UNDERGROUND METEOROLOGY - "WHAT'S THE WEATHER UNDERGROUND?" PODZEMNA METEOROLOGIJA: "KAKŠNO JE VREME V PODZEMLJU?" Giovanni BADINO1-2 Abstract UDC 551.44:551.581 Giovanni Badino: Underground meteorology - "What's the weather underground?" The aim of this work is to provide a synthetic outline of some of the processes of transient nature occurring in caves, focusing on poorly studied general aspects of underground physics and mainly making use of original experimental data. In the first part, the average climatic conditions of a caves, their connection to the external climate, and the general role played by rock, water, air and external morphology are discussed. The variation of the internal temperature with the altitude is a key parameter for the cave physics: the related energetic consequences are briefly discussed. In the second part, transient processes are considered, and a general overview of main meteorological phenomena occurring underground is given. The physics of thermal sedimentation, of underground temperature ranges, of infrasonic oscillations of cave atmospheres and, above all, of water vapour condensation in caves is synthetically described. The experimental study of these processes is extremely difficult, because they are time dependent and have very small amplitude; the first measurements show, however, that their variability from one cave to another, and from point to point inside a cave, is surprisingly high. To provide a more correct interpretation of underground climatic measurements, for their speleogenetic role and importance in cave environment protection, a better understanding of the processes described here is essential. Key words: thermal sedimentation, temperature range, infrasounds, clouds in caves, cave climate, cave meteorology, cave protection. Izvleček UDK 551.44:551.581 Giovanni Badino: Podzemna meteorologija: "Kakšno je vreme v podzemlju?" Namen prispevka je predstaviti nekatere prehodne pojave, ki vplivajo na jamsko klimo. Pri tem izhajam iz osnovnih zakonov fizike in izvirnih eksperimentalnih podatkov. V prvem delu obravnavam povprečne klimatske pogoje v jamah in njihovo povezavo z zunanjo klimo, okoliško kamnine, podzemno vodo ter zunanjo topografijo. Pokažem pomen in vlogo temperaturnega gradienta v jamah. V drugem delu obravnavam prehodne pojave in z njimi povezane ključne meteorološke procese v podzemlju. Predstavim fiziko temperaturne sedi-mentacije, temperaturna območja v podzemlju, infrazvočna nihanja jamske atmosfere in kondenzacijo v jamah. Eksperimentalne raziskave teh procesov so zahtevne, saj gre za majhna nihanja časovno spremenljivih vrednosti; kljub temu se pokaže, da so spremembe vrednosti od jame do jame in tudi znotraj ene same jame presenetljivo velike. Dobra interpretacija meritev, razumevanje nekaterih speleogenetskih procesov in učinkovita zaščita jamskega okolja, bodo mogoči le na osnovi dobrega poznavanja in razumevanja osnovnih klimatskih procesov v jamah. Ključne besede: termalna sedimentacija, območje temperatur, infrazvok, oblaki v jamah, jamska meteorologija, zaščita jam. 1 Dip. Fisica Generale, Universita di Torino, Via Giuria, 1, 10125 Torino, Italy, x@to.infn.it 2 Associazione La Venta, Treviso, Italy Received/Prejeto: 06.09.2010 INTRODUCTION The aim of this work is to provide a synthetic outline of some of the main processes of a transient nature defined as "meteorological" processes; mainly focusing on poorly studied general aspects of underground physics. Under the apparent stability of cave atmospheres, complex processes occur that, as a consequence of their very long time scale, can play an important role on the speleogenesis and on the constructive phenomena. The distinction between climatology and meteorology is outlined. If referring to an outside environment, these two terms are, respectively, defined as the study of average climatic conditions - which can include regular fluctuations, as average temperature ranges and similar - and the study of transient, non-periodic processes in the atmosphere that are, since ancient times, called "me-teora." In this work, the term "climatic" is used to indicate average values of a condition, whereas "meteorological" refers to the fluctuation around it and the transient phenomena. Inside caves, microclimates are very different from the external climate, and knowledge of these microclimates has always been very important for food seasoning and storage, medical treatments, and the use of caves as "time capsules." The importance of cave deposits for paleoclimatic studies is nowadays increasing. Although the climate records essentially refer to hypogean deposits, the direct interest is on epigean paleoclimates. The correlation between the two is, however, not simple. Up to about 20 years ago, microclimatic studies have been: 1) mainly observational, not considering physical models to be confirmed or refused; 2) quite local and relatively close to the surface; 3) lacking resolution and accuracy; 4) lacking long time series. Generally speaking, measurements represented qualitative descriptions of the cave atmosphere, referring very often to too shallow layers. They showed an apparent high stability everywhere. However, note that a measure of the Earth's temperature by a thermometer with a resolution of ±50°C, would everywhere indicate the same value, leaving unexplained why here there is a desert, here a forest and there an ice cap. The old endo-climatic measures behaved in a similar way. Now, thanks to a gigantic technological advance in the field of data acquisition and storage, in the development of physical models and to some advances in measurement techniques and data processing, the situation is quickly changing. Cave physics studies are important: 1) to estimate paleoclimatic data reliability; 2) to understand the local adiabaticity and connect it to the occurrence of complex structures; 3) to understand the speleogenetic role of condensation; 4) to characterise caves as ecological "islands"; 5) to protect caves, especially show caves, against anthropogenic impact; Here we want to outline the main micrometeoro-logical processes and the open problems. The suggestion of new ideas, fields of research and eventually the growth of the interest on cave physics are, therefore, the objectives of this work. THE CLIMATOLOGY OF CAVES Versan le vene le fummifere acque Per li vapor che la terra ha nel ventre, Che dabisso li tira suso in alto. Every stream runs with smoke-infested waves because of vapors hidden in the earth, which from deep chasms rise up to the sky Alighieri, Rime, 43 INTRODUCTION Despite the fact that the underground temperature increases with depth, caves are usually very cold. The reason is the deep vadose zone with percolating meteoric water which equilibrates the rock mass with external temperature. We can give a rough estimation of the involved time-scales. The water thermal capacity is around 4.2 kJ kg-1K-1, the thermal capacity of rock is five times less; the thermal capacity of water per volume unit is, therefore, about double of the rock thermal capac- ity. By assuming infiltration to a karst massif around 1 m3 a-1 m-2, and a rock column of 1 m2 section and 1000 m high, it takes ts=500 years until the thermal capacity of the water equals that of the rock column. The time ts represents the time scale of the cooling process which, although strictly depending on the rock thickness, ranges from hundreds to thousands of years. This is also the time scale of climatic fluctuations, which are able to penetrate in the underground (Badino 2004a). It is therefore rather important to note that the karst massifs are in thermal equilibrium with "historical" infiltrating fluids, mainly water. This explains the reason why karst massifs are quite cold and, in first approximation, unaffected by geothermal flux. As a matter of fact, the occurrence of caves in a mountain increases enormously its permeability and turns the temperature down. The temperature of the infiltrating water changes during the year, but it is easy to observe that a fluid with a flow rate F(t), function of time t, at temperature T(t) passing a thermal capacitor C approaches a temperature To, at which the average thermal energy released from the fluid to the capacitor equals zero. This can be written as: J F(t)[m-T,]dt = 0 Generally speaking, the rock behaves like a low-pass filter for the signal T(t) and it adopts the average temperature of crossing fluids. We will denote this temperature as TR. AVERAGE LOCAL TEMPERATURE In a first approximation we assume that a cave has a temperature, TC equal to the temperature of the surrounding rock, TR, which has the temperature of infiltrating water, TIW, which is almost equal to the local yearly average temperature, TL. The estimation of the local average temperature TL is then of extreme importance for understanding cave physics. TL is a complicated function of longitude, latitude, and altitude. The temperature variation with longitude is affected by the region macromorphology, such as the presence of seas, mountains, forests etc. and cannot be modeled. The latitude dependence is also connected to macromorphol-ogy, but more importantly, there is a regular tendency to lower the temperature going to the Poles; the latitude could therefore be roughly modeled. It is then possible to see (Fig. 1) that in Southern Europe the temperature TL at sea level decreases by 0.7°C per latitude degree. Quite at the opposite, the altitude dependence is very regular, the temperature decreases by about 6.5°C per kilometer of increasing elevation (this is the average "lapse rate" of International Standard Atmosphere), permitting the removal of the effect of altitude and the construction of maps of TL's referenced to sea level (Figs. 1 and 2). Table 1 gives some more precise data in the position of main cities (WorldClimate 2010), corrected to sea level. Table 1: Average yearly temperature of some European and Mediterranean airports (WorldClimate 2010) St. Petersburg 4.5 Edinburgh 8.8 Mancheste 10.9 Frankfurt 10.5 Wien 10.7 Ljubljana 11.8 Lyon 12.1 Zagreb 12.3 Geneve 12.4 Trieste 13.5 Istanbul 14.1 Marseille 14.2 Roma 15.4 Barcelona 15.6 Tunis 17.7 Knowing the local temperature TL0 of a reference point, it is then easy to estimate the TL in the neighborhood. For Europe we propose the formula: Tl = Tlo - 0-7 X ALaf - 6.5 x AAlt Where ALat is the Latitude (degrees) and AAlt the altitude (km) difference between the place and the reference station. Fig. 1: World average yearly temperatures referred at sea level (Pinna 1977; WorldClimate 2010; adapted). Fig. 2: European and mediterranean average yearly temperatures referred at sea level (Pinna 1977; WorldClimate 2010; adapted). We have stated that the cave temperature Tc is almost equal to the rock temperature TR, which is almost equal to the temperature of the infiltrated water TiW, which is almost equal to the local temperature tl- So, the cave temperature Tc is (almost)3 equal to the TL ... Let us see some of the processes capable to influence each "almost." LOCAL AND INFILTRATION TEMPERATURES The first "almost" that we are going to consider is connected to the differences between TiW and TL. The first reason for the temperature difference between infiltrating water and air is the local climate. Generally speaking, it rains during cold days, and the rainfall comes from an altitude, which is higher than the local one. Therefore, the water is colder than the local air, which often cannot warm it up during the fall. The AT depends on many things, but in average it is assumed to be around 2°C (Celico 1986). The precipitations are concentrated in certain periods, depending on the regional climate. Let us compare, for example, climatic (not "meteorological" ...) data from Florence (Italy) and Chihuahua (Mexico) (WorldClimate 2010). The city of Florence (44° N, 50 m asl) is quite far from the Equator, but Chihuahua (29° N, 1400 m asl) is at a relatively high altitude. The relative average local temperatures TL are then comparable: TL Florence: 14.6°C, TL Chihuahua: 17.8°C. The climatic type of the two regions is very different, Florence being Mediterranean, where the rainy season occurs during intermediate seasons - mainly autumn - and the summer is quite dry. In Chihuahua, the climate is highly continental with rains concentrated during the summer season. We have tried to estimate TIW as a weighted average value from the "average monthly temperature" and "average monthly precipitation" data, which is surely inaccurate, according to the above observation about the differences between air and rain temperatures. The result is nevertheless interesting: TiW Florence: 13.8°C, TiW Chihuahua: 21.6°C Indeed, not far from Florence caves have a reference temperature of about 14°C at 0 m asl (corrected with altitude) whereas a cave like Grutas Nombre de Dios, near Chihuahua, shows a temperature above 25°C. The difference between the estimated TIW and TL is, therefore, very small in Florence - and in general in Europe - but it could be quite large in other climates. It is, then, very important to carefully make these estimations, although the effect of this error is quite small if compared to the next one we are going to describe. The main reason for a difference between TL and TiW is what we denote as the "fluid selection", a huge class of processes that considers underground flow of water, and secondarily air. Water freezes below 0°C; therefore there is no infiltration when outside temperature