Spatio–Temporal Evolution of Electric Field, Magnetic Field and Thermal Infrared Remote Sensing Associated with the 2021 Mw7.3 Maduo Earthquake in China

Abstract

This study presents the spatio–temporal evolution of the electric and magnetic fields recorded by the China Seismo–Electromagnetic Satellite (CSES) and the thermal infrared remote sensing data observed by the Chinese stationary meteorological satellites Feng Yun–2G (FY–2G) associated with the 2021 Mw7.3 Maduo earthquake. Specifically, we analyzed the power spectrum density (PSD) data of the electric field in the extremely low frequency (ELF) band, the geomagnetic east–west vector data, and the temperature of brightness blackbody (TBB) data to investigate the spatio–temporal evolution characteristics under quiet space weather conditions (Dst > −30 nT and Kp < 3). Results showed that (1) the TBB radiation began to increase notably along the northern fault of the epicenter ~1.5 months prior to the occurrence of the earthquake. It achieved its maximum intensity on 17 May, and the earthquake occurred as the anomalies decreased. (2) The PSD in the 371 Hz–500 Hz and 700 Hz–871 Hz bands exhibited anomaly perturbations near the epicenter and its magnetic conjugate area on May 17, with particularly notable perturbations observed in the latter. The anomaly perturbations began to occur ~1 month before the earthquake, and the earthquake occurred as the anomalies decreased. (3) Both the magnetic –east–west component vector data and the ion velocity Vx data exhibited anomaly perturbations near the epicenter and the magnetic conjugate area on 11 May and 16 May. (4) The anomaly perturbations in the thermal infrared TBB data, CSES electric field, and magnetic field data all occurred within a consistent perturbation time period and spatial proximity. We also conducted an investigation into the timing, location, and potential causes of the anomaly perturbations using the Vx ion velocity data with magnetic field –east–west component vector data, as well as the horizontal –north–south and vertical component PSD data of the electric field with the magnetic field –east–west component vector data. There may be both chemical and electromagnetic wave propagation models for the “lithosphere—atmosphere—ionosphere” coupling (LAIC) mechanism of the Maduo earthquake.

1. Introduction

Large numbers of ionospheric observations showed that there were obvious coupling relationships between the phase of preparation or occurrence of earthquakes and the state of the ionosphere [1,2,3,4,5,6,7].

Many experts have investigated the mechanism for a long time and observed that seismic electromagnetic signals could be observed in ultralow –frequency (ULF)/extremely low frequency (ELF)/very low frequency (VLF) bands in the ionosphere above the epicentral regions several days or hours before the earthquakes [8,9,10,11,12]. Sometimes, some signals were even amplified to a certain extent [13,14,15,16]. Seismo–ionospheric perturbation signals can spread from the lithosphere to the atmosphere and the ionosphere via electromagnetic wave, chemical, and acoustic gravity wave channels. These have been concluded as the models for “lithosphere—atmosphere—ionosphere” coupling (LAIC) during signal propagation. As for the electromagnetic wave channel, we know that ULF emissions are generated before an earthquake, and these emissions propagate into the inner magnetosphere, where they interact with energetic particles. Then, we expect the precipitation of those particles into the lower ionosphere. As for the chemical channel (also known as the DC (Direct Current) electric field channel), the change in geochemical parameters (gas or radon emanation, water elevation, etc.) leads to changes in the air or in the conductivity of the air and results in a modification of the atmospheric electric field. This field will then influence the plasma density in the ionosphere. As for the acoustic gravity wave channel, we may expect the excitation of atmospheric oscillations, which propagate upward into the ionosphere, modifying the ionospheric plasma density [1,2,3,17]. The electromagnetic signals in the ionosphere contain information related to earthquake precursors. If the methods for extracting anomalies are improved, more earthquake precursors can be identified. Meanwhile, validating potential channels for LAIC mechanisms between layers has always been a research hotspot in the academic community.

