Towards Understanding Earthquake Preparatory Dynamics: A Multi-Parametric Investigation of the 2025 Kamchatka Mw 8.8 Event

Abstract

We present a comprehensive multi-parametric analysis of Lithosphere– Atmosphere–Ionosphere Coupling (LAIC) processes associated with the M = 8.8 earthquake that struck offshore Kamchatka, Russia, on 30 July 2025 (29 July 2015; 23:24:52 UTC). Thermal observations revealed coherent pre-seismic irregularities in near-surface air temperature, relative humidity, and atmospheric chemical potential (ACP), with maximum intensification occurring 1–2 days before the event, followed by rapid co-seismic dissipation and post-seismic recovery. Acoustic channel analysis revealed considerable enhancements in atmospheric gravity wave (AGW) potential energy, as computed from ERA5 reanalysis datasets, 3–5 days prior to the earthquake, with a co-seismic peak and weaker post-seismic irregularities at higher altitudes. Electromagnetic signatures manifested in both lower and upper ionospheric layers. Very-Low-Frequency (VLF) sub-ionospheric propagation from the NPM transmitter, continuously monitored at the PTK (Petropavlovsk-Kamchatsky) station in Kamchatka, Russia, exhibited both positive and negative deviations in amplitude and phase during the preparatory phase. VLF amplitude exhibited wavelike deviations consistent with AGW periods, peaking one day prior to the earthquake. Ionospheric Vertical Total electron content (VTEC) showed coherent pre-seismic maxima 2–3 days before the main shock. Together, these thermal, acoustic, and electromagnetic observations strongly suggest a consistent pre-seismic build-up, co-seismic dissipation, and post-seismic recovery, providing a robust multi-channel imprint of the Kamchatka earthquake and highlighting the importance of integrated multi-parameter approaches for understanding earthquake preparatory dynamics.

1. Introduction

Earthquakes represent one of nature’s most complex geophysical phenomena, involving far more than the simple conversion of strain energy to mechanical energy. The physical processes underlying seismic events encompass an intricate web of pre-, co-, and post-earthquake phenomena that extend well beyond the Earth’s surface and subsurface regions [1,2,3,4,5,6]. Since the 1960s, the understanding of seismic hazard mechanisms has undergone a fundamental paradigm shift. Earthquakes are no longer viewed as isolated geological events but as complex processes that create measurable connections between the Earth’s interior, atmosphere, and outer space environment [7]. This revolutionary perspective has opened new avenues for earthquake research and potential forecasting methods.

The connection between seismic activity and atmospheric phenomena operates through the Lithosphere–Atmosphere–Ionosphere-Coupling (LAIC) mechanism [3]. This framework identifies four primary channels through which pre-seismic and co-seismic processes propagate through the Earth’s atmospheric layers: (a) chemical channel, involving radioactive ionisation and gas emissions; (b) thermal channel, encompassing temperature variations and heat flux changes; (c) acoustic channel, characterised by atmospheric gravity waves and pressure variations; and (d) electromagnetic channel, including radio wave propagation and ionospheric disturbances. These channels exhibit highly non-linear, inhomogeneous, and anisotropic characteristics, necessitating a multiparametric approach to fully understand pre-seismic processes [6].

Following the establishment of the LAIC hypothesis [3,4], research expanded to encompass multiple parameters across different coupling channels. This comprehensive approach has revealed crucial insights into the complex preparation processes of earthquakes. The thermal channel plays a crucial role in pre-seismic activity, often reflected in increases in the Earth’s radiation budget prior to major earthquakes [8,9,10,11,12]. Thermal parameters in this context include variations in surface and air temperature, changes in surface latent heat flux (SLHF), deviations in outgoing longwave radiation (OLR), fluctuations in relative humidity (RH), modulation in atmospheric chemical potential (ACP), and many more. Numerous satellite-based studies have consistently reported notable increases in these parameters during pre-seismic periods, highlighting their potential as valuable indicators of earthquake precursory processes [13,14,15,16].

Thermal irregularities create convection mechanisms in the lower and middle atmosphere, generating buoyancy effects and air parcel movements that characterize the acoustic channel. Atmospheric Gravity Waves (AGWs) serve as the primary energy transport mechanism, carrying disturbances from the lower atmosphere to the stratosphere and mesosphere. During earthquake preparation phases, strain energy accumulation produces temperature modulation, altered thermal conductivity, and pressure variations that generate wave-like structures within AGW frequency ranges. These phenomena can be detected through various ground and space-based techniques, including radars, GPS systems, satellites, and low-frequency radio receivers [17,18,19,20,21,22].

The electromagnetic channel, spanning from subsurface regions to magnetospheric heights, encompasses the broadest range of parameters and exhibits highly complex behavior, with anisotropic and inhomogeneous temporal variations that often lack simultaneous responses even for the same earthquake. Among the most direct indicators are Ultra Low Frequency (ULF) emissions detected near earthquake epicenters, which provide strong evidence of pre-seismic electromagnetic activity [23,24,25,26] and are closely linked to magnetospheric variations through energetic particle bursts in the radiation belt [27,28,29]. In the lower ionospheric altitudes, seismo-electromanetics are widely explored and reported by ionospheric variability during pre- and co-seismic periods by monitoring using sub-ionospheric Very-Low-Frequency (VLF) radio-sounding techniques.

