in approximately 70% of the data points (summation of Type 1 and Type 4). The remaining observations are divided into two categories: Type 2 and Type 3, which account for approximately 30% of the observations. Type 2 represents approximately 65% of the remaining data, where the Es layer is observed in ionograms but not in radio occultation profiles. Conversely, Type 3, which represents almost 35% of the remaining observations, indicates the presence of an Es layer according to satellite data, but no signal reflection was recorded in ionograms. Examining similar tables for each station from 2007 to 2017, we observed that the disagreement rate between the two methods (summation of Type 2 and Type 3) varies between 20% and a maximum of 40%.

Figure 2 showcases two illustrative examples, specifically Type 2 and Type 3, that were examined in this study. In Figure 2A, we present an ionogram captured from the I-Cheon station on February 8, 2012, at 16:15. The station is located at 37.1° latitude and 127.5° longitude. The ionogram reveals the presence of an Es layer situated at an altitude of approximately 110 km. Notably, this Es layer appears relatively weak, characterized by an foEs value of 1.8 MHz and an fbEs value of 1.6 MHz. The subsequent ionogram taken at 16:30 does not display the presence of the Es layer observed earlier. This observation suggests the possibility that the I-Cheon station is located at the edge of the Es layer, where small horizontal displacements may influence the observations and result in the layer’s intermittent visibility. In Figure 2B, we examine the GPS L1 SNR profile and its standard deviation for the same day. This profile was obtained using a 3-min time interval and a 0.2° difference in latitude at the same longitude. Upon closer analysis, we notice a slight increase in the signal standard deviation at altitudes approximately 95 km and 110 km. However, these deviations do not meet our predetermined criteria for detecting an Es layer. Despite the observed variations in signal standard deviation, they do not exhibit the necessary strength or magnitude to fulfill our detection criteria.

The ionogram displays two distinct ionospheric echo traces, namely, the ordinary (pink) and extraordinary (green) traces, which are attributed to the ionosphere’s doubly refracting nature caused by the Earth’s magnetic field (FengJuan et al., 2022).

As an illustrative example of Type 3, we present Figure 2C, which showcases an ionogram obtained from the Grahamstown ionosonde station on October 18th, 2010, at 19:15. Upon examining the ionogram, it is evident that there is no recording of the ordinary mode Es layer. This absence of a distinct Es layer signature indicates that the concentration of plasma in the E layer is insufficient to meet the criteria for classification as an Es layer in the ionogram. This observation suggests that the conditions necessary for the formation of a well-defined Es layer were not present during the time of the recording at the Grahamstown station. In contrast, Figure 2D provides further insights by presenting the GPS L1 signal data collected from a location situated approximately 0.1° in latitude and longitude from the ionosonde station, at the same time as the ionogram recording. Notably, the GPS signal data reveal distinct disturbances in the altitude range of 100–110 km. These disturbances are indicated by a standard deviation greater than 0.2, exceeding our predetermined criteria for detecting an Es layer in radio occultation data. Therefore, based on our criteria, we can confidently confirm the detection of an Es layer using the radio occultation technique.

In order to better understand the discrepancy between ground-based and satellite technique in Es layer detection, we performed a statistical analysis using a multi-year dataset of crossing observations at six stations. The analysis has been performed within different distances from the center station, at different seasons, and local times. The subsequent sections present statistical results of conjunctions at the Pruhonice station (50° N and 14.6ºE) as an example of two separate cases: case 1 refers to conjunction data points in which an Es layer is detected through the radio occultation technique and is either confirmed by ionograms or not (Es_{radio-occultation} = 1); case 2 is characterized by the detection of an Es layer through ionograms while radio occultation signals may or may not be affected by the Es layer (Es_ionogram = 1). It is important to note that in all the following results, the agreement rate between the two methods in case 1 is always higher than in case 2. This can be simply explained by results in the confusion matrix presented in Table 2. The number of conjunction observations in which the Es layer is detected by an ionogram and not by radio occultation is almost half of those in the reverse type.

According to Table 10, when the foEs value is greater than average, the agreement between the two methods in detecting the Es layer decreases. We also noticed that 64.4% of these conjunction observations with a high fbEs value had an extremely large SNR value as well. Figure 4 provides two examples of such events. Figure 4A shows a strong blanketing Es, and in Figure 4B, there is a crossing radio occultation signal which is highly disturbed but not marked as an Es event. Arras and Wickert (2018) only showed tag signals with a standard deviation greater than 0.2 within the altitude range of 10 km as an Es layer. By applying these criteria, very strong Es layers, which cause very big standard deviation at a high-altitude band, will sometimes be overlooked.