is below zero. The negative parts of the sinusoid, which usually describes the average local temperature TL, are then eliminated. However, during the melting season the water inflows underground at temperatures not much different from 0°C; despite that the air temperature would be much higher. It is quite easy to prove that this effect cannot compensate the previous one. This means that in regions where a significant part of yearly precipitation comes as snow, the rock temperature TR is significantly higher than the TL and always above 0°C. The same can be stated for the cave temperatures. The selection of infiltrating water warms a rock up -especially in karstic areas - and extends the internal isothermal zero much above the external. As an example, the Ulugh Beg cave (Hodja Gur Gur Ata ridge, Uzbekistan), at an altitude of 3750 m asl shows an estimated TL of -6°C. The altitude of the external isothermal 0 is around 2800 m asl. The cave has important ice deposits in the first 300 meters and the measured internal temperatures are -0.8°C (3700 m asl) and 0°C (3550 m asl) (Bernabei & De Vivo 1992). This case is quite extreme (AT=5°C), but it is usual to measure smaller but significant temperature differences. Another important "selection" of inflowing water is influenced by the external morphology. Cave entrances can be extremely large and deep and can concentrate large snow deposits that, during the hot season, melt and cool the cave temperature because of the presence of melting water, always at 0°C, whereas the TL can be much higher. An example can be found in the Slovenian karst (cave Velika ledenica v Paradani), which is an ice cave at an altitude around 1000 m asl, where TL is around 6-8°C. There are other effects due to external morphology, which can produce large differences between TIW and TL, causing anomalies in the temperature of water infiltration. For instance, into the cave enters water coming from a lake -or slow and shallow- water fluxes highly exposed to sun, that during summer could attain temperatures significantly higher than air. A minor fluid selection effect, but important for ice deposit formation, is the selection of air flowing inside the cave entrances. When the air density inside a vertically extended cave is lower than the external, the air flows upward (the humidity for the estimation of air density is not considered here). The surrounding cold air cannot flow into the cave; consequently the upper parts of the underground system do not freeze. Despite the cold air being unable to enter, it however contributes to the statistics that create the average local temperature TL. During the summer, the internal atmosphere is denser than the external, and the air falls down, inhaling the external air from the upper entrances where now enters the external air, which is warmer. tte internal air, relatively cold, descends down the mountain. At the lower entrances the process is similar but inverse, and can contribute to the formation of internal glaciers; a classical example is the Kungur cave (Perm region, Russia), formed inside a relatively thin layer of gypsum, where permanent ice deposits exist near the lower entrance of this cave, and where the conduits can attain very low temperature whereas the regional average yearly temperature is around +1.6°C (Worldclimate 2010). Caves therefore "filter" the airflows on the basis of their temperature, systematically choosing between the external and internal, the hotter for the upper entrances and the colder for lower entrances. tte result is that the average temperature of airflow that goes across the entrance is not TL, but the average local temperature during the months when the internal temperature is lower (in the upper parts) or higher (in the lower parts). Usually, this is a minor effect because the total thermal capacity entering a cave through the air is greatly lower than the water thermal capacity, but it can be very important locally - for instance, near the entrances, forming underground glaciers - or in dry regions. ttis point will be thoroughly discussed later. The Tr versus T1W We can now discuss the second "almost": the difference between the rock temperature TR and the temperature of infiltrated water TIW. The thermal energy exchange between rock and water depends on the difference between their temperatures. As long as these two values of temperature become similar, the system tends to become adiabatic (no heat exchanges between water and rock) and it falls asymptotically to the theoretical, uniform temperature, which in fact it never attains perfectly. Moreover, the time-scale to obtain thermodynamic equilibrium between a mountain and the infiltrating water is on the order of centuries and it is difficult to assume TiW as constant. Therefore, the rock temperature value is always close to the external antecedent one, which it is never exactly the present. Another source of disequilibrium between the average temperature of rock and water is the airflow contribution. The specific thermal capacity of air is roughly four times less than the water one (around 1000 J kg -'K-' -at constant pressure- against the 4200 of water; the mass flows of these two fluids are, however, highly variable in time and space, then the estimation of the total thermal capacity flux (i.e. the product of the mass flow rate multiplied by the fluid specific thermal capacity) flowing underground is complex. We need therefore to briefly discuss the relative water-air contribution. The air flux ranges from zero (no air circulation in a deep karst: it is not unusual), to 10-40 m3/s in large Alpine cave systems (e.g., Corchia and Piaggia Bella, in Italy, with surfaces of respectively 3 and 20 km2), up to 100-200 m3/s along the main conduits in some large tropical caves (e.g., Hwanseon Cave, in Korea, probably with a surface around 100 km2). The air drainage is associated with the cave shape; the parameter "specific air drainage," indicated by the ratio between the whole flux and the whole karst surface is, then, almost meaningless yet the most significant parameter to be considered in the estimations. The specific air drainage is then around 10 kg/km2 in Corchia and 1-2 kg/km2 in Piaggia Bella and Hwanseon. For developed Alpine karst, we can therefore assume as a reasonable range of specific airflow of 1-10 kg/s/km2, a value comparable to the common precipitation of 500-2000 mm/a, then a water flow of 15-60 kg/s/km2. If we include the high specific thermal capacity of air and water, we infer that the air usually contributes only a few percent to establishing the cave temperature. This general statement disagrees with Luetscher & Jeannin (2004) by which the air flow can play a leading role in establishing the temperature