An extensive research has explored seismo-ionospheric perturbations in the VLF/LF domain, establishing their statistical and physical correlation with earthquake activity. Early works demonstrated the evidence of systematic correlations between subionospheric VLF/LF perturbations and seismic events [1,30,31]. Subsequent investigations confirmed these perturbations across different propagation paths and regions, highlighting both statistical consistency and physical interpretation [32,33,34,35,36,37,38]. More recent analyses emphasize quantitative correlation, criticality, and regional variability in seismo-ionospheric irregularities, offering robust evidence of LAIC processes in both case and statistical studies [39,40,41,42].

It is widely reported that Global Navagation Satellite System (GNSS) based measurements, with key parameters ionospheric Vertical Total Electron Content (VTEC), shows characteristic increases before strong earthquakes [43,44,45,46]. Subsequent studies provided statistical and case-based evidence of precursory TEC disturbances in different seismic regions, including Wenchuan, Chile, Nepal, and Tamenglong [47,48,49,50].

Several other important investigations like F-layer critical frequency variations [51], Traveling Ionospheric Disturbances (TIDs) detected via differential VTEC profiles [6,52], and magnetic field and electron density variations observed by Swarm missions [53] are found to be very effective in this channel. Complementary observations, such as GPS-based surface deformation [22] and unusual aerosol concentration [54], further strengthen the evidence of seismogenic effects within this channel.

In the past few years, several authors have emphasized the importance of adopting a multi-parametric approach when studying individual earthquakes in order to identify the most noteworthy and effective research associated with the LAIC mechanism. It is important to note, however, that not all parameters exhibit deviations for every earthquake, and the nature of the perturbations observed is not always directly comparable. Also, the temporal and spatial profiles of pre- and co-seismic perturbations are found to be different for different earthquakes. Recent comprehensive studies have applied multi-parameter approaches to analyze impactful seismic events worldwide. Notable examples include analyses of the 2021 Samos (Greece) earthquake (M = 6.9) by [6,40], as well as extensive investigations of earthquakes in Japan, Italy, Nepal, and China by various research groups [55,56,57,58,59,60,61].

2. Methodology and Data Analysis

The Kamchatka earthquake of 29 July 2025 was a magnitude 8.8 megathrust event in Russia’s Kamchatka Peninsula, occurring at a shallow depth of ∼10–20 km (https://earthquake.usgs.gov/, accessed on 30 September 2025). The rupture, spanning nearly 390 km × 140 km with slips up to 8.7 m, lasted about 3–4 min and was preceded by a Mw 7.4 foreshock on 20 July. Thousands of aftershocks followed, including several above Mw 6. The earthquake triggered a Pacific-wide tsunami warning, with waves up to 3–4 m hitting Kamchatka’s coast, causing flooding and evacuations. It was the largest global earthquake since 2011 and the strongest in Kamchatka since the 1952 Mw 9.0 event, rupturing a long-identified seismic gap in the Kuril–Kamchatka subduction zone.

A VLF observation was conducted at Petropavlovsk-Kamchatsky in Kamchatka, Russia, where signals from the NPM transmitter (23.4 kHz), located in Hawaii (Latitude 21°25′ N; Longitude 158°09′ W), were recorded. Both amplitude and phase information were collected before and during the EQ. We computed the critical radius of the EQ ($R_{cr}={10}^{0.44·M_w-0.78}$ [62]); however, no GNSS-IGS stations were available within the critical zone (CZ). To investigate possible seismogenic signatures, we selected one station, STK200, in Japan, which lies outside the CZ. In addition, we calculated the fifth Fresnel zone between the transmitter and receiver to assess potential deviation within this region. Figure 1 illustrates the locations of the NPM transmitter (cyan square), the VLF receiver (cyan diamond), the fifth Fresnel zone (cyan quasi-elliptical boundary), the IGS station STK200 (yellow diamond), the critical radius (red circle), the tectonic plate boundary (magenta lines), and the earthquake epicenter (red star).

Figure 2 shows the temporal variations of geomagnetic and solar activity parameters during day numbers 195–218 (https://omniweb.gsfc.nasa.gov, accessed on 30 September 2025). The Dst index (panel a) varied between $-30$ nT and $+45$ nT, the $K_p$ index (panel b) exhibited fluctuations mostly in the range 0–5, the 3-h averaged $AE$ index (panel c) showed values from 0 to about 40 nT, the daily average $AE$ (panel d) remained within 0–30 nT, the F10.7 solar flux index (panel e) ranged between 130 and 170 sfu and he IMF $B_z$ component (panel f) fluctuated between approximately $-20$ nT and $+20$ nT. The red dashed vertical line in each panel marks the occurrence of the earthquake on day number 211. Typically, geomagnetic storm conditions are defined by thresholds such as Dst $\le -50$ nT, Kp $\ge 5$, and ap/Ap $\ge 30$, and the available indices confirm that the earthquake occurred during a geomagnetically quiet phase several days after the preceding storm recovery. Although the F10.7 solar flux reached values around 150 sfu, this level represents a smoothly varying component of solar EUV radiation rather than an impulsive driver of ionospheric disturbances. F10.7 evolves gradually over daily to weekly scales and primarily contributes to background ionospheric trends [63]. Observational studies further show that TEC responds smoothly to F10.7 variations without producing short-term anomalies [64]. Moreover, disturbances in TEC or VLF propagation require strong geomagnetic forcing—such as prolonged southward IMF Bz or elevated Kp/Dst—rather than elevated solar flux alone [65]. In the present case, Kp remained $<5$ and Dst $>-50$ nT during the entire pre-seismic interval, indicating that the ionospheric environment was genuinely quiet and that elevated F10.7 values did not contaminate the VLF or TEC signatures observed. Thus, the Kamchatka earthquake occurred under relatively quiet to moderate space weather conditions, suggesting that any ionospheric or electromagnetic perturbations observed during this period were unlikely to be caused by external solar–geomagnetic forcing, thus supporting the interpretation of such perturbations as potential seismogenic effects.