Figure 4C is another example of a severe Es layer that blankets the echo from upper layers of ionosphere. As shown in Figure 4D, the radio occultation signal traveling from above reaches an Es layer at approximately 110 km and gets disturbed intensely. As a result of this disturbance, the normalized SNR reduces to less than one third of its initial amount. Such severe disruptions (in this case foEs = 6.3 MHz) caused by Es events can sometimes result in GNSS signal loss (Yue et al., 2016). Hence, despite the global Es data coverage that radio occultation provides, it seems that further improvements can be applied to the Es detection criteria proposed by Arras and Wickert (2018).

The discrepancy between Es layers detected in ionograms and those observed in radio occultation might also exist due to limitations in the Es layer detection capability of the ground-based techniques. The ionosonde data used in this study are all derived from the Lowell data center. Nevertheless, it should be noted that the stations do not always utilize the same software. Furthermore, there are variations in hardware and technical aspects across different stations, including variations in antenna configurations from site to site. Factors such as sounding frequency, antenna type, system maintenance, and other technical considerations can potentially influence the detectability of Es layers. As demonstrated in Figures 3A, D, although the Learmonth station follows the same temporal dependence as other Southern Hemisphere stations, it has always a higher agreement rate with radio occultation observations in comparison to others. Ionosonde stations have different limits of density sensitivity, depending on their operating mode. Therefore, the detectability of the Es layer can depend on the ionosonde station and vary from one ionosonde to another based on their different software and technical facilities. Additionally, it is always possible to have human errors, when scaling ionograms manually.

Another potential factor contributing to the observed discrepancies between ground-based and satellite measurements of Es layers is the fundamental difference in the Es detection mechanisms employed by the two methods. It is important to recognize that ionosonde and satellite observations utilize distinct approaches to detect Es layers, which can yield divergent measurements of these atmospheric phenomena. These inherent dissimilarities in detection mechanisms may offer a plausible explanation for the contradictory findings observed between ground-based and satellite techniques. In ionosonde observations, Es layers are typically identified and recorded in ionograms by analyzing the electron density of the E layer, which is derived from the transmitted ionosonde signal. The ionogram provides information about the plasma density distribution within the ionosphere, allowing for the detection of Es layers based on their characteristic electron density patterns. However, in the radio occultation technique employed by satellites, the detection of Es layers relies on the observation of electron density gradients that induce scintillations in the global navigation satellite system (GNSS) signals. In this case, it is the presence and magnitude of these gradients that are indicative of the presence of Es layers. This fundamental disparity in the underlying mechanisms for Es detection between ground-based ionosondes and satellite-based radio occultation techniques can give rise to varying measurements of Es layers obtained through each method. For instance, there may be cases where an Es layer, such as the one depicted in Figure 2C, exhibits a plasma density that is too weak to be discerned and recorded in an ionogram. However, despite the lower plasma density, the associated electron density gradient can still be significant enough to cause disturbances in radio occultation signals, leading to their detection using satellite observations. This physical distinction in the detection approaches of the two methods may also provide an explanation for the observed maximum agreement between ground-based and satellite techniques during the daytime and local summer periods, as these conditions can favor the generation of notable electron density gradients within the Es layer.

In this study, we conducted a comprehensive comparison between the Es recordings obtained from six mid- and low-latitude ionosonde stations and the corresponding FORMOSAT-3/COSMIC radio occultation measurements, utilizing the method proposed by Arras & Wickert (2018). Our analysis focused on examining the local and temporal variations in the agreement between the ground-based and satellite techniques for detecting Es layers. The key findings of our investigation are summarized as follows:

1. At different ionosonde stations, we observed a noticeable disagreement of 20%–40% between the ground-based and satellite measurements of Es layer occurrence. The extent of this discrepancy varied depending on factors such as local time, season, and altitude at which an Es event occurred. These findings emphasize the importance of considering the specific conditions under which Es layers are detected.

In conclusion, our study conducts a comparison of ground-based and satellite measurements of Es layers, shedding light on the local and temporal factors influencing their agreement. While acknowledging the limitations and specific conditions of our study, further investigations can build upon these findings to refine the detection and characterization of Es layers and enhance the overall understanding of ionospheric dynamics. Future research should explore alternative data analysis techniques, incorporate a larger number of ionosonde stations, and consider additional factors that may impact the agreement between ground-based and satellite measurements. By addressing these aspects, we can advance our knowledge of Es layers and their impact on ionospheric behavior in different regions and seasons.