In the past, several processes were considered as possible contributors to seismic thermal infrared anomalies [18]: (a) rising fluids that would lead to the emanation of warm gases [19,20]; (b) rising well water levels and CO₂ spreading laterally and causing a ‘‘local greenhouse’’ effect [21,22,23]; (c) activating positive–hole pairs during rock deformation [24]; (d) the radon ionization of the near surface air and the latent heat exchange due to variations in humidity [25]; and (e) piezoelectric and elastic strain dilatation forces [20,26]. The thermal infrared mechanism exhibits similarities to the chemical channel and electromagnetic wave channel in the LAIC system. Moreover, the thermal infrared anomalies observed during earthquakes are believed to be linked to electromagnetic perturbation. Many studies revealed that the thermal infrared anomalies were increasing notably during the fault stress accumulation and release associated with some major earthquakes [27,28,29,30,31,32]. Although Satti et al. [28], Zhong et al. [30], De et al. [31] and Wang et al. [32] have used thermal infrared and ionospheric data to study the seismo–ionospheric perturbation jointly in the aforementioned study, they did not explain or explained only one possible channel in the models for LAIC. If joint research on thermal infrared data and ionospheric electromagnetic field data is conducted to validate the LAIC mechanisms between layers, identifying potential channels will be a key focus of future studies.

Continental earthquakes have occurred with high frequency and caused severe devastation in China. According to previous studies, 33% of the world’s strong continental earthquakes have occurred in this region [33]. Since the beginning of 21st century, notable earthquakes have struck China, including the 2001 Ms8.1 Kunlunshan Mountain earthquake, the 2008 Ms8.0 Wenchuan earthquake, the 2010 Ms7.1 Yushu earthquake, and so on. The dynamic interplay between the Indian plate, the Pacific plate, the Philippine Sea plate and the Eurasian plate has resulted in the alteration of various active structural configurations. It controlled the spatial distribution pattern of the strong earthquakes in mainland China [33]. The China Seismo–Electromagnetic Satellite (CSES) has recorded more than 70 Ms ≥ 7.0 earthquakes and more than 500 Ms ≥ 6.0 earthquakes in the world since it was launched on 2 February 2018. Among these earthquakes, more than 20 Ms ≥ 6.0 strong earthquakes have occurred in China since it was launched. On 21 May 2021, at 18:04 (UTC), the Mw 7.3 Maduo earthquake occurred in the southern portion of the Qinghai Province, China, with an epicenter at 34.59° N latitude and 98.24° E longitude at a depth of 10 km (USGS). This strong earthquake occurred within the Bayan Har block, which is one of the laterally extruded active blocks on the Qinghai–Tibet Plateau since the Cenozoic Era [34,35]. The 2021 Mw7.3 Maduo earthquake disrupted the relatively tranquil seismic period characterized by events with a magnitude of Ms ≥ 7.0 in mainland China, followed by the 2022 Ms6.8 Luding earthquake and the 2023 Ms6.2 Jishishan earthquake recently (

Table 1. Catalog of Earthquakes with an Ms ≥ 6.0 in Western China from 1 January 2021 to 31 December 2023.

Magnitude (Ms)	Date	Longitude (°E)	Latitude (°N)	Depth (km)	Location
6.1	19 March 2021	92.75	31.94	10	Biru
6.4	21 May 2021	99.88	25.70	10	Yangbi
7.4	22 May 2021	98.37	34.61	17	Maduo
6.0	16 September 2021	105.34	29.20	10	Luxian
6.9	8 January 2022	101.25	37.77	10	Menyuan
6.0	26 March 2022	97.33	38.50	10	Delingha
6.1	1 June 2022	102.94	30.37	17	Yaan
6.0	10 June 2022	101.82	32.25	13	Maerkang
6.8	5 September 2022	102.8	29.59	16	Luding
6.2	18 December 2023	102.79	35.7	10	Jishishan

). There were 35 strong earthquakes with magnitudes of Ms ≥ 6.0 that occurred in mainland China, including 4 strong earthquakes with magnitudes of Ms ≥ 7.0 within 10 years. Therefore, it is urgent for experts to improve the forecasting methods. In the study, we studied the spatio–temporal evolution characteristics of electric and magnetic fields collected by the CSES and the thermal infrared remote sensing data detected by the Chinese geostationary meteorological satellites Feng Yun–2G (FY–2G), associated with the 2021 Mw7.3 Maduo earthquake.