To analyze the perturbations in surface temperature, relative humidity (RH), and the atmospheric chemical potential (ACP), we used datasets provided by the National Oceanic and Atmospheric Administration (NOAA). The reanalysis data were obtained from the NOAA Physical Sciences Laboratory portal (https://psl.noaa.gov/data/gridded/data.ncep.reanalysis.html, accessed on 30 September 2025). Our study focused on the region spanning 65° N–45° N and 150° E–170° E, which corresponds to an area approximately 10° around the earthquake epicenter. Since the NOAA database provides global data in a single file, we extracted the relevant region after downloading the necessary variables. The downloaded NetCDF (.nc) files were converted into CSV format, after which we extracted data for 15 days surrounding the earthquake and within 10° of the epicenter. ACP ($\Delta U$) values were derived from the surface temperature and RH data using the following approach [66]:

$$\Delta U=5.8\times {10}^{-10}{\left(20T_g+5463\right)}^{2}ln\left({\displaystyle \frac{100}{H}}\right),$$

(1)

where $T_g$ and H are the Air Temperature and Relative Humidity. We generate the spatiotemporal maps using the daily average of each parameter for the said period around the the EQ day.

To check the possible AGW excitation, the altitude profiles of temperature around the earthquake were obtained from the ERA5 reanalysis database. Temperature data were collected at 137 discrete geopotential levels, ranging from level 137 (10 m above sea level) to level 1 (79.30 km) (https://cds.climate.copernicus.eu/, accessed on 30 September 2025). The dataset spans 15 consecutive days (seven days prior to the earthquake, the day of the event, and seven days following the earthquake), with measurements available at one-hour intervals.

It is known that mesospheric gravity waves generally have longer vertical wavelengths (up to 30 km) compared with stratospheric gravity waves [67]. However, applying a 2–30 km filter is impractical because this range nearly covers the entire mesosphere (≈45–80 km). Moreover, large-scale features such as temperature inversions at the tropopause and stratopause could be incorrectly interpreted as wave structures. Therefore, in this study, a 2–10 km Chebyshev Type-I digital band-stop filter with 3 dB ripple was applied. This filter separates the background temperature $\overline{T}$ from the observed temperature T [68]. The fluctuation component is then defined as:

$$T^{\prime }=T-\overline{T}.$$

(2)

The Brunt–Väisälä frequency (N), which represents the oscillation frequency of an air parcel in a stably stratified atmosphere, was computed as:

$$N^2={\displaystyle \frac{g}{\overline{T}}}\left({\displaystyle \frac{d\overline{T}}{dz}}+{\displaystyle \frac{g}{c_p}}\right),$$

(3)

where z is the altitude (km), $c_p$ is the specific heat at constant pressure ($1.005\mathrm{kJ}{\mathrm{kg}}^{-1}{\mathrm{K}}^{-1}$), and g is the acceleration due to gravity ($0.0098\mathrm{km}{\mathrm{s}}^{-2}$). The vertical gradient term $\frac{d\overline{T}}{dz}$ in Equation (3) was calculated for each 100 m background temperature layer.

The potential energy per unit mass ($E_p$) associated with gravity waves was then estimated as:

$$E_p=\frac{1}{2}{\left({\displaystyle \frac{g}{N}}\right)}^2{\left({\displaystyle \frac{T^{\prime }}{\overline{T}}}\right)}_{\mathrm{layer}}^2.$$

(4)

The variance term in Equation (4), which depends on altitude, was determined as:

$${\left({\displaystyle \frac{T^{\prime }}{\overline{T}}}\right)}^2={\displaystyle \frac{1}{z_{max}-z_{min}}}{\int }_{z_{min}}^{z_{max}}{\left({\displaystyle \frac{T^{\prime }}{\overline{T}}}\right)}^{2}dz,$$

(5)

where $z_{min}$ and $z_{max}$ are the lower and upper bounds of the altitude range considered. In this study, variance was computed within each 100 m altitude layer. Finally, we plotted the temporal evolution of $E_p$ values in the altitude range of 35–50 km over the earthquake epicenter. Next, we mapped the daily mean $E_p$ values within $\pm 10°$ around the epicenter at an altitude of 46.3 km. This altitude was selected because it showed the most prominent response in the temporal evolution of $E_p$ values [68].

The earthquake epicenter was in such a location that no GNSS-IGS station was available within the CZ of the earthquake. To see the possible ionospheric response (if any) outside the CZ, we selected the nearest operational station, STK200 in Japan located at a distance of ∼1686 km. The choice of STK200 was motivated by its continuous data availability and stable operational history, which are essential for extracting reliable VTEC profiles. Mixed-observable GNSS RINEX files (including GPS, GLONASS, Galileo, and BeiDou) with a 30-s sampling interval were downloaded for each day from 22 July 2025 to 5 August 2025 from the open source archive (https://igs.bkg.bund.de/searchRINEX, accessed on 30 September 2025) and computed the VTEC using the Gopi K. Seemala’s software version 3.5 [69]. To assess seismogenic impression, we analyzed a time window of $\pm 7$ days around the earthquake, thereby incorporating both pre-seismic and post-seismic intervals. For each universal time (UT) epoch, we computed the median ($\chi$) and interquartile range (IQR) of the VTEC values over the 7 days preceding the earthquake, which served as the reference distribution. From these statistics, an upper bound (UB) and a lower bound (LB) were established, providing thresholds to quantify deviations from normal variability. As reported by [70], the median and IQR in a normally distributed dataset approximate the mean ($\mu$) and $1.34\sigma$, respectively. Therefore, a VTEC measurement is considered unusual when it crosses either the UB or LB, corresponding to a confidence level of 80–85% [71]. This methodology ensures that any detected VTEC irregularities are not artefacts of data gaps or natural diurnal variability but instead represent statistically significant departures from the background state [46].