and the internal lapse rates. They estimate the same water flux, but an air flux of 150 m3/s/km2, which can be true in special cases (see below) but in our opinion this is absolutely overestimated in Alpine karst. We would have to observe exiting or entering air flows of hundredth of cubic meter per second each square kilometer of karstic area, which is too much. It is not even possible to assume that this flux takes place almost unseen along small fractures. The friction loss to air flux depends on v2, then the flow is highly concentrated along the wider conduits: the air flow along small fracture does exist, but its contribution to the total flow rate is generally small. In some special cases the air relative weight to the thermal capacity flow can however be much higher, for example in very dry regions, but also in rainy regions inside caves with very strong air fluxes, like in the Auyan-tepui, Venezuela (Photo 1). In this case the internal lapse rates are near -5°C/km (Fig. 3b) and the air is responsible for small differences between TIW and TR, causing water condensation on the cave walls. It is nevertheless very difficult to estimate the impact of airflow on cave thermal imbalance. The air flux does not depend only on an external system (the atmosphere) that behaves quite regularly, but it mainly de- a) abisso di trebidano 12,5 12 ?11,s ü 11 i 10,5 I 10 I 9 8,5 8 200 Altitude [m asl] 400 c) Ca pitano Paff S" 1,5 ■ * • ■ ■ ■ ♦ 1 ■ » » * ■ a ♦ ■ ■ «water »sir 14da 1600 laod 2000 Altitude [m asi] e) Pozzo della Neve 9 E ' ! 6,5 6 5,5 5 ■ ■ ■ I * ♦ ♦ I » ■ « * * I ■ 600 sao looa 1200 Altituda [m asi] g) abisso di haiga fossetta 1. 1 't». • ^ f ■ aao icioo 1200 1400 laoo laoo Altitude [m asi] b) sistema de la sima aonda >1, which is usually valid for caves. Attenuation of the temperature oscillations rise with increasing frequency of the signal and the ratio between the heat capacity of the massif and heat flow given to the system. MEASURE OF THERMAL INSULATION The ratio Q between the external temperature and the local internal range (Ta/AT) indicates a direct measure of the thermal insulation (or the "thermal coupling" with the outside) of each sector in the cave. A lot of research has been done in the past both in the field of seasonal temperature variation in caves and on temperature spatial variation along the cave. tte amplitudes range from 2°C (Jernigan, 2001) to 0.10°C (Ga-doros 1989), but the most common values are around 0.15-0.3°C. If we compare this value to the typical external temperature range (10°C), we see that the seasonal Q is usually around 30-50. To measure daily temperature variations is very difficult, yet possible, with parallel sensors and special data analysis as above described. Figs. 7, 8 and 9 show the internal daily temperature range of two very different caves, Rio Martino (Piedmont), and San Giovanni Domusnovas (Sardinia, Italy). The first is a typical Alpine cave, crossed by a small river and a strong airflow. tte yearly amplitude of temperature variation is 0.12°C (Q~90), but the daily variation is much smaller, 0.012°C. ttis corresponds to a Q~5/0.012=400. tte second is a huge, natural hydro-geological tunnel, crossed by a small river and intense airflow; its daily temperature range is 1.8 °C, which corresponds to a Q=2.8, about two orders of magnitude less than Rio Martino. ^e two can be considered "high energy" caves. ttis coefficient of local thermal insulation is probably a fundamental parameter to understand many de- Fig. 8: ne Log-plot of daily temperature range frequency distribution in the Rio Martino cave. ^e distribution appears to be Log-normal. Fig. 7: ne frequency distribution of daily temperature ranges inside the Rio Martino cave, Piedmont, Italy. Fig. 9: ne frequency distribution of daily temperature ranges inside the San Giovanni Domusnovas cave, near Iglesias, Sardinia, Italy. Fig. 10: Temperature variation induced by cold inflowing water from snow melting in Rio Martino. Red line gives the daily external Positive-Degree function, which gives a measure of melting. tails about the cave morphology, as well as speleothemes formation, condensation, and so on. It can be noted that it is possible to make a different model for temperature range "fading" inside caves (Badino 2004b), assuming a cave behaves like a conductive wall where a sinusoidal temperature fluctuation propagates by diffusion. This physical model can also suggest the use of the same parameter Q for adiabaticity, although this will provide different values for the phase shift. It is interesting to note that while the first model has no "memory" about the previous fluctuation (all the inflowing flux is mixed inside C), the wall model main- tains memory of past cold or warm waves inside the system. We can conclude with a detail. The general rule that refers to external heating inducing an internal temperature increase, usually with reduced amplitude -and the opposite for external cooling-, is reasonable and generally true, but it is here necessary to carefully consider the above-mentioned concept of "fluid selection." Fig. 10 shows the underground behavior of the cave Rio Martino at an altitude of 1450 m asl, when a warm front arrived during the winter, causing snow to melt above the cave and a resulting in water inflow with a temperature around zero, which cooled down the cave. In general, we can say that if the cave temperature is above zero and the region of infiltration is covered by snow, a warm climatic episode during winter could provoke a cave cooling. Similarly, warming or cooling of external air can result in airflow changes and, possibly, in cooling or warm- ing of parts exposed to airflow. WATER VAPOR METEORA Ben sai come ne laere si raccoglie quell'umido vapor che in acqua riede, tosto che sale dove 'l freddo il coglie. Well knowest thou how in the air is gathered That humid vapour which to water turns, Soon as it rises where the cold doth grasp it. Purgatorio, F INTRODUCTION Water vapor phenomenology surely plays a key role in speleogenesis. The cave walls are very often covered by water films that play an important role in rock corrosion. The speleogenetic role of condensation is highlighted by the insoluble inclusions, which outcrop from the limestone. In this case the rock is usually uniformly corroded around the inclusion, by some isotropic proc- ess which is able to corrode in the same manner above and below the inclusion. Also the enormous quartz-ite shafts in the Venezuelan Tepui (Piccini & Mecchia 2008) have the walls uniformly covered by condensation, which seems the main cause of their very typical rock surface weathering throughout the "dissolution" of the quartzite cement for the first few centimeters of thickness. The result is a