Ventilation Conversion Mechanisms in Karst Caves

Ventilation modes in karst caves are of great signicance for exploring issues regarding "carbon sources and sinks" in karst areas. Therefore, this study conducted continuous monitoring of air temperature, humidity, and CO 2 concentrations inside and outside the Dafeng cave in Suiyang, Guizhou from August 2015 to July 2020 in order to comprehensively analyze each element using a systematic analysis method. The results revealed that: (1) the Dafeng cave is mainly divided into three ventilation modes: (cid:0) during summer and autumn, the inside of the cave is mainly in a restricted ventilation mode, wherein air exchange inside and outside the cave is suppressed, and the CO 2 concentration in the cave easily forms a cumulative effect; (cid:0)during winter, the inside of the cave is mainly in an active ventilation mode, wherein there is strong air exchange, and the CO 2 concentration inside the cave is close to that outside the cave; (cid:0) during spring, due to the variable climate and rising temperature, the inside of the cave gradually transitions from an active to restricted ventilation mode and the air exchange intensity gradually changes. The isotope data change characteristics outside the cave at the Yemingzhu monitoring site veries the ventilation mode of the tunnel during each season. (2) Due to the inuence of tourists, air changes inside and outside the cave, and the structure of the tunnel, there may be multiple ventilation modes within a single season. Tourists and tunnel structure primarily affect the structure of the tunnel. A change in the air environment outside the cave mainly affects the virtual temperature outside the cave, increasing or reducing the virtual temperature difference between the inside and outside of the cave, thereby affecting the ventilation mode. Thus, our study suggests that more consideration should be given to changes in external climate or weather conditions when studying the conversion mechanism of karst cave ventilation modes.


Introduction
Cave environment monitoring is an effective way to study karst carbon cycle processes (Peyraube et al., 2017), cave landscape formation (Fairchild and Baker, 2012), paleoclimate change (James et al., 2015), and cave environmental protection and management (Song et al., 2004;Lang et al., 2017;Lario and Soler, 2010). The cave environment (Banner et al., 2007) is mainly composed of the cave air CO 2 concentration, temperature, relative humidity, and chemical composition of the air (Yuan daoxian and Cai guihong, 1988). Among these components, the study of cave CO 2 has continually been of interest in tourist caves (Zhang Ping et al., 2017). CO 2 plays an important role in carbonate karst caves by participating in karst geochemistry (Spötl et al., 2005;Faimon et al., 2012b), karsti cation (Christoforou et al., 1996), and the formation and dissolution of cave chemical deposits (Dreybrodt et al., 1999;Frisia et al., 2011;Tarhule-Lips et al., 1998). Yuan Daoxian (Yuan Daoxian and Cai Guihong, 1988) studied the theory of karst dynamic systems, pointing out that CO 2 plays an important driving role in the three-phase imbalance system caused by karsti cation as it not only directly impacts the chemical deposit balance of caves but also impacts the system exchange process, wherein too high of a CO 2 concentration will greatly impact karst development, karst environment, and human health in karst areas between the carbon and water cycles (Kermode, 1979;Yang hankui, 1993;Zhu wenxiao et al., 1993).