To examine the VTEC profile over the earthquake epicenter, we use an alternative method. We use the Global Ionospheric Map (GIM), which provides a global representation of the VTEC, derived from a worldwide network of dual-frequency GNSS receivers. These maps are routinely generated by the IGS database with a spatial resolution of $2.5°\times 2.5°$ in latitude and longitude, and a temporal resolution of 15 min. Owing to its extensive coverage and continuous availability, GIM serves as a reliable data source for monitoring ionospheric variability on both regional and global scales. In this study, GIM data were employed to extract the VTEC values over the closest point of the EQ epicentre, and a one-dimensional time series of data was extracted from the 3-dimensional modelled grid database. Similar to the STK200 station, we computed the UB and LB and the deviations in the VTEC profiles are examined as a function of day number.

3. Results

3.1. Thermal Irregularities

Figure 3 shows the spatial and temporal evolution of near-surface air temperature (K) over the Kamchatka region during 22 July to 5 August 2025. Each panel displays daily temperature fields, with the earthquake epicenter marked by a black dot. The colour scale ranges from cooler conditions (<283 K, blue) to relatively warmer conditions (>289 K, red), capturing short-term thermal variability potentially linked to pre- and co-seismic processes. In the days leading up to the earthquake, temperatures remained slightly below 285 K over much of the study region (23–26 July), with localized warming beginning to emerge on 27 July. A marked pre-seismic warming scenario occurred on 28–29 July, when temperatures locally increased by approximately 2–3 K above those of the preceding days, peaking at nearly 289 K in the southern and central parts of the domain. On the day of the earthquake (30 July), temperatures remained elevated but began to decrease slightly in the epicentral vicinity. The post-earthquake period shows a clear enhancement in temperature on 31 July, a day after the event, followed by a gradual return toward near-background levels after 2 August. However, a weak temperature background persisted through 4 August. As mentioned in the previous section, the mainshock was followed by a series of aftershocks from 31 July to 4 August, with magnitudes ranging from 6.0 to 6.8. The continuation of the increase in the thermal energy budget is reflected in the overall profile of the temperature, and that is exactly what occurred in this case.

Figure 4 shows a similar spatio-temporal variation of relative humidity (RH) during the earthquake. RH remained consistently high over much of the region, exceeding 97–99% around 23–24 July, suggesting near-saturated conditions. A distinct drop of RH is observed on 27–28 July, with RH decreasing to 85–88% near the epicentral area, producing a pronounced negative anomaly. Moisture levels partially recovered on 29–30 July, with RH above 94–96%, peaking on the earthquake day (30 July) in the vicinity of the epicenter. The RH anomaly pattern suggests a strong coupling between local atmospheric moisture and the pre-seismic processes. The sharp decline on 27–28 July may indicate enhanced vertical mixing or increased outgassing, leading to local drying of the lower troposphere. In contrast, the re-humidification on 29–30 July could reflect moisture convergence associated with pre-seismic atmospheric instability. Post-seismic suppression of RH is consistent with a dissipative atmospheric response after fault rupture, as well as due to the similar conditions of the main shock, which originated from a series of strong aftershocks.

Figure 5 depicts the spatio-temporal evolution of Atmospheric Chemical Potential (ACP, eV), similar to Figure 3 and Figure 4. The ACP ranges from near-background levels (<0.002 eV) to strongly elevated values (>0.014 eV), providing a clear visualization of atmospheric energy fluctuations associated with the seismic cycle. Clear pre-seismic enhancement is observed beginning on July 26, with ACP increasing from ∼0.004 eV to peak values exceeding 0.015 eV on 28 July. This anomaly is spatially coherent over a 200–300 km area centered near the epicentral zone, suggesting a localized seismogenic effect. The enhancement persisted through 29 July, after which ACP sharply decreased on 30 July, with values dropping to nearly background levels (<0.003 eV) over the source region. Post-seismic recovery is evident between 31 July and 4 August, with ACP stabilizing around 0.006–0.008 eV before gradually returning to background conditions.

The pre-seismic ACP anomaly represents an increase of approximately 250–300% above the background level, consistent with enhanced ionization or trace gas release into the near-surface atmosphere prior to fault rupture. The sharp co-seismic drop suggests rapid dissipation of this accumulated energy, potentially triggered by crustal failure. Such behavior aligns with previously reported atmospheric irregularities preceding major earthquakes, reinforcing the potential role of ACP as a robust proxy for pre-seismic lithosphere–atmosphere interactions and as a candidate parameter for earthquake preparatory dynamics.

