Understanding air exchange dynamics between underground
cavities (e.g., caves, mines, boreholes, etc.) and the atmosphere is
significant for the exploration of gas transport across the
Earth–atmosphere interface. Here, we investigated the role of atmospheric
conditions in controlling air transport inside boreholes using in situ field
measurements. Three geometries were explored: (1) a narrow and deep
Understanding air transport between the Earth's subsurface and the atmosphere is a key element in the study of surface and near-surface biological, chemical, and physical processes. Air transport between the Earth and the atmosphere is commonly driven by diffusive and advective mechanisms (Choi and Smith, 2005; Ganot et al., 2014; Hillel, 1998; Kawamoto et al., 2006; Kuang et al., 2013; Noronha et al., 2017; Sánchez-Cañete et al., 2013) and has mainly been studied within soils (e.g., Allaire et al., 2009; Bayer et al., 2017; Choi and Smith, 2005; Zeng et al., 2017). However, as different types of natural or artificial boreholes and shafts also exist (e.g., Berthold and Börner, 2008; Kang et al., 2014, 2015), understanding the mechanisms that govern air and greenhouse gas (GHG) transport in such conduits is also important (Berthold and Börner, 2008; Perrier et al., 2005).
Boreholes and shafts are abundant discontinuities crossing the Earth's surface that commonly function as conduits between the ambient atmosphere and underground cavities (e.g., James et al., 2015; Moore et al., 2011; Pla et al., 2016; Steinitz and Piatibratova, 2010). These underground cavities can be more than 1 order of magnitude larger than the connecting conduit. Advective air transport in such boreholes or shafts can be governed mainly by (1) barometric pumping (BP), which is the inflow and outflow motion of subsurface air due to pressure gradients governed by changes in barometric pressure (Ellerd et al., 1999; Neeper, 2002; Neeper and Stauffer, 2012; Perina, 2014; Perrier and Le Mouël, 2016; Rossabi and Falta, 2002; Thorstenson et al., 1998), and (2) density-induced convection (DIC), which develops when there are unstable density gradients resulting from temperature and air composition differences between the atmospheric air and the borehole or underground cavity (Ganot et al., 2012; Nachshon et al., 2008; Perrier et al., 2002; Weisbrod et al., 2009; Weisbrod and Dragila, 2006).
BP will initiate airflow when pore pressure in the surrounding rock–soil differs from the pressure within the borehole or the underground cavity, which is considered as equal to atmospheric pressure (Kuang et al., 2013; Neeper, 2003; You et al., 2010). BP is dependent on the surrounding rock–soil depth and permeability and on the barometric pressure changes (Massmann et al., 2000). A water table at the lower boundary is considered to be impermeable to BP (You et al., 2010).
The onset of DIC is typically primarily determined by temperature
differences within the borehole or shaft and between them and the atmosphere
above. Temperature differences can differ between locations and depend on
several parameters, such as the surrounding rock–soil thermal properties,
the geometry of the borehole, or the atmosphere temperature cycles
(e.g., Eppelbaum and Kutasov,
2011; Klepikova et al., 2011; You and Zhan, 2012). Although air density
depends mainly on temperature, it also depends on the air humidity and to
a lesser degree on the air's gas composition (Kowalski and
Sánchez-Cañete, 2010). The integration of these three effects
(temperature, relative humidity, and air composition) into a single
parameter called virtual temperature (
Each of the above advective mechanisms was studied individually. However, to the best of our knowledge, no comparative research has been conducted to determine which mechanisms dominate different borehole and shaft geometries, e.g., different borehole diameters. Therefore, the relative contribution of each mechanism to the overall air transport within boreholes or shafts under different environmental conditions remains loosely constrained.
Here, we investigated the role of atmospheric conditions in the air transport mechanisms inside three borehole geometries: (1) a narrow-diameter shaft (0.1 m) with a PVC pipe opening into a large underground cavity (defined hereafter as “shaft” geometry); (2) the same shaft after the pipe was lowered and the link between the shaft and the underground cavity was blocked (“borehole” geometry); and (3) a borehole with a larger diameter of 3.4 m (“large-diameter borehole” geometry). Specifically, we aimed to assess the air inflow and outflow events by quantifying the oscillation of physical parameters, mainly temperature and relative humidity (RH) along the boreholes, and relate the flow events to the atmospheric forcing conditions.