Currently, there is increasing interest in characterizing the storage and ventilation of CO 2 in caves, both from external (atmospheric) and internal (speleological) perspectives. Keeling (1960) and Sánchez-Canete et al. (2013) revealed that human activities led to the accumulation of CO 2 in the atmosphere in the middle of the 20th century, which in turn led to global warming. Gregorič et al. (2014) proposed that the temperature gradient of outside air to cave air affects the gas exchange process, wherein cave air mainly exchanges with the outside atmosphere via convection (Serrano-Ortiz et al., 2010;Sanchez-Canete et al., 2011), thereby altering the CO 2 content inside the cave (Baldini et al., 2008) and affecting the dissolution and deposition of its chemical deposits (Badino et al., 2011). In brief, the convective movement between gas inside and outside the cave is called the ventilation effect (Sanchez-Canete et al., 2011). The driving force of cave ventilation can be primarily divided into dynamic and static driving forces (Cigna, 1968). The dynamic driving force mainly includes wind and water movement (Nachshon et al., 2012), while the static driving force includes pressure, temperature, and air composition (including water vapor, CO 2 , and CH 4 ). Caves act as cold air traps. In the cold season, outside air enters caves for gas exchange, whereas, in the warm season, caves appear to be more isolated and gas exchange is blocked. This phenomenon explains the high CO 2 level in caves (Peyraube et al., 2017). As of late, with the improvement of research techniques, certain progress has been made regarding ventilation effects research. However, but there still exist differences in the discrimination of ventilation modes, and the interaction between CO 2 and cave CO 2 remains to be explored in depth (Lang et al., 2015). Fernàndez-Cortes et al. (2006) and Milanolo and Gabrovšek (2009) estimated the air density change based on the temperature difference between the air inside the cave and the outside atmosphere before determining the ventilation mode of the cave. Based on a simulation experiment, Faimon et al. (2012) suggested that it is advisable to directly use the temperature difference between the inside and outside of the cave to determine the ventilation of the cave. However, as early as the 1980s, De Freitas et al. (1982 pointed out that the classi cation of the cave ventilation effect is based on the air inside and outside the cave, not temperature. Thus, it is not advisable to directly use the temperature difference for evaluation. Therefore, Kowalski and Sanchez-Canete (2010), based on the traditional de nition of virtual temperature and the error of water vapor, improved measurements of internal/external air density, revealing that when the internal CO 2 mole fraction signi cantly exceeds the atmospheric value, the calculation of the virtual temperature should consider the weight of the CO 2 . Mattey et al. (2016) and Gregoric et al. (2013) also utilized the virtual temperature, nding that there are two main ventilation modes in karst caves: active and restricted ventilation. However, further studies by Faimon et al. (2012) and Vieten et al. (2016) found that, in addition to these modes, there is also a third ventilation mode: the transitional ventilation mode.
However, due to the complexity and uniqueness of the cave structure, as well as the diversity of climatic conditions that determine the time and intensity of cave ventilation (Fairchild and Baker, 2012), the mechanism of cave ventilation effect is still not well understood. As the ventilation effect is the primary driving factor of CO 2 concentration changes in caves, it is important to further explore this topic (Kowalczk and Froelich 2010).
Overall, the estimation of cave ventilation can be achieved by a variety of methods, but many ignore the effect of high CO 2 concentration, leading to various errors in the discrimination of ventilation modes. Therefore, this research, using the Guizhou Suiyang wind Dafeng cave as an example, accurately calculates air buoyancy based on virtual temperature and analyzes the interaction between the cave and external environment in combination with the daily and seasonal CO 2 changes in the cave. Further, this study analyzes the ventilation mode's in uence on the air composition of the cave, and explores the mechanism of ventilation mode conversion, identifying the time in which the cave releases or stores CO 2 .

Study site
The study area is located in the Suiyang county of Zunyi city in the northern Guizhou province (28°08′00″N -28°20′00″N, 107°0230″E -107°2500″E), which belongs to the Furong River tributary Chiwu River underground river system. The Shuanghe cave system is the fth longest cave in the world and the longest cave in Asia (2019). It is located in the area of the Huangyujiang anticline and the Tuping complex syncline in the northern part of the Nanchuan-Guiding fold belt in the north and south of Sichuan and Guizhou. Because of the existence of tectonic forces in multiple directions, northeast, northwest, and south-north fold fault zones exist, wherein the cave area is surrounded by three-fold fault zones in a relatively rising triangle block (Li po et al., 2008). The strata exposed in the study area are primarily the Middle-Upper Cambrian Loushanguan Formation and the Lower Ordovician Tongzi Formation dolomite, lime dolomite, intercalated int, and argillaceous dolomite (Li Po et al., 2008). The climate is mid-subtropical monsoon, wherein the monthly average temperatures in January and July are 1 ℃ and 22.5 ℃, respectively, and the annual average temperature is 15.5 ℃. The annual average rainfall is 1210 mm, and the rainfall intensity is high, concentrated mostly from April to October. Vegetation types are mainly subtropical evergreen deciduous and broad-leaved mixed forests.
