Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 If we are to understand global warming, and in particular global water-cycling, then it is vital to explore the links between atmospheric conditions, earth processes and major global cycles. One arena that has been heretofore ignored is the effect on global dynamics of earth fractures that are open to the atmosphere. Historically, these fractures have been studied merely as participants in aquifer recharge or aquifer contamination during periods of infiltration. In general, they are considered inactive when there is no precipitation. This paper puts forward in-situ continuous field measurements demonstrating that during no-flow periods, fractures breathe via convection on a daily basis, enhancing atmospheric exchange by several orders of magnitude compared to the non-fractured crust. We quantify the timing, persistence and characteristics of this mechanism. The convective exchange mechanism is pervasive, occurring daily with peak flux exchange at night and in winter, the reverse of most other surface processes.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Although most will agree that terrestrial fluxes of CO2 and other gases play an important role in regulating global climate change, quantifying those fluxes remains a challenge [Ryan and Law, 2005; Kowalski et al., 2008; Dabberdt et al., 1993]. Mechanisms that control gas production within the soil and the transfer of gases across the soil-atmosphere interface are complex and interdependent. For example, increased soil moisture and temperature enhance microbial activity and gas production/consumption rates, but also reduce transport by blocking soil pores with water [Riveros-Iregui et al., 2007]. Historically, the process of gas exchange between soil and atmosphere has been assumed to occur by diffusion [Campbell and Norman, 1998], but enhanced to shallow depths by advective transport driven by wind and atmospheric pressure fluctuations [Massman et al., 1997; Weeks, 2001; Albert, 2002]. Atmospheric disturbances can also drive deep venting of fractures and bore holes in mountainous terrain [Weeks, 2001], caves [Fernandez et al., 1986] and tunnels [Thorstenson and Pollock, 1989]. While, these advective events are infrequent and depend upon local weather variability, this paper presents a daily recurring mechanism that vents cracks and fractures at rates that significantly exceed diffusive venting. Thus, in regions of our planet where vadose zone fractures, karst terrain, or cracking soil and desiccating playas prevail, assuming that the primary mechanism is that of diffusive gas exchange may lead to an underestimation of the flux of greenhouse gases to the atmosphere.
 For more than a half-century it has been readily accepted that soils dry noticeably faster when cracks are present has been common knowledge [e.g., Ritchie and Adams, 1974]. Yet, the mechanism that controls enhanced drying was never elucidated. Moreover, it has been observed that salt crust accumulation within natural fractures is much greater than can be explained solely by diffusive venting of evaporated water [Weisbrod et al., 2000]. Weisbrod et al.  and Weisbrod and Dragila  proposed that thermally driven convection (rather than diffusive venting) might be the primary mechanism for venting of fractures. This new hypothesis is able to successfully account for the amount of enhanced soil drying and salt accumulation observed. Nachshon et al.  and Kamai et al.  quantified this mechanism under controlled laboratory conditions for different thermal gradients and showed that fractures with apertures greater than 1 cm convect freely, while thinner apertures are affected by thermal and viscous dissipation. If, in fact, this mechanism exists under natural field conditions, its importance to atmospheric gas exchange could be quite significant, because convective venting would affect not only gas flux but also speciation.
 Under what conditions do soil cracks and surface fractures vent convectively? Convective venting is facilitated by the upper vadose zone thermal gradient. The sun's heating of the soil surface propagates slowly downward due to the low thermal diffusivity of the vadose zone so that around midnight it is warmer below the surface (∼30 cm) than at the surface [Hillel, 1998]. At night, warm air inside cracks is less dense than atmospheric air. Thus, as the atmosphere cools at dusk, an unstable air-density gradient develops causing an overturning of air within the crack, much like the overturning of lake water in winter. Venting of warm, moist air from the crack and entrainment of cool, relatively dry atmospheric air continues unabated until dawn. Thus, whereas diffusive vapor flux and gas exchange from the soil surface to the atmosphere peaks during the day, the component from the cracks is delivered to the atmosphere by convective flux at night.
 The objective of the work presented here was to determine whether or not the natural vadose zone thermal profile is sufficient for development of convection cells within a natural surface-exposed fracture (SEF). Presented here is continuous, high resolution, temperature and relative humidity data taken within a SEF over a two-year period.