We examined two sites: (1) a narrow-diameter (0.1 m) 27 m deep borehole that
reaches a large underground cavity located above the local water table of
the southern part of the Israeli coastal aquifer and (2) a large-diameter
(3.4 m) 59 m deep borehole that reaches the local aquifer near the Nabatean
archaeological site of Avdat in the Negev highlands of southern Israel. The
distance between these sites is
The first borehole was drilled into a sequence of alternating layers of
sand, sandstone, and silt comprising the unsaturated zone of the Israeli
coastal aquifer (Goren et
al., 2014; Schwarz et al., 2016). A PVC pipe (case) was inserted into the
narrow-diameter borehole to prevent potential soil collapse. The pipe
reached the top of the underground cavity, which is at least 2 orders of
magnitudes larger in volume than the pipe and is located well above the
local groundwater level of
The second borehole site (large-diameter borehole) is an archaeological site that was excavated
into Eocene chalk formations with an upper part of loess soil (Nativ
et al., 2003; Shentsis et al., 1999). The water table in this site is at
a depth of
Schematic illustration of the three studied geometries
Sensors installed at the first site included four thermocouples (type T, Omega Engineering, UK) at depths of 0, 6, 18, and 24 m and two RH–temperature sensors (Hygroclip 2, Rotronic AG, Switzerland) at a depth of 12 m and at the lower part of the borehole at its connection point to the cavity (27 m of depth); see Fig. 1b for an example of sensor locations. Outside the borehole, a meteorological station was installed at 2 m aboveground, including the following sensors: (1) wind velocity and direction (WindSonic, Gill Instruments, UK); (2) barometric pressure (CS106, Vaisala, Finland); and (3) RH–temperature (same type as within the borehole). Data from all sensors were measured at 5 s intervals and averaged and logged at 10 min intervals (CR1000, Campbell Scientific, UT, USA). In addition, a televiewer was lowered into the pipe to verify that the pipe was intact and was either connected to or disconnected from the underground cavity in the shaft or borehole, respectively. Measurements at the second site included a similar RH–temperature sensor at the depth of 10 m and a temperature sensor at 59 m.
Absolute humidity (AH) was used as a tracer for the air transport within the
three geometries and was calculated from the measured temperature and RH
data using Eq. (1) (Hall et al., 2016):
A 1-week time series of results from the shaft is shown in Fig. 2a. Atmospheric
temperature and RH presented typical daily cycles, as expected. During
daytime, atmosphere air temperatures were higher with lower RH values
(25–3
Barometric pressure typically varied with two diurnal cycles; the average
barometric pressure was
Time series results from four representative days for the
shaft
Atmospheric AH was stable during the measurement period and values ranged
between 10 and 15 g m
To quantitatively define each inflow or outflow event, a classification
algorithm was built and solved for the 42-day data using MATLAB
Classification of inflow and outflow transition events at 12 m of depth
for the shaft
Time series results from a single transition event. Panels
To identify the physical parameters that control the transition between air
inflow and outflow events, we focused on and analyzed in detail single events,
one of them given as an example in Fig. 4. In a typical event, with both
inflow and outflow, three stages were observed: (1) transition of
d
Time lags between changes in d
Figure 6 examines the correlation between the direction of air transport
(inflow–outflow) and the atmospheric forcing, i.e., changes in barometric
pressure and thermal stability. The general distribution of
d
Our results indicate that changes in atmosphere barometric pressure determine the advective airflow direction. This is presumably due to the difference between the barometric pressure and the cavity pressure caused by the vertical propagation of the barometric pressure (Neeper, 2003; Neeper and Stauffer, 2012; Perina, 2014; You et al., 2011). In the case of positive barometric pressure changes (i.e., increase in barometric pressure over time), the barometric pressure will be greater than the cavity pressure and thus inflow of air will develop. In contrast, outflow of air will start when negative pressure changes occur (i.e., decrease in barometric pressure over time).