The Dafeng cave is a branch cave of the Shuanghe Cave system. It is a tourist cave with an entrance elevation of 734 m, width of 7.6 m, height of 4.5 m, and length of approximately 696 m. The height of the tunnel is between 1.7 m and 22.6 m, wherein the average height is 8.9 m. The total volume of the cave is approximately 64518 m 3 , with a total area of 4805 m 2 . In general, the cave tunnel is developed to the south, gradually widening toward the south. As shown in Fig. 1, the monitoring sites Yemingzhu and Shenquan are special as the cavity of Yemingzhu, which is a junction of three tunnels, is large (approximately 28008 m 3 ) wide, and the air circulation is fast. Meanwhile, Shenquan is located on the branch of the Dafeng tunnel, wherein it extends 0.05 km to the end of the branch cave. This tunnel is relatively closed off, making it di cult for the air to circulate. Conversely, the tunnel structures of several other monitoring sites are straight and near the cave entrance.

Data sources
Four routine monthly monitoring sites ( Fig. 1) were placed from the entrance of the Dafeng cave to the inside of the cave, wherein they were monitored from August 2015 to July 2020. The temperature, relative humidity, and CO 2 concentrations were monitored using the American Telaire-7001 portable infrared CO 2 meter. The temperature measurements ranged from -20 ℃ to 70 ℃ with an accuracy of ± 0.35 ℃, and the humidity measurement ranged from 5% to 100%. The CO 2 resolution was 0.001% with a measurement accuracy of ± 0.05%, and a concentration range of 0% -10%. Standard (380 mg/L) gas was used for calibration before monitoring, wherein the instrument was placed 2 m from the operator during operation to limit human interference. The data of the air inside and outside the cave were mainly collected using a DEV31-1 gas sampling bag. The indoor air was primarily measured using a Finnigan MAT company model MAT252 gas isotope mass spectrometer with a resolution of 200, accuracy of 0.01‰, and a mass range of 1-150 AU. The results were expressed using the international Vienna Pee Dee Belemnite (V-PDB) thousand-point difference standard. The meteorological data outside the cave were monitored by the United States Kestrel-4500 portable small weather station, collecting air temperature, relative humidity, and CO 2 concentration data with a measurement accuracy of ± 1.0 °C, ± 3%, and ± 0.05%, respectively. The rates are 0.1℃, 0.1%, and 0.001%, respectively. The number of visitors was calculated by a person at the ticket gate at the entrance of the cave with a time interval of 30 min. The collected data were analyzed and processed by several softwares, including ArcGIS, Origin 2017, Spss 19.0.

Virtual temperature calculation formula
The virtual temperature was calculated using the comprehensive temperature, humidity, and CO 2 concentration (Sánchez-Canete et al., 2013) as follows: where T is the temperature (°C), r v is the water vapor mixing ratio, and r c is the CO 2 mixing ratio. The virtual temperature difference between the inside and outside of the cave △Tv = Tv inside -Tv outside when △Tv > 0 indicates that the cave is in a state of active ventilation. Conversely, when △Tv < 0, the cave is in a state of restricted ventilation. δ 13 C calculation formula The data processing method of δ 13 Cco 2(g) was used as follows (Kang zhiqiang;and He shiyi, 2011): where δ 13 C represents the abundance value of the sample isotope (‰), R sample denotes the isotope ratio of the sample, and R PDB is the Pee Dee Belemnite (PDB)isotope ratio of the standard sample. Each sample was tested ve times, wherein the average was taken, and every 4 samples were measured in parallel to ensure the accuracy of the test results.