2. Data Selection and Processing

2.1. Thermal Infrared Remote Sensing Data

Zhong et al. [30] compared the outgoing longwave radiation (OLR), the temperature of brightness blackbody (TBB), and the medium–wave infrared brightness (MIB) data detected by the FY–2G satellite. They found that the MIB data contained significant noise away from the anomaly area and the OLR anomalies emerged with weak amplitudes. Thereby, they confirmed that TBB data was best for extracting infrared anomalies. We selected the TBB data from thermal infrared remote sensing data. The FY–2G satellite was launched on 31 December 2014 and located at a fixed point of more than 35,000 km above the equator. It has an effective observation range of 45°–165° E and 60° S–60° N. The band of TBB data we selected was 10.3–11.3 μm. The spatial size for computing the TBB data of the FY–2G satellites is longitude 0.05° × latitude 0.05° [36].

We selected the power spectrum data to extract TBB anomalies. The power spectrum reflects the power energy of each frequency component. It can reveal the periodicity and the spectral peaks. To deal with extensive waveform data, we adopted the power spectrum estimation to obtain the anomaly frequency and amplitude. The purpose was to investigate the difference between the wave radiation changes before or after an earthquake and other periods. In this study, to avoid the impact of solar radiation on the land surface and considering the CSES ascending node at LT 02:00, we selected the data recorded from LT 1:00 to 2:00 to calculate the mean value to study the anomalies. Therefore, we collected data from LT 1:00 to 2:00 from 1 January 2020 to January 2023. The main steps for extracting anomalies from power spectra were as follows [37,38,39].

(1)

Our main concerns revolved around the reconstructed signal and the error of the original signal. We used the db8 wavelet basis in the Daubechies (dbN) wavelet system to wavelet transform the data. To achieve this, we experimented with various wavelet bases to decompose and reconstruct the signal. The characteristic of the dbN wavelet system is that it can divide the frequency band better with the increase in the order. However, as the calculation greatly increases, the real–time performance deteriorates. Conversely, a dbN wavelet system with too small an order (such as Db3) divides the frequency band roughly [40]. We also considered the dbN wavelet system calculation, so we selected the db8 wavelet basis. It could exclude, as far as possible, the Earth’s annual, daily, and basic temperature fields, temperature variations caused by hot or cold air currents and rain clouds, and minor temperature variations caused by other factors. Cold or hot air currents and rain clouds caused by temperature changes generally occur on short time scales (hours to days). They include high–frequency information and could be removed by a second–order wavelet transform based on the db8 wavelet basis. The Earth’s basic and annual temperature fields (notable annual variation information) could also be removed by applying the seven–order wavelet transform based on the db8 wavelet. They include low–frequency information. Since the second–order wavelet scale subtracted the seven–order wavelet scale, the information in the middle frequency band could be retained. The information in the high and low frequencies was omitted, and this step was equivalent to band–pass filtering.

(2)

We carried out a Fourier transform process to data within a moving window (window length n = 64 days, step m = 1 day). We obtained each power spectrum by windowing the start time of the data and the time series data for each pixel. In this way, the time–frequency spatial data were obtained. The similarities and differences between the thermal power spectra of the aseismic and seismic periods were analyzed using frequency and amplitude.

(3)

The power spectrum frequencies of each pixel were calculated from relative amplitudes using Equations (1) and (2) to generate spatial data with relative time–frequency variations.

$${\mathrm{A}}_{\mathrm{ik}}=\frac{1}{\mathrm{l}}{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{l}}{\mathrm{W}}_{\mathrm{ijk}} \left(\mathrm{i} = 1, 2\dots , \mathrm{n}; \mathrm{k} = 1, 2\dots , \mathrm{m}\right)$$

(1)

$${\mathrm{R}}_{\mathrm{ijk}}=\frac{{\mathrm{W}}_{\mathrm{ijk}}}{{\mathrm{A}}_{\mathrm{ik}}} \left(\mathrm{i} = 1, 2,\dots , \mathrm{n}; \mathrm{j} = 1,2,\dots , \mathrm{l}; \mathrm{k} = 1, 2,\dots , \mathrm{m}\right)$$