3.2. Acoustic Irregularities

Figure 6 shows the temporal–altitudinal variation of the $E_p$ of AGW, derived from ERA5 reanalysis, over the epicentral region of the EQ. The color shading and contours represent the daily averaged $E_p$ (J/kg) between 35–50 km altitude, spanning the period 22 July to 5 August 2025. A pronounced enhancement of AGW potential energy is observed between 25–27 July, peaking at values above 19–20 J/kg around 46–49 km altitude on 26 July. This anomaly occurs just 3–5 days prior to the earthquake, suggesting distinctive atmospheric variability in the middle to upper stratosphere above the epicentral region. On 29–30 July, coinciding with the earthquake event, another notable increase is observed, with $E_p$ exceeding 18 J/kg near 46–48 km altitude. The background conditions, in contrast, remain considerably lower at ∼0.5–8 J/kg below 42 km altitude. After the event, enhancements appear sporadically (e.g., 31 July and 2–4 August), but these are weaker (around 8–14 J/kg) and displaced in altitude relative to the pre-seismic maxima. The persistence of strong perturbations in the days leading up to and during the earthquake, followed by a decline, supports the presence of atmospheric perturbations temporally linked to seismic activity.

Figure 7 presents the spatiotemporal evolution of the $E_p$ at an altitude of between 46–49 km. Each panel displays the daily averaged distributions of $E_p$, in J/kg, from 22 July to 5 August over the Kamchatka region, with latitude ranging from ∼41° N to 63° N and longitude from 150° E to 170° E keeping the EQ epicenter at the centre. The colormap shows variations from near-zero background levels up to ∼20 J/kg, highlighting regions of strong AGW activity. Notably, a pronounced increase in $E_p$ is observed around 25–27 July, peaking around 20 J/kg on 26 July, just days prior to the earthquake, suggesting deviant wave energy propagation in the atmosphere. A moderate intensification of $E_p$ is observed on 29 July. There are some examples of $E_p$ enhancement observed on 30–31 July, 2 and 4 August 2025, but the region is far from the epicenter region. the elevated $E_p$ values exceeding 18–20 J/kg coincide spatially with the seismic region, while surrounding days exhibit background levels closer to 8–12 J/kg, providing quantitative evidence of perturbations potentially linked to the EQ.

3.3. Electromagnetic Irregularities

3.3.1. VLF Perturbations

Figure 8 and Figure 9 show the NPM nighttime differential amplitude and phase from 21 July to 30 July. The term ’differential’ refers to the difference of the signal (amplitude or phase) during the day of interest from the monthly average, calculated using undisturbed days [72]. The black dashed line marks the $\pm 2\sigma$ level, computed along with the monthly average. The shaded areas highlight intervals where the signal exceeds this threshold. On 21–22 July, both amplitude and phase slightly surpass the $2\sigma$ level. On 23–24 July, clear deviations appear in both parameters. On 25 July, the signal remains largely undisturbed. On July 26, the amplitude and phase exceed two standard deviations for about two hours. On 27 July, both characteristics show minor disturbances.

In the following two days, the amplitude and phase behave differently. On 28 July, the amplitude remains within the dispersion limits, while the phase shows apparent disturbances and a wavelike structure. On 29 July, the amplitude rises well beyond the dispersion boundaries, but no noticeable perturbations are seen in the phase. Both positive and negative deviations are present in amplitude and phase across the interval. Additionally, a wavelike structure in the phase is observed on the evening of the earthquake day.

To investigate possible wavelike structures in the nighttime VLF amplitude, we applied both Fourier and wavelet analyses. As noted in the introduction, atmospheric gravity waves (AGWs) can be generated before and during earthquakes, and their signatures may be extracted from VLF signals perturbed by seismogenic disturbances in the lower ionosphere. For the AGW analysis, only nighttime VLF data were considered, since during daytime the effects of seismic activity are masked by the dominant influence of solar radiation. The dataset from 21 July to 5 August was analyzed using both fast Fourier transform (FFT) and wavelet methods. The FFT analysis was performed with a rectangular data window to obtain the normalized FFT spectrum. For the wavelet analysis, we use Morlet mother Wavelet function to keep the wavelet scale almost equal to the period of FFT.

Figure 10 presents the normalized FFT spectrum of nighttime VLF amplitude variations for the period 21 July to 5 August. The X-axis indicates the oscillation period (in minutes), while the Y-axis shows the normalized Fourier amplitude. Each curve corresponds to one night of observations, allowing for direct comparison of spectral characteristics over the study interval. Distinct enhancements in spectral amplitude are visible on 29–30 July exhibit strong peak with amplitudes exceeding 0.8, particularly within the 40–120 min period range. These periodicities are consistent with the typical timescales of AGW. Some comparatively lower magnitude oscillation was also observed on 22–23 July. On most other nights, the spectra remain relatively flat, indicating the absence of strong periodic modulations in the VLF signal.

Figure 11 presents the wavelet power spectra of nighttime VLF amplitude variations for the period 21 July to 5 August. Each panel corresponds to one night of data, showing the temporal evolution of oscillation power across different periods (in minutes). X-axis axis represents time of the observation in hours after 20:00 Local Time (LT) and the Y-axis represents the oscillation periods in minutes. The colour scale indicates the normalised power. The dashed white curves mark the cone of influence (COI), within which the results are considered statistically reliable. Clear enhancements in wavelet power are observed on two nights, particularly 22–23 July with weaker spectral power and with the most prominent power on 29–30 July. On these dates, strong oscillations appear within the 60–120 min period range, consistent with AGW activity. There are some other scattered wave intensification observed on 1–4 August 2025, but they are not reliable as they are outside of the COI. Therefore, both the FFT and wavelet show a prominent seismogenic impression before the EQ.