Analysis of characteristics of cave ventilation modes
During the monitoring periods from March to October (Fig. 2), the virtual temperatures of the air outside the cave were higher than the virtual temperature of the air inside the cave, indicating that the air density inside the cave is greater than that outside the cave. Therefore, during this period, the cave ventilation is restricted as the air exchange between the inside and outside of the cave was blocked. For the rest of the monitoring period, the air density outside the cave was higher than that inside the cave, indicating a state of active ventilation, wherein the air exchange was relatively strong. The CO 2 concentration change curve of the Yemingzhu monitoring site (Fig. 2) reveals seasonal changes during the monitoring period, mainly characterized by relatively high CO 2 concentrations during summer and autumn, and relatively low CO 2 concentrations in winter and spring.
In summer and autumn, the virtual temperature difference between the inside and outside of the cave varies from -10.0 ℃ to -2.7 ℃ and -8.9 ℃ to -0.2 ℃, respectively, wherein the average values are -6.4 ℃ and -4.6 ℃, respectively. In addition, the air density in the cave is higher than that outside, meaning air exchange is suppressed. Here, the cave is acting as a "cold air trap" and air ow will only occur near the entrance of the cave. Deeper into the cave, air exchange is suppressed, and therefore it is in "restricted ventilation mode". At the Yemingzhu monitoring site, the CO 2 concentration range is 672 -1351 ppm and 532 -1289 ppm. Note that existing studies have shown that the air CO 2 concentration at the Dafeng cave Yemingzhu monitoring site is 1150 ppm in summer and autumn (Pan Yanxi et al., 2016).
In general, the CO 2 concentration in the air uctuates around its background value (1150ppm). Based on the intensive data of the wind tunnel from October 1 to 7, 2018 (Fig. 3), the air CO 2 concentration in the cave experienced clear diurnal changes with the changing number of tourists. However, during some periods, the internal CO 2 concentration had high and low values that were different from the background value of the cave air CO 2 concentration (Fig. 3). It is assumed that this is largely caused by the cumulative effect of the CO 2 concentration and self-puri cation ability of the cave air environment.
During the winter, the virtual temperature difference between the inside and outside of the cave varies from 1.3 °C to 12.1 °C with an average of 6.7 °C. As this difference is greater than 0 °C, the virtual temperature of the air outside the cave is lower than that of the air inside, That is to say that the air density outside the cave is higher than that inside. The relatively cold air ow outside travels along the bottom of the cave to penetrate deeply, and exchanges with the relatively warm air ow inside the cave Page 7/15 until the two are in equilibrium (Fig. 2). This phenomenon is called "active ventilation mode". Note that the CO 2 concentration in the cave mainly uctuated between 397 -837 ppm (Fig. 2).
Finally, during the spring, the virtual temperature difference between the inside and outside of the cave varies from -2.1 ℃ to 10.4 °C with an average of 4.1 ℃. At this time, the virtual temperature difference is basically less than 0 °C, indicating that, for most of the spring, the cave is mostly in a restricted ventilation state as the air exchange is weak, and only a small part of the cave is in a state of active ventilation (February). As shown in Fig. 2, during the spring, the ventilation mode of the wind tunnel gradually transitioned from active to restricted, meaning the air exchange inside and outside the tunnel gradually weakened. During this period, the CO 2 concentration in the cave varied from 468 -871 ppm.
By comparing the CO 2 concentrations in the four seasons (Fig. 2), it is clear that the CO 2 concentrations in the summer and autumn are higher than those in the winter and spring. The highest CO 2 concentrations generally appear in August (autumn), while low values appear in January, February, and March (winter and spring). This is likely due to the conversion of the cave's ventilation mode and the change in the outside air environment.