compacted sand rock surface. Finally, the position of many speleothemes at the bottom of convex surface on the roof, suggest a possible formation due to local rock solution and deposition -without water infiltration- driven by condensation processes. For a caver, a cave is usually perceived as "active" when a water film covers the walls. It is therefore important to analyze the condensation processes occurring in caves, because in a closed system at the equilibrium -maximum entropy- there would not be condensation, and the temperature would strictly be homogeneous. We need now to analyze the small terms, which are able to open a cave system, and create disequilibria. SUPERSATURATION PROCESSES tte main reason for disequilibrium is a sudden cooling of a fluid -air or water- when this penetrates underground. In general, the difference of temperature between fluids and rock creates an imbalance. This can also happen at large depths, but it is mainly a process at the "interface" between the epi- and hypogean environments. This process can also affect the energy balance of most shallow areas of a cave as well as the stability of internal glaciers. It can generate a net water flow into a cave -usually very small- but it is typically a local and low depth process. We are instead going to focus our attention on global processes that are due to the continuous fluctuation of fluids around equilibrium and that are active at large depths too (Badino 2005b). Inside a cave system, essentially five processes can create condensation, all connected to the presence of airflow: 1) upward movements of air parcels (adiabatic cooling) 2) mixing of air fluxes 3) air expansion in throttles 4) condensation from hazes 5) "Raoult" mixing. UPWARD MOIST AIR MOTION Outside, the cooling generated by the upward movement of air parcels is the main reason for cloud formation. Underground, the situation is rather similar but it is always very difficult to observe the clouds there. We have already seen that in the free atmosphere, the average air cooling along an upward external rise is around 6.5°C/km, very near to the moist adiabatic lapse rate. ttis means that each cubic meter of air cools 6.5°C in 1 km of rise, therefore: i) it does not release thermal energy to the surrounding environment and, ii) it condenses 6 grams of water vapor. Inside caves the situation is slightly different, because the lapse rate is usually around -3.5°C/km. tte transformation is not adiabatic, and this means that the rise of an air particle underground subtracts thermal energy, cooling the environment, and releases only 3.5 grams of water per kilometer of rise. Haze, clouds and "rain" appear: the cave walls get wet. It is easy to see the haze in suspension, looking at a distant light in large grottos, or watching the bright droplets in suspension in the few centimeters of air in front of a strong light. The "rainy season" then happens during the upward transfer of air parcels, generally when the external temperature drops below the internal temperature. This explains why large temperature fluctuations of the external atmosphere directly affect the underground energetic balance and condensation, even if the net air inflows into the cave are very small. Descending branches in a globally ascending cave obviously show an opposite phenomenology, as can be easily detected by looking at the hazes in suspension along ascending and descending conduits. As previously noted, there are evidences that these seasonal processes affect the thermal stratification inside the conduits. Let us focus on the specific deposition points on the cave walls. Condensation causes a local energy release that stops the condensation process, unless the energy excess is removed inside the rock and to the air. At the equilibrium, the condensation processes are therefore driven by the efficiency of these two processes of energy removal, which depends on the rock thermal diffusivity, the surface morphology, its orientation and so on (Dreybrodt et al. 2005). It is quite common to observe that, in a regular conduit, the condensation is usually quite uniform, but this can be an interesting field of research. Unfortunately, caves are far from being regular, and water deposition usually tends to concentrate at specific points. To start, the development of a droplet always needs a condensation nucleus. Inside an air particle flowing upward, the condensation nuclei -impurities- are quickly captured by droplets and eventually transferred to the walls. This is, by the way, the reason for the usual extreme air purity inside caves, although it depends on the direction of the airflow motion. Becoming poor in nuclei, depleted from the air by droplets formation, the air can locally become supersaturated, and water condenses on the cave walls as soon as it reaches them. The deposition is thus concentrated on the relatively narrow passages following a wide area. This is the first process that indicates that relatively narrow passages are preferred points of condensation, and therefore preferred points of rock dissolution. AIR FLUXES MIXING The condensation created by adiabatic cooling of moist air particle is, by far, the most common in the free atmosphere, but in the caves atmosphere other processes become important. The most common to be observed is the condensation due to the mixture of two saturated air particles at different temperatures. The resulting "mixing clouds" are so common underground, to the point that cavers do not observe them: the clouds created by breathing or moist cloths are typical examples of mixing clouds. The scientific literature has plenty of references about the mixing clouds formation mechanism and its common misinterpretation; the opinion that they form "because the warm breath is cooled...", and so on, is not true. "Heating and cooling itself are irrelevant... mixing clouds are formed by mixing of different air parcels... because of the shape of the saturation vapour pressure curve two parcels can mix to form a supersaturated parcel" (Bohren & Albrecht 1998). ^eir role in cave atmospheres has been only recently considered (Lismonde 2002). This is, in fact, a mechanism connected to the non-linearity of the Clapeyron curve (Badino 2005b), and essentially the same as the so-called "Bögli's mixing corrosion" (Bögli 1965). Temperature drops in caves are commonly due to different "histories" of water and air columns flowing along different branches, often characterized by different temperature gradients along the flow, but mainly by different temperatures at the entrance (Badino 1995), especially in the presence of very steep external topographies. It is easy to see that super-saturations are low and not directly measurable, some 0.01-0.1% of