2. Materials and Methods
 The field site (Figure 1) consisted of a natural fracture that was open to the atmosphere, with the following dimensions: variable aperture of ∼1–5 cm, fracture trace ∼2 m long, and variable depth >1 m. Located in the northern part of the Negev Desert, Israel, the geologic unit is the Avdat Group, composed of alternating beds of chalk and some limestone (Total porosity: 40–45%. Permeability: 1–2 mDarcy) [Weisbrod et al., 1999]. Mean annual precipitation is 200 mm with an average of 41 rainy days spanning October to April. Mean annual temperature is ∼19°C, ranging between 44.8 and −1.8°C. Annual pan evaporation is ∼2.3 m.
 Instruments were installed in an underground laboratory (1.2 wide × 2 m long × 2 m deep) that intersected the fracture plane (Figures 1a and 1b). The ceiling and walls were sealed with epoxy cement (Duralite®, Bolidet, Netherlands) to prevent exchange of water vapor between the laboratory and the fracture. In the vicinity of the underground laboratory the fracture was open to a depth of about 1.25 m; below that depth the fracture split into two thin, parallel, mostly clogged, fractures with apertures of less than 0.002 m. An instrument suite of thermocouples (TCs) and relative humidity (RH) probes (TCs: Copper-Constant thermocouple wire, Omega Engineering Limited, Manchester, UK; RH: Hygroclip SC04, Rotronic, Zurich, Switzerland; accuracy 1.5% at 23°C) was installed and connected to a multiplexer and data logger (CR10X, Campbell Scientific, Logan, Utah) (Figure 1). Data were collected every 10 minutes. Instruments installed in 2004 were: A 1-D string of TCs to measure fracture air temperature; an identical 1-D string of TCs to measure rock temperature; and a 1-D string of RH probes measured fracture air humidity. TC probes were positioned at 10-cm intervals and RH probes at 40-cm intervals, vertically from the soil surface to a depth of 120 cm. Later (in 2006), one 2-D 60 × 60 cm array of 25 TCs (15-cm spacing) was installed forming a square grid along the plane of the fracture (Figure 1c). Data from the 1-D arrays were collected during 2004–2005, and 2-D arrays from 2006 to present.
Figure 2 shows an example of a thermal map of fracture air as it develops convection cells at night. To our knowledge this is the first time that the formation of convection cells associated with thermally driven convective venting of air in a natural fracture has been directly measured and visualized in situ. The thermal map was generated from data obtained from the 25 TCs of the 2-D array shown in Figure 1c. Figure 2 clearly shows that during the day (e.g., 1700 h) air inside the fracture is, as expected, warm on top and cool below (i.e., lighter over denser air, stable conditions). But as dusk approaches and the atmospheric air cools (e.g., 1900 h), unstable conditions develop, two convection cells form and a finger of cool atmospheric air invades the fracture (also see Animation S1 of thermal dynamics during a typical 24-h period). Cool air entering the fracture is drier, resulting in a net removal of water vapor from the fracture and the surrounding matrix. Figure 3a shows the RH measured at a depth of 10 cm inside the fracture over a typical 48-h period. Note that as dusk approaches and venting begins, the RH of the fracture air begins to decrease (100% to 65–80%) and stays relatively low until venting ceases at dawn. A very good correlation was observed between the time at which atmospheric air density is higher than that of the fracture air, and the time at which RH begins to decrease within the fracture. We have monitored the SEF in the Negev Desert continuously for several years now and have confirmed that this diurnal venting pattern repeats itself consistently every 24 h with some seasonal variability.
 Superposed onto the daily venting cycles is also a strong seasonal signal driven by the earth's deeper seasonal thermal profile (Figure 3b). During the summer, the deeper vadose zone is cooler than the average surface temperature, limiting convective venting to the period of nighttime cooling. In winter, however, the deep vadose zone is continually warmer than the average surface temperature (Figure 3b), resulting in an almost continuous upward air density gradient within fractures. Consequently, convective venting in winter is expected to persist throughout the night and well into the day. The 1-D thermal data set was used to calculate the predicted seasonal variability for the length of time each day that thermal convection should occur (Figure 4a). Calculations for Figure 4a take into account the minimum thermal gradient needed to exceed the stability criteria for thermal convection; in all cases the Rayleigh number exceeds the critical value (∼40) needed for convective instability [Nield, 1982]. Figure 4a is effectively a predictive model using only vertical-profile temperature data from 2004–2005 and significant simplifications for aperture geometry. Nevertheless, the prediction from Figure 4a compares very well with the data from the actual convection captured by the 2-D array of thermocouples from 2006–2008 (Figure 4b). Here, the time of active convection was determined from a frame-by-frame analysis of ∼70,000 thermal frames of continuous data similar to those shown in Figure 2 and in Animation S1. On average, convection occurs for up to 18 h per day during the winter and 10 h a day during the summer (Figure 4b). Small differences between calculated (Figure 4a) and observed (Figure 4b) hours of convection are expected due to the impact of wind, internal fracture structure, surface roughness, different locations within the fracture for the 1-D vs. 2-D TC arrays, and different years over which the data were taken for each graph. Considering the potential impact of all of these differences, the predictions from Figure 4A are remarkably accurate.
