Edited by: Raquel Nieto, University of Vigo, Spain
Reviewed by: Zhao Li, The Hong Kong Polytechnic University, Hong Kong; Penka Vlaykova Maglova Stoeva, Space Research and Technology Institute (BAS), Bulgaria
†Deceased
This article was submitted to Atmospheric Science, a section of the journal Frontiers in Earth Science
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Mountains that contain subterranean voids can inhale fresh and clean air, and their breath is a fascinating natural phenomenon that speleologists know very well. Air flow through the entrances of underground systems is also an interesting geophysical problem. Basically, it is caused by temperature and pressure gradients between the internal and external atmospheres, but the dynamic interplay between these two driving forces is still not well understood. Our contribution dissects the physics of underground winds. Wind velocity, internal and external temperature and pressure have been measured synchronously at two entrances of a vast (∼64 km) underground system beneath the Mount Corchia, Apuane Alps, Italy. The data shows that, within time scales of minutes to days, pressure fluctuations of the external atmosphere primarily force air to flow underground, whereas temperature gradients play only a minor role. We modeled the cave as a system that takes the external atmospheric pressure as the input signal and outputs wind from its entrances. This wind, in turn, contains information about the system’s response, and hence on the structure of the subterranean voids. This information can be extracted using standard signal processing techniques and by using deconvolution methods we identify the same infrasound resonances in signals sampled at both entrances. These are the characteristic frequencies of the cave, and by using the Helmholtz resonance formalism it can be estimated that the explored volume of this important underground system is less than half of its probable real extension.
Caves are natural underground cavities and in most cases they are not isolated physical systems. The vast majority of caves on Earth are in fact connected to the external world by one or more entrances and by water flows, and in general they exchange energy that ultimately can sustain life in the form of a specialized biological community. Air flow in caves is frequently observed and is caused by temperature and pressure gradients between the internal and the external atmosphere. Periodic wind reversal within cave entrances has been observed and measured (
Deep (∼−1 km) and long (>>1 km) caves with more than one entrance show variable, often intense, air circulation, and these underground winds guide the explorations of speleologists through connected, yet unknown branches of subterranean voids. The physics of such winds is much more complex.
The basic physical principles of underground air flow have started to be unraveled (
No comprehensive models have been developed that can explain the physical details of underground winds in vast underground systems. Direct observations on wind speed flowing through the entrances of different cave systems might help in dissecting the driving forces and important parameters. Unfortunately, recording a single sample of wind velocity at only one entrance of different caves is not of much use. As shown in
The Antro del Corchia underground system develops within the Corchia Mountain in the Apuane Alps, Tuscany, Italy. Overall, the known part of the system is ∼1.2 km deep and ∼64 km long, and 16 entrances are presently known that provide access to an intricate network of huge galleries, chambers, shafts, and narrow passages that roughly develop along three main levels at different depths (see
Wind speed was measured using two biaxial Windsonic (Gill Instruments) sonic anemometers (accuracy 2% at 12 m/s, resolution 0.01 m/s). The anemometers were placed inside the cave at ∼1–5 m from both the Eolo and the Serpente entrances, and in the middle point of the smallest conduit section in order to increase accuracy and to reduce noise. Data were sampled at 0.2 Hz and stored electronically using a GP1 data-logger (Delta-T Devices).
External temperature and pressure were measured at 1 min intervals approximately halfway between the two cave entrances using a digital P-Log 125B barometer/thermometer equipped with an PT100 temperature sensor (resolution of 0.01°C) and a piezoresistive pressure sensor with a precision of 0.05 hPa in the range of 800–1200 hPa. The internal temperature was measured along the path connecting the two entrances. During the measurement campaign the internal temperature was 9.7 ± 0.2°C and for the purpose of the present work was considered constant (fluctuations in the order of 0.2° are not expected to contribute significantly to global underground air flow, see below and
The measurement campaign took place between July 7 and 14, 2008, and during this period we kindly asked speleologists/tourists not to visit the cave. Indeed, no transient perturbations of measurements due to the presence of people in proximity to the instruments could be observed in the signals (
All analyses were carried out using the software Mathematica (Wolfram Research Inc., v. 11.1.1.0), and its signal processing package, running on an iMac with 3.4 GHz Intel Core i7 processor.
Wind speed and external temperature and pressure measured synchronously at the Eolo and Serpente entrances of the Antro del Corchia cave. Top, raw wind speed data recorded (0.2 Hz sampling rate) at the two entrances. Middle, external temperature. Bottom, external pressure.