relative humidity. These values, nevertheless, correspond to a total release of some milligrams per cubic meter of air reaching the mixing point. A flow rate of 10 m3/s, quite usual underground, can release 103 kg of unsatured water per year in the mixing region. tte areas of confluence of different conduits, where different airflows are mixed, often present supersaturated atmospheres. There, we can expect to find a sudden conduit dimension increase and observe water films around us. This process depends on the airflows strength and direction and therefore it is dependent on the external temperature. It is then another seasonal effect. AIR EXPANSION IN BOTTLENECKS The clouds that sometimes appear leeward large peaks are called "banner clouds." Their phenomenology is not completely understood, but they appear to be the result of a sudden (i.e. adiabatic) air expansion that locally cools the gas below the dew point (Friedlander 2000). Let us now consider the airflow in a cave system. The flux through a mountain, as a whole, can be considered a Joule-Thomson expansion because the air parcel is de facto "throttled" through the cave. tte general transformation is isenthalpic. tte actual process seems extremely complex, showing energy dissipation by friction, temperature and flow periodicities, thermal exchanges on the cave wall. The whole transformation results in a small temperature increase due to the dissipation of mechanical energy (air pressure drop at the passage ends), but in the proximity of the downstream end the cooling generated by the adiabatic expansion can prevail, as in the case of external banner clouds. The pressure drop at the edge of a narrow conduit can be measured with an altimeter. In Corchia (Tuscany, Italy) we have measured values around 100 Pa for an air speed of 5 m/s, which is roughly equivalent to an upward movement of some 10 m for the air particle. It has been stated above that the adiabatic rise of an air particle causes 5°C of cooling per kilometer of rise, then flowing through the throttle with this pressure drop, the air has a rapid cooling of some hundredths of a degree, and the related water condensation (some hundredths of gram per cubic meter of flowing air) represent the result of this process. The air throttling down bottlenecks creates stationary and invisible clouds, downstream of the narrow passages ("banner clouds"). This is a second process indicating preferred condensation on relatively narrow passages. CONDENSATION FROM HAZE The Clapeyron law describes the equilibrium pressure above a flat water surface. If the surface shows a radius of curvature comparable to the intra-molecular interaction length -a small droplet- then a molecule on the surface is less bound to the liquid than to a flat surface, resulting in a strong tendency to evaporate. The equilibrium pressure, described by the Kelvin equation, is then higher (Fletcher 1969; Rogers & Yau 1989). Incidentally, we note that the very common presence of a stable aerosol (or haze) in a cave atmosphere is a direct evidence of supersaturation of moist air around droplets: in a simple "saturated" atmosphere the haze would quickly evaporate. In the case of a negative radius of curvature - a concave surface - the surface needs a smaller vapor pressure to reach equilibrium and becomes a preferred condensing point for the surrounding air in equilibrium with a flat surface. This means that small rock fractures tend to be filled up with aggressive condensed water. This process of a preferred point of condensation is very evident on the giant gypsum crystals of the Ojo de la Reina cave in Naica, Mexico (Photo 3) (Badino et al. 2010). A droplet produced by mechanical fragmentation, as happens at the base of a waterfall, is able to create a local super-saturation; the water is transferred from the waterfall to relatively distant walls not only by direct droplet deposition but also by condensation, because Photo 3: Ojo de la Reina, Naica, Mexico. Micro-crystals formation on the surface of gypsum megacrystals due to condensation in microfractures. the water vapor around droplets is always supersaturated relative to flat rock surfaces. RAOULT CONDENSATION Raoult's Law describes the decrease of vapor pressure for saline water. Practically, salty water evaporates with lower rates than pure water, and if pure salt is exposed to moist air, some water will condense on the crystal surfaces. This happens because the water vapor in the air becomes supersaturated in comparison to the salty solution. A typical carbonate concentration value in karst water is around 100 mg/l, which corresponds to a mole fraction Xs=2x10-5; the equilibrium pressure above karst-ic water is some 0.001% lower than above pure water. tte correction is very small, but does exist. The global effect is the water transfer from a less to a more salty water surface and therefore, for instance, to soak the conduit walls if the water flowing inside has a lower salt content than the water film on the rock. The water exchange between internal waters with different salt concentrations commonly happens and it represents again a speleogenetic process occurring at the intersections of different water and air drainages. AIRFLOW METEORA Io venni in loco dogne luce muto, che mugghia come fa mar per tempesta, se da contrari venti e combattuto. I came into a place mute of all light, Which bellows as the sea does in a tempest, If by opposing winds't is combated. Inferno, III A similar, more precise measurement, made in coincidence between two different entrances of the same huge cave (Corchia, Tuscany, Italy), shows a similar behavior (Fig. 13). Why? OSCILLATION PHENOMENOLOGY Cave atmospheres are essentially quasi-closed air masses very close to thermodynamic and buoyancy equilibrium, connected by relatively small conduits to the INTRODUCTION We have seen that convective air circulation in caves depends on the buoyancy and friction of internal air, and that wind velocities would depend on the square root of AT, the difference between the instantaneous internal and external temperatures. If we plot the wind velocity versus the square root of the difference of temperature, we expect to obtain a straight line. We made this measurement at the entrance of the Su Bentu cave system (Sardinia, Italy) obtaining the plot as shown (Fig. 11)- not exactly a straight line ... Plotting the ratio (wind velocity)/(square root of AT) versus time (Fig. 12), it is possible to see that the anomalies are actually concentrated; in fact, the general trend does not generally differ from a straight line; however, during two short periods of time the system has not behaved as expected in case of convective circulation. Fig. 11: Su Bentu cave, near Nuoro, Sardinia, Italy. In