Characteristic of δ 13 Cco 2(g) inside and outside the cave One year monitoring of stable carbon isotopes outside the cave showed that the value of δ 13 Cco 2(g) in the atmosphere outside the cave uctuated between -12.25‰ and -9.39‰ Fig. 4 . The mean value of δ 13 Cco 2(g) in spring, summer, autumn, and winter was -11.45‰, -11.09‰, -10.84‰, -10.78‰, respectively. Compared with the δ 13 Cco 2(g) outside the cave, the air in the cave, originating from the degassing of cave water, cave ventilation effect, tourist respiration, and the overlying soil of the cave, was mainly composed of the atmosphere and rich in light CO 2 . Generally, δ 13 Cco 2(g) ranged from -19.89‰ to -7.00‰. δ 13 Cco 2(g) was lighter in summer and autumn, with the value ranging from -19.89‰ to -15.83‰. The value of δ 13 Cco 2(g) in winter uctuated between -12.83‰ and -7.00‰, showing a trend of gradual heaviness. Conversely, the value of δ 13 Cco 2(g) in spring uctuated between -17.96‰ and -11.22‰, showing a yrend of gradual lightness.

Tourist in uence on ventilation mode
Studies have shown that an adults' hourly exhaled CO 2 concentration value is 17 -40 L (Lang et al., 2017;Zhang Yingju, 1988), as calculated on the basis of 25 L/person exhaled. The CO 2 concentrations from October 1 to 7, 2018 produced by tourists and the actual CO 2 concentration measured are summarized in Table 1. From this table, it is clear that the CO 2 concentration obtained by monitoring is much lower than the CO 2 concentration produced by tourists. However, this is contrary to the actual situation as the cave is in a restricted ventilation mode in autumn, and visitors will inevitably produce a certain amount of CO 2 . Therefore, the nal monitoring result should be the sum of both concentrations, even though the monitoring result is much lower than the sum of the two. This indicates that the air inside and outside the cave underwent convection, where the air inside the cave was replaced by the air ow entering the cave, which is a positive ventilation effect. Thus, the air ow in the cave is not only affected by the ventilation effect but also by the tourist activity.
Because of the effect of cave ventilation and the cumulative effect of the CO 2 concentrations, the air CO 2 concentration in the cave displayed obvious diurnal changes (Fig. 5). In general, the CO 2 concentration undergoes four stages of changes: a few tourists enter, a large number of tourists enter or stay, tourists start to leave and no tourists, and no tourists or before tourists enter stages. The CO 2 concentration value C(n) begins to rise slowly when a small number of tourists enter while it rises rapidly when a large number of tourists enter or stay in the cave, which is attributed to the cumulative effect. When the tourists leave, the CO 2 concentration, as a result of cave ventilation, gradually decreases to the concentration level it was before tourists entered, or even lower. As shown in Figs 2 and 3, the cave is in a restricted ventilation state, but the CO 2 concentration value is low at this time. This result is contradictory to the actual situation because a large number of tourists entering the cave will contribute a certain amount of CO 2 . Meanwhile, the temperature inside the cave will increase from the increased heat emitted by the tourists, and the air exchange inside and outside the cave will be greatly enhanced. However, the monitoring results do not display such changes. This data contradiction could be due to the "piston effect" caused by the activities of tourists, wherein the air ow in the cave is caused by the movement of tourists.
During the active ventilation period, the "piston effect" caused by tourists occurs when the direction of natural air ow is consistent with the direction of air ow generated by the tourists' activities. When the ventilation intensity in the cave is the sum of the two, the air in the cave does not easily form a cumulative effect, and the peak CO 2 concentration will not be very high, sometimes falling below the background CO 2 concentration (indicated by the blue line in Fig. 5). Conversely, during the restricted ventilation period, when the direction of the air ow generated by the activities of tourists is inconsistent with the direction of the natural air ow, the nal air ow direction of the cave depends on the strength of each ow. If the natural outward air ow in the cave is greater than that caused by the movement of tourists, the cave is still in a restricted ventilation state (Fig. 6), inhibiting air exchange and forming a cumulative effect (shown by the red line in Fig. 5). Therefore, the CO 2 concentration on the October 5 is higher because the natural air ow in the cave is less than the air ow generated by visitors' activities, causing the cave's ventilation mode to change from restricted to active. This veri es the phenomenon of different ventilation modes during the intensive monitoring process in October 2018.
The cave air CO 2 is primarily attributed to two aspects. The rst is CO 2 escape from the supersaturated cave water, wherein the δ 13 Cco 2(g) value mainly depends on the isotope fractionation of the CO 2 escape process and the δ 13 Cco 2(g) of the cave water sample Dissolved Inorganic Carbo (DIC). However, the δ 13 Cco 2(g) is relatively light because part of the carbon source is derived from organic origins that are rich in 12 C soil CO 2 , and 12 C preferentially enters the gas phase during CO 2 escape. The second aspect is the air exchange between inside and outside the cave. Here, the contributed δ 13 Cco 2(g) is equivalent to the δ 13 Cco 2(g) outside the cave, which is relatively heavy. Based on the change characteristics of δ 13 Cco 2(g) at the Yemingzhu monitoring site, the δ 13 Cco 2(g) is mostly light during summer and autumn and heavy in winter, uctuating between the two in spring. This is consistent with the ventilation mode of the cave during these seasons. When the cave is in a restricted ventilation state, air exchange is inhibited, and CO 2 mainly comes from dripping water and rock ssures in the cave, making the δ 13 Cco 2(g) value light. When the cave is in a state of active ventilation, air exchange is strong, and CO 2 mainly comes from outside the cave, making δ 13 Cco 2(g) heavier. When the cave is in transitioning from active to restricted ventilation, the air exchange intensity gradually weakens, and thus the δ 13 Cco 2(g) value gradually becomes lighter.
Therefore, the in uence of tourists on the ventilation mode and form is mainly re ected in the breathing of tourists and the "piston effect" caused by the movement of tourists in the cave. The characteristic change of δ 13 Cco 2(g) in the cave veri es this result.

Impact of air environment changes on ventilation modes
Cave air is the most important media connecting the inside and outside of the cave. When the temperature outside the cave rises and falls due to weather changes, corresponding changes will occur inside the cave, causing circulation between the air ow inside and outside the cave. As the depth of the cave increases, the temperature of the cave air will increase or decrease, but the amplitude of its uctuations will gradually decay until it reaches a constant temperature throughout the year. In winter, the temperature outside the cave is lower than that inside the cave, and the air density is higher. Therefore, a cooler air ow is formed outside the cave and ows inside along the entrance of the cave at the lower part of the tunnel. Then, the warmer air ow inside the cave ows out along the top of the cave. This cycle is repeated until the temperature inside and outside the cave is basically the same, and the air exchange stops. Meanwhile, during summer and autumn, the temperature inside the cave is signi cantly lower than that outside. Acting as a cold air trap, exchange between the cold air ow inside the cave and warm air ow outside hardly occurs. In the spring, as the climate changes from cold (winter) to hot (summer), these two phenomena co-exist. However, most of the time, the cave is in a state of restricted ventilation. When the weather starts to rise slowly outside the cave, the cave changes from restrictive to active ventilation.
Note that because the temperature change inside the cave is much smaller than that outside the cave, the magnitude of the temperature change outside the cave is basically the same as the temperature difference between the inside and outside of the cave. In Fig. 7, it is clear that the ventilation effect in the wind tunnel is more evident, and the farther away from the opening, the worse the ventilation. This occurs because it not only has the aforementioned wind direction characteristics but also the ventilation rate is proportional to the virtual temperature inside and outside the cave, wherein the difference is positively correlated. The multi-year data show that wind speeds are higher in midsummer (the hottest month) and midwinter (the coldest month), and smaller in spring and autumn. In terms of intensive data, combined with the virtual data inside and outside the cave during the monitoring period, it is clear that the air buoyancy difference inside and outside the cave was negative during the early morning of October 1 and 2, 2018. After that, the air buoyancy difference became positive, and there were two short-term negative values during the transition. After this change, the cave was mostly under active ventilation. There was continuous rainfall during October 2-7, which caused the temperature outside the cave to decrease (Fig.  3). When the temperature inside and outside the cave is equal or when the temperature outside the cave is lower than that inside, the direction of air exchange begins to change, thereby causing the cave to change from restrictive to active ventilation. Note that the two short-term negative values are mainly due to tourists in the cave. The air disturbance from tourists causes the air ow in the cave to produce turbulence, which causes the cave's self-puri cation of air CO 2 to be stronger than the CO 2 produced by tourists breathing and heat. This phenomenon shows that the cave ventilation mode depends on the cave's external climate changes due to changes in the environment, which explains why both ventilation modes can occur at times other than the spring.
In summary, changes in the air environment outside the cave will inevitably cause changes in the virtual temperature difference between the inside and outside of the cave, aggravate or slow down the air exchange between the inside and outside of the cave, and lead to a change in the ventilation mode.
In uence of cave space structure on ventilation mode Due to its special tunnel structure and shape, the Dafeng cave has different effects on CO 2 diffusion and circulation at different monitoring sites. As the tunnel structure of the Dafeng Tunnel is complex and changeable, the tunnel cross-section varies greatly. The ventilation intensity of different monitoring sites in the tunnel varies correspondingly. Where the tunnel is narrow, the heat and CO 2 generated by tourists are not easily diffused, and can easily accumulate. Where the tunnel is spacious, the air ow is smooth, and heat and CO 2 can be diffused in time, which is the main reason for CO 2 concentration uctuations. As shown in Fig. 7, it was found that the ventilation rate of each monitoring site gradually decreased from the entrance to deeper in the cave. As it is the connecting hub of three tunnels, the Yemingzhu monitoring site has a wide and at tunnel structure with a large cavity (28008 m 3 ) and many branch tunnels. Therefore, its ventilation intensity is larger than that of other monitoring sites. The CO 2 at Yemingzhu rises slowly and is not prone to accumulative effects. Conversely, the Shenquan monitoring site has a relatively high altitude, with a concave tunnel structure that is high on both sides and low in the middle. The right side of Shenquan is at the top of the branch tunnel. It is closed, and the air is not easy to circulate, causing the heat and CO 2 generated by tourists to not be easily diffused. According to the CO 2 density, it is easy for sedimentation to make the CO 2 produced by tourists to continuously increase on the basis of the former CO 2 , forming a cumulative effect on the cave. In Fig. 6, it is clear that the ventilation rate at Shenquan is much smaller than that of the other monitoring sites. This is the main reason why the CO 2 concentration at the Shenquan site is higher than that at the Yemingzhu site.

Conclusions
By examining the characteristics of long-term and short-term ventilation modes in wind tunnels, the following conclusions can be made: The wind tunnel is mainly divided into three ventilation modes. During summer and autumn, the inside of the cave is mainly in a restricted ventilation mode, where air exchange inside and outside the cave is suppressed, and the CO 2 in the cave easily forms a cumulative effect. During winter, the inside of the cave is mainly in active ventilation mode, meaning there is strong air exchange, and the CO 2 concentration inside the cave is close to that outside the cave. During spring, owing to the changeable climate and warmer temperature, the inside of the cave gradually transitions from an active to a restrictive ventilation mode, and the intensity of air exchange gradually weakens. The change characteristics of the isotope data outside the cave and at the Yemingzhu monitoring site con rm the aforementioned ventilation modes.
Due to the in uence of tourists and the structure of the tunnel, air changes inside and outside the cave.
Over the course of a season, there may be multiple ventilation modes. Tourist respiration can cause the CO 2 concentration to increase in the cave as a result of the cumulative effect of CO 2 and the "piston effect", which is caused by the movement of tourists. These in uences may cause a change in the ventilation mode in the cave. Further, climate changes outside the cave will cause changes in the cave's air movement, disrupting the exchange of air ow inside and outside the cave, and causing a change in the ventilation mode. If the tunnel is spacious, the ventilation conditions are good, allowing heat and CO 2 to be diffused with time, which barely impacts the ventilation mode. If the tunnel is narrow and the cavity is small, the CO 2 and heat in the cave easily accumulate, changing the virtual temperature, and thereby affecting the ventilation mode in the cave. Thus, our study suggests that more consideration should be given to changes in external climate or weather conditions when studying the conversion mechanism of karst cave ventilation modes.