(2)

where ${\mathrm{A}}_{\mathrm{ik}}$ is the average power spectrum amplitude within statistical time (length l) in the kth frequency of the ith pixel. ${\mathrm{W}}_{\mathrm{ijk}}$ is the power spectrum amplitude of the ith pixel on the jth day in the kth frequency and n is the total number of pixels, m is the number of frequency points and points, and l is the total number of time series data. R_ijk is the relative power spectrum amplitude of the kth frequency on the jth day of the ith pixel calculated by Equation (2). The obtained time–frequency spatial data were used to scan the entire spatial and temporal range and all frequency bands to identify frequencies (i.e., the characteristic period) corresponding to large changes in amplitude.

2.2. Electric Field Data

The electric field data in this study were recorded by the CSES which was launched on 2 February 2018, with a circular Sun–synchronous orbit. It is at an altitude of ~507 km with an inclination of ~97.4°. The descending node occurs at LT 14:00, and the ascending node occurs at LT 02:00. Its major scientific objectives are to detect the electromagnetic environment and earthquake–related perturbations in the topside ionosphere [41]. The CSES carries eight payloads. The electric field detector (EFD) is one of the actual payloads for space electromagnetic properties research and measures the electric field between the DC and 3.5 MHz [9], providing calibrated scientific data in four frequency bands: ULF (DC −16 Hz), ELF (6 Hz–2.2 kHz), VLF (1.8–20 kHz), and HF (high–frequency, 18 kHz–3.5 MHz).

We used the perturbation amplitude method of electric field power spectrum density (PSD) to extract the electric field anomalies. In this study, to avoid the impact of solar radiation, space electromagnetic environment disturbance, and considering the TBB data at the same time, the data during the local nighttime were selected to study anomalies. Thereby, we collected data during ~LT 2:00–2:30 considering only quiet geomagnetic field time (Dst > −30nT and Kp < 3). The main steps for extracting anomalies were as follows [42].

(1)

Our observational study of orbital data passing near the epicenter exhibited an enhancement of the proton cyclotron frequency at ~600 Hz. In order to ensure the credibility of the data, we divided the frequency bands of the power spectral density into 371 Hz–500 Hz and 700 Hz–871 Hz [42,43], as shown in Figure 1.

(2)

We constructed the statistical background of observation (80° E–120° E; 12° N–52° N). The CSES has a revisit period of 5 days. The distance of its neighbor track is ~4.73° in longitude [41]. Thereby, in order to ensure that every grid exhibited data, we used a sliding step of 4° in longitude and 1° in latitude with at least one week of data in the region based on our tests. There were 10 × 40 grids in the region with a precision longitude 4° × latitude 1°.

(3)

We calculated the PSD data in the ELF band from 1 January 2021 to 12 December 2022. We got the median matrix β as the background and the standard variance matrix σ as the background perturbation. Two groups of 10 × 40 matrices could be obtained.

(4)

We selected data every 7–days to calculate the median to get the median matrix α as the real–time data. Finally, we used Equation (3) to extract perturbation amplitude θ.

$$\mathsf{\theta }=\left(\mathsf{\alpha }-\mathsf{\beta }\right) /\mathsf{\sigma }$$

(3)

where θ is a dimensionless parameter, which represents the multiple of standard deviation.

2.3. Magnetic Field Data

The magnetic field data in this study were also from the CSES. We selected the magnetic field data detected by the high–precision magnetometer (HPM) payload onboard CSES. The total geomagnetic field is measured by the HPM, which consists of two tri–axial fluxgate sensors (FGMs) [44] and one coupled dark–state magnetometer (CDSM) [45]. The FGMs provide the total geomagnetic field vector between the DC and 15 Hz, and the CDSM provides the scalar value of the total geomagnetic field [46]. We can get three–component data of the magnetic field from the HPM payload. In the study, we selected the –east–west vector component from FGMs because of its notable seismic anomalies [11,47,48]. Orbit data of the up–orbit group nearest to the epicenter were selected from 30 days before to 15 days after the earthquake under the condition of space magnetic disturbance (Dst ≥ −30 nT and Kp < 3). As the FGMs’ sampling rate is 1 Hz, a sliding average of every 50 s of data was performed to establish the background data B for each track data [49]. Sliding window selection will directly affect the smoothing effect of the data. A larger window enhances the smoothing effect, aiding in suppressing high–frequency variations, whereas a smaller window is less effective at mitigating high–frequency changes, making it more challenging to discern the overall trend of the waveform. In this study, we opted for a sliding window of 50 points, corresponding to a data duration of approximately 50 s. Consequently, the trend waveform obtained through the sliding average represents changes with periods greater than 50 s [49]. The real–time orbital data are N for each track data and we used Equation (4) to extract perturbation D.