3.3.2. VTEC Perturbations

Figure 12 shows the temporal variation of the Vertical Total Electron Content (VTEC) recorded at the STK200 station from Day of Year (DoY) 203 to 217. In the top panel, the black curve denotes the observed VTEC profile, while the red and green curves represent the upper (UB) and lower (LB) bounds, respectively. An arrow indicates the EQ day (DoY 210). The bottom panel presents the corresponding differences in VTEC, expressed as deviations from the expected background values. Localized enhancements are observed on DoY 204, 211, and 213; however, these variations appear relatively sporadic and do not reveal a coherent pre- or post-seismic signature. Unlike thermal, acoustic, and other electromagnetic parameters, which display clearer perturbations on the days preceding the earthquake and the consistent fluctuations due to the effects of the aftershocks, the VTEC deviations remain inconsistent. The primary reason could be the distance of the IGS station from the epicentre, for which the ionospheric disturbances are outside of the CZ and thus the VTEC has no such consistent effect, which is not seismogenic in nature.

Given this large epicentral distance, it is reasonable that the VTEC does not exhibit distinct pre- or co-seismic perturbations, as this aligns with established observations that TEC anomaly amplitudes systematically decay with distance from the source, often following an exponential or power-law attenuation trend. This spatial attenuation effect has been independently confirmed by [73,74]. Accordingly, while the TEC variations observed at STK200 may reflect regional ionospheric fluctuations and are not seismogenic in nature.

Figure 13 illustrates the temporal variation of the VTEC profile, analogous to Figure 12. Unlike Figure 12, the GIM results here appear more convincing. A sharp enhancement in VTEC is observed on DoY 208 (about 8 TECU) and on DoY 209 (about 6 TECU). On the earthquake day (DoY 210), the increment is comparatively smaller, with an increase of around 3 TECU. Notably, the pre-seismic VTEC enhancement is quite prominent. Following the earthquake, the VTEC exhibits a substantial rise on DoY 211, though with slightly lower values compared to DoY 208. This behavior closely resembles the thermal and AGW perturbations, which also show prominent increases immediately after the earthquake, likely due to a series of aftershocks.

The spatiotemporal GIM anomaly as computed from GIM is shown in Figure 14. A clear, time-localised perturbation is observed beginning several days before the EQ and evolving through the first week of August. The daily anomaly maps indicate positive VTEC values reaching peak amplitudes of order 10–15 TECU on 25–27 July, contrasted with the much smaller background/quiet values of order 0–3 TECU over most other days. On the co-seismic window, (29–30 July), the pattern is mixed. The epicentral region itself is dominated by lower perturbations (∼2–5 TECU), while a strong positive patch of 8–12 TECU appears displaced to the north-east of the epicenter. in the post-EQ period, (31 July–5 August), positive deviations recur intermittently, again with amplitudes of a few to ∼10 TECU, but with changed location and vertical/areal extent relative to the pre-seismic maximum.

Meteorological records for the late-July 2025 period indicate no major weather disturbances affecting the Kamchatka Peninsula. Active tropical systems in the Western North Pacific, including Typhoon Wipha, remained far to the south and did not influence the region (https://www.jma.go.jp, accessed on 30 September 2025). Synoptic surface analyses from JMA demonstrate a typical summertime pattern with migrating extratropical lows and frontal systems over the North Pacific, without any indication of a rapid pressure drop or intense storm development near Kamchatka during 25–31 July (https://www.jma.go.jp, accessed on 30 September 2025). Global climate reports for July 2025 describe above-average temperatures and enhanced Indo-Pacific convection associated with ENSO variability (https://www.ncdc.noaa.gov, accessed on 30 September 2025), but no localized extreme meteorological event capable of explaining the observed RH and thermal transitions. Additional checks using ERA5 reanalysis confirm the absence of strong cyclonic activity or unusual synoptic-scale features over the study region during this period (https://www.ecmwf.int, accessed on 30 September 2025). Thus, the outcomes are free from any substantial methodological contamination.

4. Discussion

In this study, we found that the observational period of April–June 2025 was marked by strong geomagnetic disturbances and frequent seismic activity, making it unsuitable for establishing quiet-time baseline conditions. Major geomagnetic storms occurred on 16 April (Dst $- 138$ nT, Kp $\approx 7.7$), 1 June (Dst $- 118$ nT, Kp $\approx 8$), and 13 June (Dst $-101$ nT), while several moderate to large earthquakes were recorded in the Kamchatka–Kuril region during the same interval (e.g., 18 April, M5.4; 13 June, M6.1; 20 July, M6.5–7.4). The simultaneous occurrence of these disturbances suggests that the background geophysical state was highly perturbed, rendering statistical averaging or anomaly detection unreliable due to the overlapping effects of geomagnetic and seismic disturbances. Therefore, this study adopts a case-specific approach, emphasizing consistent multiparametric responses associated with the selected event and supporting the physical plausibility of the proposed LAIC mechanisms.

The thermal parameters reveal a coherent set of irregularities in near-surface air temperature, RH, and ACP associated with the earthquake. The near-surface air temperature showed a pre-seismic warming phase, rising by ∼2–3 K on 28–29 July to nearly 289 K, before slightly decreasing on the earthquake day and peaking again during the aftershock sequence on 31 July. RH exhibited a complementary pattern, with near-saturated conditions (97–99%) on 23–24 July dropping sharply to 85–88% on 27–28 July near the epicenter, followed by strong re-humidification to above 94–96% on 29–30 July, and post-seismic suppression thereafter. The most pronounced signal appeared in ACP, which increased from background levels of ∼0.004 eV to more than 0.015 eV (∼250–300% enhancement) on 28 July, before collapsing to <0.003 eV on the earthquake day and stabilizing at ∼0.006–0.008 eV in the following days. Together, these irregularities support a coherent pre-seismic build-up of atmospheric energy and instability, rapid co-seismic dissipation, and post-seismic recovery.