the convec-tive model of air circulation, the wind velocity and the square root of the difference between internal and external temperature are proportional, then plotting one vs. the other would give a straight line. It looks far to be true. Su Bentu: Wind Speed vs Buoyancy 5,00 ™ 4,00 « e ll' 3,00 2,00 1,00 0,00 s * • • • •fV • 0,00 0,50 1,00 1,50 SqrttTeJit-TInt) 2,00 2,50 Fig. 12: Su Bentu cave, near Nuoro, Sardinia, Italy. ^e ratio between wind velocity and the square root of the difference between internal and external temperature has to be constant. In fact, it is quite constant except during definite periods. The whole mass of internal air atmosphere starts oscillating back and forth. It has been well known for a long time that cave winds are sometimes periodic (Plummer 1969). In the field, observation of these air fluctuations is only possible when the average wind speed is very small, nearly equal zero. The theory has, however, shown that the oscillations must also occur in presence of strong winds, and that every cave atmosphere contains its intrinsic harmonics and "timbre" that depend on the cave morphology (Badino 1995). It is therefore expected that an underground wind measured at the cave entrance should contain information on both the internal and external atmosphere dynamics, as well as on the geometry of the underground system (Plummer 1969; Lismonde 2002). In fact, it can be theoretically shown that the shape of each single conduit has effects everywhere inside the cave, therefore the air movement in each contains information about the whole cave. Elementary acoustical models can describe the origin of infrasound harmonics; however, cave morphology is usually very complex, and the harmonic spectra of its "voices" are extremely complex. Fig. 13: ne ratio between wind velocity and the square root of the difference between internal and external temperature vs. time, measured in coincidence at the two main entrances, Eolo and Serpente, of the Corchia cave system (Tuscany, Italy). Sometimes the airflow has the opposite direction than expected by the convective model. variable external atmosphere. This causes a continuous energy transfer between the outside and the inside (Badino 1995). A quasi-closed air mass is able to store energy because of the difference of pressure between itself and the air masses connected to it. On the other hand, a moving air parcel has some inertia and cannot stop immediately when the driving pressure is switched off. When an internal-external imbalance occurs, an energy release to the system takes place, and the energy flows and oscillates between the different internal kinetic and pressure energy "reservoirs", like a moving swing. CAVES HARMONICS AND NOISE The presence of rhythmic behavior of airflow at the entrance of caves has usually been associated with very big caves, located at intermediate altitude. This could be fairly true in general, although such events can be noticed - without the instrumental support (anemometer) - only if a complete reversal of flow occurs, meaning an average flow around zero -intermediate entrances - and only if the period of oscillations is comparable with the time that usually cavers spend in the windy conduit. Our measures (more than 3 million in 12 entrances of 5 cave systems) have shown that apparently very regular airflows can have a complex structure (Fig. 14) and that every cave is able to emit infrasounds. The "cave player" is obviously the external atmosphere, which, during windy and perturbed days, can excite the main cave harmonics. tte energy is stored and the air masses start to oscillate, releasing energy to other harmonics and to distant cave regions. Although a short period of stimulation can result in hours of oscillations, the general rule is that under common atmospheric conditions only noise signal is acquired, which on average overwhelms the harmonics. Although the data analysis is extremely complex and we are still "tuning" our experimental approach, we can however already make some observations based on the available measurements. Fig. 14: Corchia cave complex (Tuscany), Eolo entrance. ^e apparently constant airflow has in fact an extremely complex structure connected with the cave morphology. Fig. 15: Corchia cave complex (Tuscany), Eolo entrance. A typical air flow spectral power. ^e intrinsic harmonics are usually submerged by a pink noise. The first observation is that caves emit infrasounds with frequencies around 10-2 and 10-3 Hz, on discrete (quantized) harmonics. Low frequencies seem to characterize the whole system, whereas "high" frequencies are emitted only "locally," i.e. by surrounding conduits. For instance, the Mt. Corchia cave system was recently studied by placing two identical anemometers on the two far system entrances (Eolo and Serpente). Spectra show a significant common low frequency ("Deep harmonic"? "Fundamental harmonic"?) of 0.000295 Hz, which corresponds to a period of 3400 s. It would represent the note emitted by a 280 km tube, closed at one side... The second result is about the noise spectrum. The power spectra of sampled signals show an overall decay at increasing frequencies, following a power law f-n, often with n=2.3-2.7, but which depends on the frequency scale. Depending on the value of the exponent of the power law, different scenarios can be drawn: in the 1/f regimen (pink or flicker noise) the energy decreases linearly with frequency, then each octave contains the same energy. Pink noise is intermediate between the 1/f" noise (white noise), where the system has no memory of previous states and the energy is equally distributed among frequencies, and the 1/f2 noise (Brownian noise), where the system has no memory of previous increments. We then show that the overall power spectra of large underground systems usually tend to behave as pink noise (Fig. 15). Pink noise is rather ubiquitous and occurs in very different natural systems. It has been shown to appear in complex systems with superposition of dynamic processes that relax exponentially with different characteristic times (Milotti 2002). These results are very interesting because from the infra-acoustic point of view, complex caves can be considered as systems of many different damped and coupled oscillators, and this could probably be considered the origin of the pink noise. CONCLUSIONS A cave is commonly considered to be a nearly-invariable environment: daily or seasonal climatic cycles cannot infiltrate, the weather stays perfectly stable, and meteorological fluctuations are absent. Furthermore, this stable condition would homogeneously occur anywhere inside a cave; the temperature would not change along the year, also maintaining constant along the entire cave length. Caves