4. Discussion and Conclusion
 Evidence presented here categorically shows that convective venting of cracks and fractures is a natural and pervasive process that may have a pronounced impact on earth-atmosphere gas exchange in areas where surface cracks and fractures prevail, such as desert playas, cracking soils in agricultural regions or rock fractures. The mechanism presented here is not the only mechanism capable of generating advective venting of fractures; atmospheric turbulence has also been shown to enhance soil and snow air venting. However, while these events are infrequent and depend upon the capriciousness of local weather patterns, the mechanism described in this study, i.e., thermally driven free convection, occurs regularly on a nightly basis (Figures 3c and 3d) and should be considered a primary transfer mechanism in areas where the surface is cracked or fractured.
 How important is this process to global atmospheric fluxes? With respect to water vapor, a known greenhouse gas, convective venting from cracks and fractures makes a marked contribution to atmospheric vapor flux. The vapor flux expected to vent from these fractures [Kamai et al., 2009] assuming a 1-m fracture spacing, is of similar order of magnitude as the daytime soil evaporative flux observed in field studies [Cahill and Parlange, 1998]. Consequently, to ignore this mechanism is to ignore potentially half of the vapor flux to the atmosphere in regions of fractured soil surface. Salt deposition data also indicates a significant contribution. Within a 1-m deep, 2-cm aperture fracture in the arid Negev Desert, we measured an accumulated salt-crust mass corresponding to a vapor venting rate 200 times greater than that which would be predicted by diffusive venting alone [Weisbrod and Dragila, 2006].
 Convective venting may also influence fundamental ecological processes that are involved in soil respiration by increasing the soil depth that contributes to soil respiration and changing the soil moisture conditions at those depths (Figures 3c and 3d). We measured venting activity in a fracture to a depth of 80 cm. At night, gas generated ∼1 m below the surface near a crack will experience diffusive resistance similar to that for gas generated only a few centimeters below the soil surface. Gas species generated by aerobic processes (e.g., volatilization of NH3) are different from those generated anaerobically (e.g., dentrification to N2, N2O, NO). And, because soil commonly exhibits a moisture gradient (i.e., more moisture with increasing depth), venting and drying by venting will change the relative proportion of the components comprising the atmospheric flux. Potential changes to soil respiration in cracked soil are important to consider since soil respiration is the main terrestrial source for the flux of gases (e.g., CO2, methane, etc.) [Ryan and Law, 2005].
 Soil cracks and fractures on the earth's surface are not rare; they are ubiquitous features that can be commonly found in arid, moist and frigid climatic settings. Moreover, karstic systems provide almost 25% of the world's potable groundwater supply [Ford and Williams, 1989]. Preliminary measurements within an ancient borehole (52 m deep, 3.5 m in diameter, Negev Desert, Israel) suggest unstable thermal conditions to a depth of 40 m most of the year (data not shown). The salt crust that accumulates in cracks because of convective venting could pose a significant risk for water quality.
 In an era where even small changes in greenhouse-gas fluxes are thought to have large repercussions on global climatic function, a mechanism with a potentially high contribution begs to be quantified. We submit to the scientific community data showing the existence of a daily convective venting mechanism in situ, with quantification of this process, with the hope that it will be taken into consideration in studies of terrestrial-atmospheric gas exchange.
 Supported by the National Science Foundation, grants 0208384 and 0510825, the Binational Science Foundation (BSF), grant 2002058, and the Israeli Science Foundation, grant 70/06. Comments provided by S. Tyler and an anonymous reviewer helped to improve this manuscript.