The velocity of air flowing through the two entrances showed similar kinetics, but the magnitude of wind speed was generally higher at the Eolo entrance. On three occasions, at ∼65.000, ∼570.000, and ∼590.000 s, the direction of air flowing through this entrance reversed and, at the same time, local minima could be recorded at the Serpente entrance. The Eolo entrance is directly connected to large underground galleries (i.e., with a cross-sectional area of ∼10 m2), whereas the Serpente entrance is connected to them through a ∼40 m long narrow tube. The lower magnitude of wind velocity and the absence of wind reversal in the Serpente signal might have been caused by friction forces dumping underground wind flows (
No evident common patterns could be observed between wind speed data and in the external temperature or pressure signals. As expected, the external temperature showed daily periodic variations, and on two occasions – i.e., at ∼89000 s and ∼183000 s – a peak in the external temperature corresponded to a peak in wind speed flowing through the Eolo entrance.
The external temperature fluctuations give rise to density differences that increase linearly with ΔT = Text − Tin, where
where ρ is the air density and
Because of its dependence on external temperatures, convective air flow is expected to show seasonal variations. These are actually well known to speleologists that exploit this fact to guide their explorations. During the summer season, as in the present case, when the external temperature is on average warmer than that of the internal atmosphere, external air is expected to flow into the cave from its entrances placed at the highest heights and to exit from the lowest entrances. Obviously, daily variations due to atmospheric perturbation and drop of external temperatures are also possible. The Serpente entrance is the lowest known entrance of the Corchia underground system, so we would expect to observe different wind directions at the Serpente and Eolo entrances due to convection. This was not the case in our data (
In principle, to test whether convection could have forced air to flow through the two entrances of the Antro del Corchia cave, one might exploit the proportional relationship between
Since the external temperatures were measured at 60 s intervals whereas wind velocities were measured at 5 s intervals, we first resampled air speed data by averaging the 12 samples within each 1 min time interval. Plots of raw wind speed data vs.
Correlation between wind speed data and temperature gradient.
We then reasoned that if the two variables were somehow correlated, then both wind speed time series should contain information at least of the daily oscillation of the external temperature, information that could be extracted by analyzing the frequency components of the signals. To this purpose we resorted to a spectral analysis of the de-trended signals.
Spectral analysis of de-trended temperature gradient and wind speed signals. Signals were de-trended using a moving average filter (see the main text for details). The figure shows the periodograms computed for
Pressure gradients can play a key role in forcing underground air to flow through the caves’ entrances. The underground atmosphere is connected to the external atmosphere, after all, and thus it must somehow sense and react to variations of the external pressure. The underground atmosphere is confined within closed spaces of various morphologies, like huge chambers, galleries, wells, narrow conduits, and so forth, that influence the movement of air masses and thus the dynamics of underground winds. These winds, therefore, must contain information on the caves’ shape, as well as the sound which carries information on the structure of the musical instrument that has produced it. The external pressure is the input signal that forces the cave system to output winds. Thus, the caves’ response could be identified through the deconvolution of output wind speed and input pressure signals. Since the two entrances where wind speed was measured are connected to the same system we expected to find the same system response after deconvolution analyses.
We used the discrete Fourier transform algorithm to compute the deconvolution in the frequency domain.
Spectral analysis of the system’s response. The response of the cave system has been estimated by deconvolution of wind speed signals recorded at the Eolo and Serpente entrances (considered as the system outputs) and the external pressure signals (considered as the system input). The analyses reveal common peaks. In the Serpente case, the two peaks at frequencies <1 mHz are only vaguely visible on this scale but are well above the baseline (arrows). Note that in both spectra the
No comprehensive physical models of air circulation in vast underground systems have been developed so far, and this is in part due to the high complexity of the problem. Vast caves are not straight tubes but show intricate morphologies and are in general connected to the external side of the mountains through different entrances placed at different heights. Thus, underground air flow is likely to be characterized by turbulent chaotic motions, and to be influenced by complex non-linear interplays between temperature and pressure gradients among different parts of the system. In addition, measurements of air flow taken at the entrances of different underground systems are noisy and at first sight do not appear to contain regular and common patterns. We show here that synchronous and careful measurements of internal and external temperature, pressure and wind speed at two connected entrances of a cave can help in dissecting the main forces that drive underground air flow.
Barometric forces dominate convective forces and mainly affect wind signals, at least at the considered time scales of 1 week. This might not be true at longer scales. For example, the seasonal variations of air circulation in caves, that are well known to speleologists, are likely to depend on the temperature gradients between the internal and external atmosphere (
where
Overall, our analyses show that caves are the biggest natural wind instruments on Earth.
Giovani Badino passed away on August 8, 2017 while we were preparing the draft of this paper. He collected all the data shown here and many more during lonely intense measuring campaigns. He was an expert speleologist, a passionate explorer, an excellent scientist with solid, and vast roots in humanities. He was a very good friend.
The data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher. A substantial part of the data is, nonetheless, provided as
GB contributed to the conception and design of the study, recording the measurements, organizing the database, and writing the first draft of the manuscript. RC performed the analyses and wrote the manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at: