Volcanic activity monitoring by unique LIDAR based on a diode laser

The early detection of volcanic activity is a central goal of volcanology; however, reliable prediction methods are still absent [1, 2]. Typically, volcanic activity is accompanied by local seismic tremors, while fracture-emitted aerosol emission can occur a few hours before the eruption [3, 4]. For example, in January 1993, a strong emission of aerosols was detected 6 h before the eruption of the Galeras volcano (Colombia), while six seismometers failed to detect any unusual seismic activity before the event [5]. It was suggested that volcanic activity can be monitored by the changes of aerosol emission near the volcano slopes. When the crater remains 'sealed' between eruptions, volcanic gases in the magmatic feeding system tend to seep through cracks of different sizes in the vicinity of the volcanic dome. In fact, the monitoring of volcanic gas (hydrogen, helium, carbon dioxide, methane, radon etc) emanation is a challenging task due to different factors. The first problem is a challenge to locate the most intense volcanic gas flows: a system of cracks and holes inside the volcano slope is extremely complicated and dynamic by its nature. Thus, it is difficult to find 'a sensitive place' to install gas detectors in the vicinity of 'the strong gas emission points'. The second problem is that the volcanic gases are mixing with the air fluxes, and thus only a fraction of volcanic gas contributes to the signal of a gas sensor. The third problem is that meteorological changes usually shield the volcanic gas emission. Summing up all these factors, it becomes clear that a tunnel near the volcano could be a good place to install volcanic gas detectors. To predict volcanic activity, one needs to carry out dynamic measurements. Thus, a tunnel with volcanic gases has to be slightly ventilated to prevent the air saturation with gases as radon and to avoid detecting the increasing gas concentration. The above-mentioned problems (where to detect and how to maximize the signal) are also important for the instruments installed inside a tunnel. A good way to accomplish these goals is to map the tunnel area at multiple locations. This can be accomplished with the optical technology called LIDAR (Light Detection and Ranging), which is capable of detecting different gases or aerosols by time and space-resolved measurements [6, 7]. A typical LIDAR consists of a transmitter (most commonly laser) and a detector, and operates as follows: the transmitter emits a light pulse which propagates through the medium; on its path the emitted photons may scatter in different directions or can be absorbed by the medium. The photons scattered backwards are registered by the detector and the LIDAR's circuits measure the time delay between the laser pulse emission and the backscattered photon registration, which allows calculation of the distance from the LIDAR to non-transparent objects located on the sensing track. Specially designed LIDARs (i.e. having highly sensitive detectors) are also capable of measuring the distribution of weakly scattering media along the sensing track (e.g. fogs, smoke plumes and other aerosols) [8, 9]. Much effort has been made to use LIDARs in volcanological studies [3, 10–12]. Generally, these studies were performed outdoors using LIDAR systems with very high laser pulse energies (up to several joules per pulse), which are highly dangerous for human eyes.

In the current study we suggest using a LIDAR for the fracture-emitted aerosol monitoring in an underground laboratory in the hot tunnel over the Elbrus volcano magmatic core. As the LIDAR beam can be directed along the tunnel floor, photons will scatter at aerosol particles emanating from multiple cracks and can be summed over a long distance (up to few 100 m), increasing the sensitivity to low aerosol concentrations. Specifically, for this experiment we developed a highly sensitive aerosol LIDAR. Due to safety protocols for human eye protection, our instrument's transmitter energy density was kept below 1 μJ cm⁻² [13]. In a deep isolated tunnel, the LIDAR instrument would provide a highly sensitive monitoring of space-resolved fracture-emitted aerosol variations that should be multiple times more sensitive compared to single spot gas detection by conventional sensors.

The original LIDAR (figure 1(b)) based on a semiconductor (GaAlAs) laser was specifically designed to meet eye safety protocols [14]. Conventional LIDARs operate with laser pulses of high energy to improve the remote sensing sensitivity and range limits. In contrast, we have developed a LIDAR that emits low-energy pulses at a high repetition rate, and the detector (single photon avalanche photodiode, SPAD) counts single photons. The laser pulse emission triggers 'opening' of the detector (the voltage is turned on) in order to get it ready for photon registration, and if a single photon is captured, the detector goes to 'a closed state' (the voltage is turned off). The described cycle is repeated at 4 kHz frequency to build the backscattered photon range distribution (so-called 'histogram'). During a single measurement the laser emits 100 000 pulses of 3 ns duration at 910 nm wavelength. To suppress noise emissions, we have used an interference optical filter (with 3 nm FWHM at 910 nm) before the detector. The measurements were carried out every 30 min and the resolution along the LIDAR sensing track was 10 cm. To improve the stability of LIDAR measurements we control the laser emitter and the SPAD detector temperature using a small thermostat based on Peltier thermoelectric elements. The experimental site with the set of instruments featuring the LIDAR sensor (figure 1) was located in a hot (ambient temperature ∼40 °C) auxiliary adit of a deep underground laboratory hosted by the Baksan Neutrino Observatory of the Institute for Nuclear Research and was operated in collaboration with the Schmidt Institute of Physics of the Earth (IPE), Russian Academy of Sciences [15].

Zoom In Zoom Out Reset image size

Laboratory premises are represented in the form of a side streak (length ∼50 m, diameter ∼3 m, north-south orientation) blasted in host rocks with walls and floor left untreated, ∼4000 m horizontally from the entrance to a system of adits of the underground research facility. The LIDAR was mounted on a flat instrumental platform ∼40 cm above the floor on a rocky base at ∼32 m distance from the adit dead end and ∼20 m from the entrance. The laser beam was aligned ∼3° upwards relative to the adit floor.

Several different sensors were also installed inside the adit so that the LIDAR data could be compared with other parameters of the laboratory environment. One of the sensors was the precision thermometer (0.005 °C resolution) developed at the IPE Laboratory for fundamental problems of ecological geophysics and volcanology [16]. The precision thermometer was installed on the geophysical instrumental basement located 15 m from the dead end in the middle of the adit (concrete cube in figure 1(a)). The temperature data were recorded with 0.5 Hz quantization frequency and stored on a remote server with public access option. Atmospheric pressure, humidity and temperature in the tunnel were also quantified every 30 min by a digital weather station (incorporating barometer, hygrometer and a thermometer, figure 1(b)) installed near the LIDAR.

Radioactive radon gas emanates through cracks in the rock [17, 18]. The radon accumulation in closed chambers such as an underground geophysical laboratory can affect highly sensitive LIDAR measurements due to the light emission induced by the radon spontaneous radioactivity, so we also measured the radon concentration every hour by a commercially available radon sensor (RADEX MR107). Due to its design the developed LIDAR had a valuable feature: the sensitive detector (single photon avalanche photodiode) cannot be damaged by bright light (i.e. during personnel operation in the adit) and is ready for normal measurements within few minutes after lights switching off.

The LIDAR installation location was chosen specifically since the experimental site is positioned directly over the deep magmatic source in the lower crust [19], so the process of hot fluid flow (water vapour, carbon dioxide, radon, methane, hydrogen, helium etc) including some of the deep origin can be observed there. The mentioned fluids emanate into the adit through crustal cracks and pores in the surrounding rock, heating the air and saturating it with aerosol particles [20]. The excess pressure in the adit due to gas emanations creates a general air outflow through the main entrance and prevents the colder outside air entering the laboratory.

Long-term monitoring of the Earth's crust deformation in the observatory is carried out by the 75 m long Michelson laser interferometer supervised by the Sternberg Astronomical Institute [21, 22]. The instrument is located ∼500 m from the entrance to the underground research facility and ∼3500 m from the LIDAR installation site. The Michelson laser interferometer-strainmeter provides the magnitude and the sign of the local Earth's crust deformation. This information is essential for fundamental studies of the Earth's rotation and tidal waves [23]. The absolute values of the local expansion/compression of the Earth's crust provided by the strainmeter were used in our study for comparison with variations of the aerosol emission measured by the LIDAR.