Atmospheric conditions during the borehole measurements presented daily temperature
and RH cycles (Fig. 2b nos. 1 and 2, black lines). During daytime, air
temperatures and RH values were 20–25
In the inflow events the AH values in the middle borehole sensor (12 m) were similar to the upper atmospheric values (Fig. 2b no. 3, purple and black lines), whereas in the outflow events they equalled those in the lower part of the borehole (Fig. 2b no. 3, purple and black dash lines). Nonetheless, the airflow effect was observed only at the 12 m depth and not at 27 m (Fig. 2b no. 3, green line). Therefore, we conclude that inflow events reached the depth of 12 m but did not reach depths of 27 m.
While all shaft and borehole parameters at site 1, other than the connection to the lower
cavity, were identical and the atmospheric conditions in the two
measurement periods were similar, there were still clear differences between
the two geometries. The borehole temperature readings at 12 m exhibited only half of
the standard deviation compared to the same depth in the shaft (
Histograms of changes in atmospheric barometric pressure (d
The shaft–borehole differences at the 27 m sensor can be explained using a simple two-reservoir model. In the case of the shaft, we can define both the atmosphere and the cavity as two infinite air reservoirs connected via a finite-volume shaft. Therefore, air transport between the two reservoirs is unlimited and controlled only by the boundary conditions (i.e., barometric pressure). In this case, all sensors throughout the shaft will show similar AH decrease and increase, as evident from the similar changes in the purple and green lines in Fig. 2a no. 3. On the other hand, in the borehole, there is only one upper infinite reservoir (i.e., the atmosphere) and each inflow air transport is limited by the soil resistivity at the lower boundary (i.e., the soil capability to enable penetration of air inflow events – the soil permeability and porosity). Here, the effect of AH decreases with depth and indeed the deepest sensor of 27 m showed no change in AH compared to the changes in AH at the 12 m depth (Fig. 2b no. 3).
Time series results from the large-diameter borehole for
1 week. Gray columns represent periods of thermal stability inside the
large-diameter borehole. Values in line 5 represent
Results from the large-diameter borehole are presented in Fig. 7. In contrast to the
shaft–borehole, AH changes inside this large-diameter borehole (measured at 10 m of depth) were correlated with the
We posit that the main parameter controlling which transport mechanisms
govern advective air movement is the borehole or shaft
diameter. A small borehole diameter will
decrease the magnitude of the DIC. This is because DIC magnitude in a
cylinder geometry is positively proportional by the fourth power of the
cylinder radius (e.g.,
Berthold, 2010; Berthold and Börner, 2008; Berthold and Resagk, 2012;
Rayleigh, 1916). Therefore, in our case of a small-diameter borehole of 0.1 m, DIC
had a minor influence on air transport inside the shaft–borehole. The following equations
provide the theoretical basis for our conclusion that borehole diameter (
For BP, under the assumption of unidimensional cylindrical flow, the volume
flow rate per unit length (
Conceptual model for airflow inside boreholes and shafts.
Panel
For the case of DIC, the thermal instability number (
The use of Eqs. (5) and (10) for comparison purposes cannot be addressed
without considering the differences in
It should be emphasized that the threshold value of
A conceptual model was developed to present the advective transport mechanisms of the three geometries (Fig. 8). The differences between the borehole and the shaft are illustrated in Fig. 8a and the differences between them and the large-diameter borehole in Fig. 8b. The borehole diameter will define which advective transport mechanism is more significant: at a small diameter of 0.1 m, BP controls gas transport (Fig. 8a), whereas at the large-diameter borehole of 3.4 m, DIC is the dominant mechanism (Fig. 8b).
The mechanisms controlling the subsurface–atmosphere air exchange have several important implications. These include, for example, volatile organic compound (VOC) transport from the subsurface to the atmosphere in contaminated sites (Boothroyd et al., 2016), natural aeration (oxygen supply) of underground quarries or tunnels and the need for artificial, enhanced air exchange facilities in such environments, and changes in RH values in karst systems. For example, RH changes in a mine underground atmosphere have a great influence on the rock physico-mechanical parameters and stability (Auvray et al., 2008). Commonly used mine shafts can induce rapid RH changes at the shaft–cavity interface as presented above, which can then lead to rock stability problems. Shafts can also be used for the fast removal of water vapor from deep soil layers in order to lower its hydraulic conductivity and subsequently cease the downward transport of contaminants.