In the acoustic channel, the AGW potential energy ($E_p$) as computed from ERA5 exhibited distinct seismogenic impression. The altitude profile shows a strong pre-seismic enhancement on 25–27 July, with $E_p$ peaking at 18–20 J/kg near 46–49 km, compared to background levels of only ∼0.5–8 J/kg below 42 km. A second amplification occurred during the co-seismic window on 29–30 July, again exceeding 18 J/kg, while post-seismic fluctuations on July 31 and August 2–4 were weaker (4–5 J/kg) and displaced in altitude relative to the pre-seismic maxima. The spatiotemporal maps at 46–49 km confirm these patterns, showing coherent enhancements centered near the epicenter on 25–27 July (around 18–20 J/kg on 26 July), moderate intensification on 29 July, and later deviations on 30–31 July and 2 and 4 August occurring farther from the epicentral area.

The electromagnetic perturbations show different characteristics in lower and upper ionospheric altitudes. At lower ionospheric heights, the nighttime NPM VLF amplitude and phase show prominant deviations prior to the EQ. Small deviations above the $\pm 2\sigma$ level occur on 21–22 July, followed by more pronounced fluctuations on 23–24 July. After a relatively quiet July 25, brief excursions are observed on July 26, with minor disturbances on 27 July. On 28 July, amplitude remains within normal limits while phase shows prominent wavelike fluctuations, and on 29 July, amplitude rises well above the dispersion boundaries while phase remains stable. Both positive and negative deviations occur, with wavelike structures in phase particularly evident on the evening of the earthquake. Spectral analyses reveal strong periodicities in the 40–120 min range on 29–30 July, consistent with AGW timescales, with weaker oscillations on 22–23 July and mostly flat spectra on other nights. Wavelet analysis confirms these results, showing enhanced power within the 60–120 min band on 29–30 July, while other scattered intensifications fall outside the cone of influence and are not statistically reliable.

In the upper atmospheric layers, the VTEC variations recorded at the STK200 IGS station exhibit sporadic and inconsistent deviations, with localized enhancements of approximately 3–5 TECU on DoY 204, 211, and 213. These deviations do not form a conventional pre- or post-seismic pattern for a single EQ and aftershocks, likely due to the large distance from the epicenter and outside of CZ. The GIM-derived VTEC shows more pronounced signals, with clear pre-seismic enhancements reaching ∼8 TECU on DoY 208 and ∼6 TECU on DoY 209, while the EQ day (DoY 210) exhibits a smaller increment of ∼3 TECU. Post-seismic increases of ∼6 TECU are observed on DoY 211, closely matching the temporal behavior of thermal and AGW excitations, likely linked to aftershock activity. Spatiotemporal GIM maps reveal time-localized positive deviations of 10–15 TECU several days before the earthquake (25–27 July), with the epicentral region showing lower co-seismic signature of 2–5 TECU and displaced maxima of 8–12 TECU to the northeast. During the post-seismic period (31 July–5 August), positive deviation recur intermittently with amplitudes up to ∼10 TECU and varying spatial distribution.

Our multi-parametric analysis of the Kamchatka earthquake reveals consistent pre-seismic signatures across thermal, acoustic, and electromagnetic channels, consistent with the LAIC mechanism. In the thermal channel, near-surface air temperature, relative humidity, and ACP exhibited maximum pre-seismic intensification during 28–29 July (1–2 days prior), followed by co-seismic dissipation and post-seismic recovery. In the acoustic channel, AGW potential energy showed strong pre-seismic enhancement during 25–27 July (3–5 days prior), with a co-seismic peak on 29–30 July and weaker post-seismic irregularities in the following days. Electromagnetic perturbations were observed in both lower and upper ionospheric layers: VLF amplitude and phase showed prominent deviations on 23–24 July and 29 July (around a week). The wavelike structures, similar to the AGW wave period in VLF amplitude, exhibit a very short-term intensification, with the maximum wave intensity observed on 29 July (one day prior). The GIM-derived VTEC exhibited coherent pre-seismic maxima on 27 July (3 days prior). In contrast, IGS-station VTEC remained sporadic and inconsistent due to its distance from the epicenter. Overall, the combined thermal, acoustic, and electromagnetic observations demonstrate a consistent multi-channel pre-seismic build-up, rapid co-seismic dissipation, and subsequent post-seismic recovery, providing a clear seismogenic imprint of the Kamchatka earthquake.

The observed disturbances reveal a physically coherent coupling chain connecting lithospheric, atmospheric, and ionospheric processes within the LAIC framework. The pre-seismic enhancement in atmospheric parameters such as ACP, temperature, and relative humidity indicates increased thermodynamic and ionization activity in the lower atmosphere, likely caused by radon emanation and latent heat release during crustal stress accumulation. This thermal and ionization build-up promotes buoyancy-driven convection, generating upward-propagating atmospheric gravity waves (AGWs). The observed enhancement in AGW potential energy thus reflects mechanical energy transfer to the stratosphere, while the subsequent VLF amplitude and phase disturbances represent ionospheric responses to these AGW-induced density fluctuations and electric field perturbations. Together, the consistent timing and spatial correspondence of ACP, AGW, and VLF fluctuations support a diffusion-type, multi-layer coupling mechanism characteristic of the LAIC process.