$$\mathrm{D}=\mathrm{N}-\mathrm{B}$$

(4)

where D is perturbation, which represents the difference between the real–time data N and the background data B.

2.4. Research Area

The Maduo earthquake occurred on the Bayan Har block in the central Tibetan Plateau, which is a seismically active region in mainland China. This area is surrounded by a number of active faults, including the East Kunlun Fault, the Xianshuihe Fault, the Ganzi–Yushu Fault, the Margaichaka–Ruolagangri Fault, the Longmenshan Fault, and the Arkin Fault [50,51], as shown in Figure 2. The causative fault for the Maduo earthquake is the Jiangcuo Fault, a deep fault that spreads nearly parallel to the Dari Fault, the Maduo–Gande Fault, and the East Kunlun Fault [52,53]. The Jiangcuo Fault extends ~700 km with a general strike of NWW, a predominantly NE dip direction, a nearly vertical dip angle, and exhibits a left–lateral strike–slip motion [53]. It is categorized as a secondary fault south of the East Kunlun Fault [54,55]. The epicenter of the Maduo earthquake was located in the central section of the surface rupture zone, where branching rupture phenomena were observed at both ends of the rupture zone [52]. The earthquake resulted in a 150–km–long rupture zone along the Jiangcuo Fault, characterized by branching phenomena. The high level of stress accumulation was the primary factor driving the development of the Maduo mainshock and subsequent aftershocks [56].

The seismogenic area of the lithosphere was estimated based on the seismogenic formula proposed by Dobrovolsky et al. [57].

$$\mathsf{\rho }={10}^{0.43\mathrm{M} }$$

(5)

where ρ is the straight–line distance from the epicenter (the unit is km) and M is the magnitude of the earthquake. The earthquake we studied was an Ms7.4, so according to Equation (5), the straight–line distance from the epicenter in the seismogenic area is more than ~1520 km, which corresponds to a distance of ~13° from the epicenter. Seismo–ionospheric perturbation signals can spread from the Earth’s surface to the ionosphere, causing the ionosphere perturbations based on the LAIC mechanism. In addition, seismo–ionospheric perturbation signals may shift in the ionosphere during propagation. The maximum offset can be more than 10° in longitude or in latitude [13,14,15,16]. Thereby, we selected the research area which was outside ±10° in latitude and in longitude from its epicenter.

3. Results

3.1. Spatio–Temporal Evolution of Thermal Infrared Anomalies

The TBB radiation began to increase notably along the northern fault of the epicenter since 1 April. It spread and increased along the southeastern side of the fault gradually. It notably achieved its maximum on 17 May, and then began to decrease gradually. On 21 May 2021, at 18:04 (UTC), the Mw 7.3 Maduo earthquake occurred in the southern portion of the Qinghai Province, China, with an epicenter at 34.59° N and 98.24° E at a depth of 10 km (USGS) during the gradual disappearance of TBB anomalies. The TBB anomalies along the fault disappeared completely on June 6. The anomalies lasted for ~2 months, as shown in Figure 3.

It was observed that the maximum anomaly region was located in the southeast of the location (~90° E, ~39° N) during the period from 23 April to 29 April 2021. This region lies on the edge of the Tarim basin, which is rich in resources, particularly petroleum and natural gas. The basin is highly sensitive to pre–earthquake stress changes. When the stress accumulates to a certain degree, the peripheral active tectonic zones and microfractures in the basin become gas upwelling channels. As a result, greenhouse gases such as methane and carbon dioxide escaping from the surface create an obvious radiation warming effect [58,59,60,61]. Cicerone et al. [61] confirmed that pre–seismic thermal anomalies are more likely to occur in areas with geothermal heat and hydrocarbon–rich deposits when analyzing pre–seismic thermal radiation anomalies. Subterranean gases diffuse to the surface at specific rates and proportions [62]. Regional stress changes can lead to the formation of new fissure channels within fault zones due to stress extrusion, facilitating the rapid diffusion of deep–source gases and existing fissure gases to the surface, resulting in pre–earthquake warming. Meanwhile, the deep heat energy also conveys upward unevenly along the fissures to form warming [22,63]. Zhang et al. [60] analyzed the thermal radiation anomalies associated with eight earthquakes which occurred along the edges of the basins in Sichuan Province and Xinjiang Province since 2008. The findings revealed that the thermal radiation anomalies were similar to the basin morphology. Consequently, the maximum anomaly region might also be attributed to this basin effect.

3.2. Spatio–Temporal Evolution of Electric Field Anomalies

We selected the PSD data in the ELF frequency band detected by the EFD payload in the nighttime without the space magnetic environment perturbation. It was observed that the PSD in the 371 Hz–500 Hz and 700 Hz–871 Hz bands exhibited anomaly perturbations near the vicinity of the epicenter and its magnetic conjugate area on 17 May. Particularly notable were the anomalies in the magnetic conjugate area, which are shown in Figure 4, Figure 5 and Figure 6. The anomalies were found to be more notable in the horizontal –north–south and vertical directions, while being less notable in the horizontal –east–west direction (only three–component observations in the band of 371 Hz–500 Hz were shown in the manuscript due to the large numbers of figures).

A total of 300 orbital data in the two frequency bands from three months before to one month after the earthquake were selected from tens of thousands of available datasets for spatial–temporal evolution characterization. Using a 7–day time window and the above data processing, it was found that the perturbation amplitude θ in the 371 Hz–500 Hz band exhibited anomaly perturbations in the northwest of the epicenter from 20 April to 26 April as shown in Figure 7 and Figure 8. Additionally, the anomaly perturbations were detected northeast of the epicenter from 11 May to 17 May, corresponding to the locations of the anomalies in Figure 4, Figure 5 and Figure 6. As shown in Figure 8, the vertical component anomaly perturbations in the 371 Hz–500 Hz band were more notable compared to those in the horizontal –north–south component shown in Figure 7. The anomalies from 27 April to 3 May were similar to those from 11 May to 17 May. Aside from the periods mentioned above, notable anomalies were also detected in the epicenter region from 4 May 10 May, as shown in Figure 8. The two–component anomalies gradually decreased as the earthquake occurred and disappeared post–earthquake, as shown in Figure 7 and Figure 8. The spatial–temporal evolution characterizations of the horizontal –north–south component in the 700 Hz–871 Hz band were similar to those in the 371 Hz–500 Hz band, but more notable. Notable anomalies were observed in the epicenter region from May 4 to May 10, as shown in Figure 7 and Figure 9. The vertical component’s spatial–temporal evolution characterizations in the 700 Hz–871 Hz band were similar to those in the 371 Hz–500 Hz band, as shown in Figure 8 and Figure 10. However, the perturbation amplitude θ of the vertical components in the 700 Hz–871 Hz band was the most notable among all the components and frequency bands studied, as shown in Figure 10. Furthermore, there was no observable trend of decreasing perturbations during the earthquake occurrence period due to the superposition of pre–earthquake and co–seismic data, as shown in Figure 10. Separate calculations of pre– and post–earthquake data revealed that the perturbation amplitude θ tended to decrease after the earthquake, remain quiet during the earthquake, and increase before the earthquake, as shown in Figure 11. The spatial orientation distribution of the anomalies shown in Figure 7, Figure 8, Figure 9 and Figure 10 indicated that the anomalies after May were located in the east–southeast region of the epicenter, whereas those before May were situated in the northwest region of the epicenter. This distribution is similar to the spatial distribution of the TBB anomaly region in Figure 3. Specifically, TBB anomalies occurred on the northwestern flank of the epicenter before May, and shifted primarily to the region east of the epicenter after May. The temporal–spatial evolution of the TBB anomalies is approximately consistent with that of electric field anomalies.

3.3. Spatio–Temporal Evolution of Magnetic Field Anomalies

We selected two groups of up–orbit trajectories passing near the epicenter from 1 April 2021 to 15 June 2021, as shown in Figure 12. Based on previous studies [11,47,48], the magnetic field –east–west component vector data were selected for the study. Twenty–four ascending orbital data were selected from more than tens of thousands of orbital data for the study using the HPM data processing method above, as shown in Figure 13. The two sets of red dashed lines above in Figure 13 represent the latitudes of the epicenters for the 2021 Yangbi Ms6.4 earthquake and the 2021 Maduo Mw7.3 earthquake, respectively. The two sets of red dashed lines at the bottom represent the magnetic conjugate region, while the two pentagrams are the epicenters of the two earthquakes, respectively. Notable anomaly perturbations were observed in the vicinity of the Maduo epicenter and its magnetic conjugate area on 16 May, one week prior to the earthquake. Similar anomaly perturbations were also detected near the epicenter of the 11 May earthquake in Maduo, although there were no anomalies in the magnetic conjugate area at that time. Notable anomaly perturbations were also observed near the epicenter and in the magnetic conjugate area on the occurrence of the 22 May Maduo earthquake. For the Yangbi Ms 6.4 earthquake on 21 May, anomaly perturbations were also exhibited near the epicenter and the magnetic conjugate area, though they were less notable compared to those on 22 May. This pattern of simultaneous anomaly perturbations near the epicenter and in the magnetic conjugate area raises the question of whether such phenomena are indicative of a higher probability of seismic anomalies. We will select more additional cases to analyze this hypothesis in the subsequent studies. Aside from the periods mentioned above, the data recorded were quiet. The small steps in all orbits near the equator were attributed to the switching of the satellite and were not related to earthquakes.

4. Discussion

The anomaly perturbations observed in both the thermal infrared TBB data and the CSES electric and magnetic field data occur during the same temporal anomaly perturbations period, within close proximity to the same region, which can indicate that there may be some LAIC relationship. The thermal infrared anomalies during earthquakes are also types of electromagnetic perturbations. LAIC models are generally categorized into three propagation channels approximately, including electromagnetic wave propagation models, additional chemical (DC electric field) models and acoustic gravity wave models, each of which changes ionospheric plasma in the ionosphere [3]. The CSES flies at an altitude of ~500 km and detects physical parameters in the F layer of the ionosphere. If earthquakes affect the plasma in the D and E layers of the ionosphere near the epicenter, the plasma will rise to the F layer. Due to pressure gradient forces and gravity, the plasma will spread out toward the poles along the magnetic lines, thereby generating currents along this direction. Then, the horizontal component of the magnetic field will change, and this physical process could explain the obtained conclusions, i.e., Ampere’s law [11].

The parameters detected by the PAP (plasma analyzer package) payload onboard the CSES were considerably perturbed during earthquakes [11,64]. To verify the physical mechanism, we selected ion velocity Vx data recorded by the PAP payload associated with the two groups near the Maduo epicenter from April to May 2021 (the Vx velocity direction matches the direction of the satellite flight velocity), as shown in Figure 14. Both the magnetic –east–west component data detected by the HPM payload and the Vx data detected by the PAP payload exhibited anomaly perturbations near the epicenter and in the magnetic conjugate area on 11 May and 16 May, as shown in Figure 13 and Figure 14. However, only the magnetic –east–west component data exhibited notable anomaly perturbations on 16 May, whereas Vx data exhibited notable anomaly perturbations on 11 May, as shown in Figure 13 and Figure 14. This may indicate that there were other influences besides the upward transport of particles, which took on greater weight on 16 May. Consequently, when the Vx data exhibited minor anomaly perturbations, the magnetic –east–west component data exhibited notable anomaly perturbations on 16 May. Upon the occurrence of the earthquake, the magnetic –east–west component data exhibited notable anomaly perturbations in the vicinity of the epicenter and its magnetic conjugate area, while the Vx data exhibited notable anomaly perturbations in the magnetic conjugate area, with almost no anomaly perturbations detected near the epicenter, as shown in Figure 13 and Figure 14. This may indicate that the changes in the magnetic –east–west component data were influenced not only by the upward transport of particles but also by some other factors. The upward transport of particles may be attributed to the release of subsurface material, similar to (a)–(d) for the possible contributors to the seismic thermal infrared anomalies, and is consistent with the chemical (DC electric field) model of the LAIC. The release of subsurface material induces alterations in the atmosphere or its conductivity, consequently modifying the atmospheric electric field. This field, in turn, will influence the transport of plasma in the ionosphere. Electromagnetic perturbations can occur during periods without particle transport, with the electric field notably perturbed in the horizontal –north–south and vertical directions, and the magnetic field notably perturbed only in the horizontal –east–west component, consistent with the right–hand rule. Piezomagnetism and piezoelectricity generate outwardly radiating electromagnetic waves [65] and also produce thermal anomalies similar to the (e) for the possible contributors to the seismic thermal infrared anomalies. There appears to be a temporal and spatial correspondence with the Maduo earthquake, suggesting the upward propagation of electromagnetic waves from the surface to the ionosphere, thus supporting the electromagnetic wave propagation channel of LAIC models.