One of the important implications is the potential role of shafts and boreholes as conduits for air exchange to overall GHG emissions and related mechanisms such as carbon capture and storage (CCS) processes. Two basic assumptions are considered here: first, the BP air transport rate is up to a few orders of magnitude greater than diffusion (You et al., 2011), and second, these conduits can act as “pipes” to the Earth's subsurface, connecting elevated GHG sources to the atmosphere (e.g., Serrano-Ortiz et al., 2010).
One example of a significant GHG type emitted from boreholes is methane, whose
emissions were quantified for 19 narrow boreholes in Pennsylvania
(Kang et al., 2014, 2015). After upscaling their results to the state level, it was proposed
that these borehole emissions represent 4 %–7 % of the total methane emissions in
Pennsylvania. The research focused mainly on the production function of
methane and not on the physical transport mechanism. Implementing our
conclusion that BP was the main air transport mechanism can indicate that
the methane emissions presented by Kang et al. (2014, 2015) likely occurred
mainly during periods of d
Three borehole geometries were compared to explore air transport mechanisms under naturally variable atmospheric conditions. The first case was a 27 m vertical shaft with a 0.1 m diameter that connected a large underground cavity to the atmosphere; the second case was the same borehole but with a connection to the underground cavity that was blocked and the pipe ended in the unsaturated soil matrix. The third was a large-diameter borehole 3.4 m in diameter and 59 m deep. In the first two, the shaft and borehole, the air inflow and outflow at 12 m were found to be correlated with the changes in barometric pressure (BP). However, in the large-diameter borehole, the air transport at a similar depth (10 m) was correlated with density–instability (DIC) rather than barometric pressure.
Use of AH changes during the winter and spring seasons was shown as a practical tool to identify the source of air parcels within the three geometries, namely atmospheric vs. lower borehole–cavity, and thus to determine the direction and effect of the air transport. Water vapor concentrations in the atmosphere vary throughout the day, while they are almost constant in underground cavities and can therefore be used as a natural tracer for air source and flow directions without injecting additional gases.
A conceptual model is presented to describe the induced airflow in all three geometries. In the shaft, the atmospheric air entered through the shaft to the cavity and vice versa. In other words, the shaft connects between two large air sources, and inflow and outflow via the shaft is determined according to the barometric pressure changes. In the borehole, the atmospheric air entrance was limited by the soil resistivity at the lower boundary. Thus, the inflow of atmospheric air was observed only at 12 m of depth and not at the deeper 27 m sensor. BP was found to control air advective transport in both geometries. On the other hand, in the third geometry of a large-diameter borehole, thermal instability initiated DIC advection, while BP did not play a significant role. This caused the circulation of atmospheric air into the borehole to a depth of 10 m whenever the thermal instability occurred. This mechanistic explanation was validated using the winter and spring season dataset. Although we show that theoretically the transport mechanism observed for winter and spring should hold with reduced significance for summer and autumn, further data are needed to verify the theoretical calculation.
In summary, our observations improve the understanding of the governing mechanisms controlling air movement in boreholes and shafts as a function of their geometries and diameters as well as the ambient atmospheric conditions. In addition, our observations assist in better calculating GHG fluxes from these domains and estimating the time periods when these fluxes are enhanced.
The dataset used in the analyses can be found in the Supplement to this article.
The supplement related to this article is available online at:
EL, NGL, AM, and NW performed the data analysis. EL and NW wrote the first draft of the paper and all authors contributed to the final version.
The authors declare that they have no conflict of interest.
This work was funded by the Bi-National Science Foundation (BSF), contract number (2014220), the Israeli Ministry of Agriculture, contract 857-0686-13, and the Israeli Science Foundation (ISF), contracts 678/11 and 1471/18. We also acknowledge the Sam Zuckerberg scholarship provided to Elad Levintal and the fruitful comments provided by Andrew Kowalski and the two anonymous reviewers who helped to improve this paper. The field observations were conducted with the Geological Survey of Israel team: Hallel Lutzky, Uri Malik, Haim Chemo, Ziv Mor, and Haggai Eyal, with Raz Amir from the Ben-Gurion University of the Negev. Edited by: Axel Kleidon Reviewed by: Andrew Kowalski and two anonymous referees