Recent research underscores the value of multi-parameter precursor detection in the lithosphere–atmosphere–ionosphere coupling (LAIC) framework. For example, ref. [75] demonstrated synchronized atmospheric and ionospheric perturbations detected via machine learning techniques, offering a methodological advance in precursor identification. Ref. [76] found GNSS-TEC disturbances at epicentral sites, while [77] further refined the detection of mid-scale AGWs from GNSS/atmospheric temperature profiles. Ref. [78] presented a comprehensive multi-parametric and multilayer analysis of the 2019 Mw 7.2 Kermadec Islands earthquake, establishing LAIC processes through integrated seismic, atmospheric, and ionospheric observations. Ref. [79] quantitatively assessed spatiotemporal TEC patterns preceding large earthquakes, establishing distance- and magnitude-dependent scaling of perturbations. Together, these works establish the interconnected behaviour of thermal, acoustic, and electromagnetic precursors and indicate how systematic cross-parameter analysis yields deeper insight than single-channel studies.

Building on these interrelated observations, our study contributes by quantifying magnitude scaling, mapping the coupling sequence of ACP → AGW potential energy → VLF/TEC fluctuations, and improving temporal alignment across layers. Unlike many earlier studies that reported TEC or thermal changes in isolation, we provide coherent vertical coupling from surface thermodynamics (ACP) through stratospheric acoustic-gravity waves (∼18–20 J kg⁻¹) to lower/upper ionospheric signals. Methodologically, our use of high-resolution near-surface meteorological data, ERA5-derived AGW potential energy, and dense propagation-path VLF analyses strengthens the chain of evidence. By comparing anomaly magnitudes (∼250–300% for ACP; ∼18–20 J kg⁻¹ for AGW) with earthquake magnitude and proximity, our work advances previous empirical scaling hypotheses and sets a higher standard for multi-parameter seismo-precursor studies.

5. Conclusions

The multi-parametric investigation of the Kamchatka M_w 8.8 earthquake provides a unique opportunity to examine the complex interplay of lithospheric, atmospheric, and ionospheric processes preceding a major seismic event. Such a study is crucial because different channels of the Lithosphere–Atmosphere–Ionosphere Coupling (LAIC) mechanism—thermal, acoustic, and electromagnetic—exhibit distinct spatio-temporal signatures, with variations in pre-seismic timing and intensity. By analyzing multiple parameters simultaneously, including near-surface temperature, RH, ACP, AGW, and VTEC, we can obtain a coherent picture of pre-seismic build-up, co-seismic dissipation, and post-seismic recovery. We were fortunate to study this earthquake during a period of moderate solar activity, while geomagnetic conditions remained quiet enough to minimize solar-terrestrial contamination. This allowed us to isolate genuine seismogenic impression without interference from external space weather effects. Similarly, potential disturbances from notable meteorological phenomena, such as typhoons or thunderstorms, were absent during this period. In this case, the scarcity of GNSS stations within the defined Critical Zone necessitated the use of an alternative, indirect approach. By employing the GIM database, we extracted VTEC information over the epicentral region, effectively simulating a virtual GNSS–IGS station. The resulting patterns are consistent with previous findings, confirming the reliability of our approach. Although formal error quantification was not performed here, the GIM product itself provides a nominal accuracy of 2–5 TECU [80], which is within acceptable limits for detecting pre-seismic ionospheric disturbances. Hence, the observed variations can be considered robust within the known uncertainty range of the dataset.

The temporal differences observed among various pre-seismic parameters arise naturally from the physical laws governing earthquake preparation and energy migration. As stress accumulates in the lithosphere, mechanical deformation and frictional instability develop according to the principles of rock mechanics and fracture dynamics [81,82,83]. These processes follow the fundamental laws of elasticity, thermodynamics, and energy conservation, leading to the gradual accumulation and eventual release of strain energy. Consequently, thermal, acoustic, and electromagnetic parameters exhibit their maximum unusual intensifications at different times before the earthquake. This variation reflects both the distinct sensitivities of each parameter and the complex timing of multi-physics energy transfer across layers, emphasizing the necessity of a multi-parametric approach to capture the complete seismogenic process.

In summary, acoustic irregularities appeared first about 3–5 days before the earthquake, followed by thermal responses in near-surface air temperature, relative humidity, and ACP about 1–2 days before. Electromagnetic perturbations in both VLF and VTEC occurred around 1–3 days prior, indicating distinct times and delays of migration of seismogenic energy through the atmosphere–ionosphere system, consistent with the lithosphere–atmosphere–ionosphere coupling mechanism. Overall, this multi-parametric approach enhances our understanding of earthquake preparation processes, providing insights into the migration of seismogenic signals and offering a robust framework for future observational and modelling studies of seismic hazards. The complex temporal variations among different pre-seismic parameters pose critical challenges for numerical modeling of earthquake preparatory processes. Each parameter responds differently to the underlying lithospheric and atmospheric perturbations, making it difficult to develop a unified model that accurately reproduces all observed fluctuations. For ultimate understanding of earthquake preparation, it is essential to examine all available parameters in real-time within a uniform dashboard, enabling coherent multi-channel analysis. Such an integrated framework will be implemented in the near future to enhance insights into seismogenic mechanisms. For a clearer and more robust understanding of LAIC-related multiparametric behavior, we plan to undertake a more extensive statistical investigation in future work. Such an analysis will allow us to examine long-term variability and distribution characteristics in greater depth, leading to more objective and reliable interpretation of seismic–atmospheric–ionospheric interactions [84].