In this study, it can be approximately hypothesized that three channels in the LAIC models may have some weights before or after the earthquake, indicating that there may be more than one LAIC mechanism involved in the preparation and occurrence of seismic events.

5. Conclusions

In the study, we analyzed the spatio–temporal evolution of the electric field and magnetic field data recorded by the CSES and thermal infrared remote sensing data observed by the FY–2G satellite associated with the 2021 Mw7.3 Maduo earthquake. Some characteristics can be concluded as follows:

(1)

We selected the TBB data to be the thermal infrared remote sensing data observed by the FY–2G satellite. The TBB radiation began to increase notably along the northern fault of the epicenter ~1.5 months prior to the occurrence of the earthquake. It spread and increased along the southeastern side of the fault gradually. It achieved its maximum intensity on 17 May, and then began to decrease gradually. The Maduo earthquake occurred as the anomalies decreased. The anomalies lasted for ~2 months.

(2)

We selected the PSD data in the ELF frequency band detected by the EFD payload in the nighttime without the space magnetic environment perturbations. It was observed that the PSD in the 371 Hz–500 Hz and 700 Hz–871 Hz bands exhibited anomaly perturbations in the vicinity of the epicenter and its magnetic conjugate area on 17 May, with particularly notable anomaly perturbations in the magnetic conjugate area. We observed that anomaly perturbations began to occur ~1 month prior to the occurrence of the earthquake. The earthquake occurred as the anomalies decreased, and this trend continued post–earthquake. Among them, the vertical component perturbations were more notable as the dominant component, and the 700 Hz–871 Hz perturbations were more notable as the dominant frequency band.

(3)

We selected the magnetic field –east–west component vector data detected by the HPM payload, as well as the ion velocity Vx data recorded by the PAP payload for the study. The most notable magnetic field anomaly perturbations were exhibited in the vicinity of the Maduo epicenter and its magnetic conjugate area on 16 May, one week prior to the earthquake, and on 22 May, the occurrence date of the earthquake. In addition, magnetic field anomaly perturbations were also exhibited near the 2021 Ms6.4 Yangbi earthquake on 21 May. Both the magnetic –east–west component data and the Vx data exhibited anomaly perturbations near the epicenter and the magnetic conjugate area on 11 May and 16 May. However, the magnitudes of their perturbations were not consistent across the observations.

(4)

The anomaly perturbations observed in both thermal infrared TBB data and CSES electric and magnetic field data occur within a consistent perturbation time period and spatial proximity, suggesting a potential presence of LAIC relationships. There may be both chemical and electromagnetic wave propagation models involved in the LAIC mechanism related to the Maduo earthquake. It can be approximately hypothesized that three channels in the LAIC models may have varying degrees of influence before and after the earthquake, suggesting that there may be not more than one LAIC mechanism for the preparation and occurrence of an earthquake. In the future, we will integrate data from a broader range of sources and altitudes to investigate different types of earthquakes more comprehensively.