A typical histogram of the photocounts distribution with a resolution of 10 cm along the LIDAR sensing track is shown in figure 2. Laser photons are scattered by aerosols along the entire track from the LIDAR sensor to the adit wall (figure 2, dashed rectangle on the left), while photons scattered from the adit wall represent a distinct peak in the histogram at a distance of about 32–35 m (vertical dashed rectangle in figure 2). Photocounts recorded for distances greater than 35 m correspond to the intrinsic noise of the LIDAR's detector because there are no laser photons received from such distances. During the LIDAR data processing, the signals were determined as follows: backscattering by aerosols—the total amount of photocounts received at distances of 0–30 m, round-trip transmission signal (or just round-trip signal)—the total amount of photocounts received at distances of 32–35 m, noise—the total amount of photocounts for distances of 100–200 m.

Zoom In Zoom Out Reset image size

Thus, the LIDAR histogram processing provides three values: the aerosol backscattering signal, the round-trip transmission signal and the noise level (indicator of the photodetector stable operation). The LIDAR signal time series are presented in figure 3(a). Aerosol backscattering and round-trip signals are synchronous but negatively correlated: when the aerosol concentration in the air increases, the number of photons scattered by the aerosol particles grows as well, so the number of photons reaching the adit wall and returning back to the detector decreases accordingly. The signals correlation coefficient (figure 3(b)) equals 0.961, from which we deduce that LIDAR photons are elastically scattered at aerosol particles or at the adit walls, and the absorption by gas molecules can be neglected. Furthermore, a strong negative correlation between aerosol backscattering and round-trip transmission (figure 3(b)) confirms that the observed phenomenon is not a technical issue, but the process of aerosol emanation weakening.

Zoom In Zoom Out Reset image size

For the first time, we detected a two-fold decrease in aerosol emanations in the underground laboratory within two months (August 25–27 October 2019). During the observation period, the noise level remained insignificant and merely constant (lower line in figure 3(a)). Three short bursts on the noise curve correspond to the personnel visiting the laboratory in August and October, which was accompanied by the lights turning on.

The LIDAR data, the results of Earth's crust deformation (laser strainmeter) and different sensors measurements (radon radioactivity, air humidity and temperature) are compared with each other in figure 4. The radon emanation (thick line) and variations in relative air humidity (thin line) were rather noisy, but a general decreasing trend of humidity and radon level can be observed (figure 4(a)). On some days, for example, August 31, September 14, September 19, and October 21, the coincidence of increased humidity and radon emissions is rather noticeable. Remarkably, the radon and water vapour signals coincide by sign and by magnitude (∼1%–3%) over two months of measurements. The same dependence was previously observed at the Gran Sasso observatory (Italy) [18]. Supposing that radon gas is produced by volcanic fluids originating from the Elbrus volcano magmatic feeding system, a small decrease of air temperature in the adit should appear. Air temperatures measured above the concrete cube in the hot tunnel (the precision thermometer) and near the laser strainmeter (3500 m closer to the exit from the underground research facility) are compared in figure 4(b). One may notice that both thermometers demonstrate the temperature decrease over two months but with a 40-fold difference in absolute values. Thus, the climatic temperature oscillations outside the underground observatory do affect the sensor near the strainmeter (daily variations are 3–4°С), while in the isolated adit similar variations are almost absent (<0.1°С).

Zoom In Zoom Out Reset image size

To explain the observed temperature and fracture-emitted aerosol variations we suggest the following model. Since there is no chance for atmospheric air transfer into the isolated adit, the ambient temperature decrease in the experiment is most likely due to the reduced flow of hot volcanic fluids coming from large depths through crustal pores and cracks. The fluid flow reduction is in turn caused by the Earth's crust compression. This hypothesis fits well with the observed two-fold decrease in aerosol emission (figure 3(a)): decreased volcanic fluids flow brings fewer aerosol particles into the adit air, which is detected by the LIDAR. The inset in figure 4(c) shows several examples of aerosol emission bursts (thick line) coinciding with temperature rising (thin line from figure 4(b)) over a week (Aug 28–Sep 7). At the same time, the contribution of remaining hot fluids to air warming inside the adit is comparable to one of radon and water vapour (figure 4(a)). Otherwise air temperature variations inside the adit would have been more evident. It should be noted that the LIDAR aerosol backscattering signal (figure 3(a)) appeared to be 1–2 orders of magnitude more sensitive to crustal deformations than radon concentration/relative humidity and temperature variations inside the adit.

Figure 4(c) shows a fragment of the local Earth's crust deformation signal measured by the laser strainmeter (thin line) and the corresponding LIDAR aerosol backscattering data (thick line) in August–October, 2019. For the given observation interval (∼60 d) both processes demonstrate a decrease of measured parameters. Thus, compression of the Earth's crust by ∼20 μm is accompanied by an almost two-fold decrease in aerosol emission driven by migration of fluids of deep origin. The revealed decrease in aerosol concentration may be explained in terms of tightening of cracks and pores in rock due to the local crustal layers' compression. Moreover, the seasonal Earth's crust compression [21] measured by the laser strainmeter (figure 4(c), thin line) has to be modulated by diurnal, semi-diurnal, weekly-biweekly variations of ambient temperature, atmospheric pressure and tidal waves [23]. Taking into account the expected variations in the aerosol concentration one may expect their manifestation in the LIDAR data as well.

The Fourier analysis of the strainmeter and LIDAR data time series (see figure 5) has revealed a good correlation in the modulation of these signals by tidal waves P₁K₁ and S₁ [23] (figures 5(a) and (b), respectively). Note that sharp aerosol emission bursts (see figure 4(c)) yield additional noise in the spectrum (figure 5(b)) compared to the deformation spectrum (figure 5(a)). Due to additional spectral noise the effect of tidal modes M₁, N₁ and O₁ in the LIDAR data is not as evident as in the strainmeter data (figure 5(a)). The only solution here is to continue the LIDAR monitoring so the missing modes may reveal themselves in the spectral domain eventually [21, 22].

Zoom In Zoom Out Reset image size

While the Fourier spectra of both atmospheric temperature variations and aerosol scattering data have demonstrated clear 12 and 24 h harmonics (figure 5(c)), such harmonics are missing in the spectrum of air temperature variations inside the adit (figure 5(d)). This discrepancy strongly suggests that forced ventilation of the underground research facility creates temperature oscillations near the strainmeter yet produces zero effect on the air temperature in our deep-seated geophysical laboratory.

The results discussed above demonstrate that the LIDAR data analysis reveals correlations between the aerosol backscattering signal and other signals obtained by well established conventional techniques, i.e. precision thermometry and Earth's crust deformation measurements by laser strainmeters. Thus, the aerosol LIDAR may be considered as a new perspective sensor for fracture-emitted aerosol monitoring and hence as a new instrument for studying the geophysics of deep magmatic sources.

For the first time we detected the Earth's crust deformation by fracture-emitted aerosol emanations, which were quantified by a highly sensitive LIDAR system. The specially designed compact and eye-safe LIDAR instrument (910 nm, 3 ns, <1 μJ cm⁻²) was developed to detect tiny changes in fracture-emitted aerosol concentration in an underground geophysical laboratory located over the deep magmatic source of the Elbrus volcano in Northern Caucasus.

Simultaneous measurements of radon gas emanations, air humidity and air temperature in an isolated adit of the geophysical laboratory, as well as the Earth's crust deformations (quantified by a laser strainmeter located at the Baksan Neutrino Observatory), have shown a steady decrease by 2%–3% of its corresponding values. The obtained dependences may indicate the magmatic nature of aerosols carried by fluid migration of deep origin. Interestingly, but according to the LIDAR measurements, the aerosol emanations decreased two-fold during the observation period. Such a difference in aerosol emanation, the Earth's crust deformation, radon emission and air humidity/temperature may be explained in terms of the tightening of cracks and pores in rock due to the local Earth's crust compression. It was shown that diurnal and semi-diurnal periods of the crustal layers' compression due to tidal waves P1K1 and S1 are manifested in modulation of aerosol emission.

We conclude that considering the actual combination of experimental data a new aerosol indicator can be introduced in geodynamic and seismotectonic studies. It is at least twice as sensitive to crustal deformations compared to observed variations of radon and water vapor emanations due to the cumulative effect of all aerosol emission events along the LIDAR sensing track.

The authors are grateful to academician E.I. Gordeev and to academician A.D. Gvishiani for fruitful discussions.

Russian Science Foundation (agreement №19-19-00712).

The authors declare no conflicts of interest.