appear to be the realm of "invariability". This belief is mainly generated by the observation of relatively small amplitude variations for both internal temperature and humidity, if compared to the external variations. The variations of amplitude inside caves are in fact fairly small; however, this does not justify the common thought that, in a cave, either climatic or meteorological processes do not actually take place. In the real world, external cycles will appear underground in the form of: variations of airflow direction and intensity; imperceptible variations of seasonal and daily temperature; local air and water temperature disequilib-ria; local thermal air sedimentation, and water film condensation on the cave walls. The best way to highlight these processes is to carefully analyze the subtle parameters responsible of underground meteorology: the spatial and temporal temperature variations; the energy exchanges, that will always locally happen if small differences in temperature occur; the condensation-evaporation processes, which cover walls with thin films of water and create internal clouds; the air fluxes, with their imperceptible but continuous changes of speed. tte phenomenology of the two main flowing fluids in a cave -water and air- is completely different. Water can only flow downward and on the lowest part of conduits; due to its enormous thermal capacity, water generally has the leading role in establishing the underground temperature. Nevertheless, its flow is not adiabatic and water normally subtracts energy from the underground systems. Air can flow everywhere, reverse its flowing direction and, consequently, the thermal exchanges; air can make the rock surfaces wet; by its water vapor content air can smooth the temperature variations and dominates the energy exchanges. More importantly, along the cave length the nature of thermodynamic transformations of air is not adiabatic. This means that a thermal exchange between air and water will continuously take place. Finally, airflow generally releases energy to the cave as a whole. Measurements suggest that the meteorological variability among caves, and also among different points within a cave, is relatively high, and probably can explain many details of their structure. These variations from one cave to the other are relatively much larger than the variability of the external environment, where the astonishing changeable Earth landscape are created by relatively small differences in precipitation, temperature ranges and absolute humidity. tte characterization of these micro-meteorological variations is, probably, the key for a better understanding of the underground world. ACKNOWLEDGEMENTS This study would have not been possible without the enormous help, during years, given by many cavers for the field measurements; without the financial support of Associazione Gruppi Speleologici Piemontesi for the instruments; without numberless observations, questions made and suggestions provided by numerous speleolo- gists. Special thanks are due to Roberto Chignola and Ar-rigo Cigna for discussions, and above all to Daniela Pani and Franci Gabrovsek for their general observations, encouragement and help in writing. But very special thanks are owed to the caves, which have patiently helped to let me measuring them, for such a long time. REFERENCES Badino, G., 1995: Fisica del Clima Sotterraneo - Memo-rie IIS, 7, II, pp. 137, Bologna Badino, G., 2000: I Gradienti di Temperatura nei Monti, un Indicatore Esplorativo.- Talp-FST, 21, 72-80. Badino, G., 2004: L'Influenza del Clima Esterno sulle Terre della Notte. - Memorie IIS, II, xVIII, 2004 Badino, G., 2004:Cave temperature and global climatic chnages.- Int. J. Speleol., 33,103-114. Badino, G., 2005: Underground Drainage Systems and Geothermal Flux. - Acta Carsologica, 34/2, 1, 277316. Badino, G., 2005: Clouds in Caves.- Speleogenesis and Evolution of Karst Aquifers.- Speleogenesis and evolution of karst aquifers, 2/2, [Online] Available from http://www.speleogenesis.info [Accessed 30«^ .11.2010]. Badino, G., Calaforra, J., Forti, P., Garofalo, P. & L. San-na, 2010: tte Present Day Genesis and Evolution of Cave Minerals inside the Ojo de la Reina Cave (Naica, Mexico) - IMA 2010, submitted to Int. J. Speleology. Bakalowicz, M., 1972: La Riviere Souterraine de Pinar-gozu - Annales de Speleologie, 27/1, 93-103. Bernabei, T. & A. De Vivo (ed.), 1992: Grotte e Storie dell'Asia Centrale - Centro Editoriale Veneto, 310 pp., Padova. Bögli, A., 1965: tte Role of Corrosion by Mixed Water in Cave Forming - in: Stekl O. (ed.), Problems of the Speleological Research. Czechoslovak Academy of Science, pp. 125-131, Prague. Bohren, C. & B. Albrecht, 1998: Atmospheric ^ermo-dynamics.- Oxford University Press, pp. 402, New York. Celico, P., 1986: Prospezioni Idrogeologiche.- Liguori, pp. 736, Napoli. Fleagle, R. & J. Businger, 1980: An Introduction to Atmospheric Physics.- Academic Press, pp. 432. Fletcher, N., 1969: tte Physics of Rainclouds - Cambridge University Press, pp. 389, Cambridge. Friedlander, S., 2000: Smoke, Dust, and Haze - Oxford University Press, pp. 408, New York. Gadoros, M., 1989: tte Physical System of Speleocli-mate.- In Proceedings of 10*^ international conference of Speleology, Vol 3, 752-754, Budapest. Jernigan, J. & R. Swift, 2001: A Mathematical Model of Air Temperature in Mammoth Cave.- Journal of cave and karst studies, 63(1), 3-8. W.C. Lewis, 1991: Atmospheric pressure changes and cave airflow: a review.- National Speleological Society Bull., 53, 1-12. Lismonde, B., 2002: Aerologie des Systemes Karstique.-CDS Isere, pp. 361, Grenoble. Luetscher, M. & P.Y. Jeannin, 2004: Temperature distribution in karst systems: the role of air and water fluxes.- Terra Nova, 16, 344-350. Milotti, E., 2002: A Pedagogical Review of 1/f Noise.-Arxiv preprint physics/0204033 [Online] Available from arxiv.org [Accessed November 30'h, 2010]. Pflisch, A., Wiles, M., Horroks, R., Piasecki, J. & J. Ringeis, 2010: Dynamic Climatologic Processes of Barometric Cave Systems Using the Example of Jewel Cave and Wind Cave in South Dakota, USA.-Acta Carsologica, 39/3, xx-xx. Piccini, L. & M. Mecchia, 2008: Solution Weathering Rate and Origin of Karst Landforms and Caves in the Quartzite of Auyan-tepui (Gran Sabana, Venezuela).- Geomorphology, 106, 15-25. Pinna, M., 1977: Climatologia.- UTET, pp. 442, Torino. Plummer, W., 1969: Infrasonic Resonances in Natural Underground Cavities - Journal of the Acoustical Soc. of America, 46 (5), 1074-1080. Provalov, D., private communication, 2010. Rogers, R. & M. Yau, 1989: A Short Course in Cloud Physics.- Pergamon Press, pp. 227, Oxford. WorldClimate, 2010: [Online] Available from: http:// www.worldclimate.com [Accessed 24 February 2010]