Geysers Valley CO 2 Cycling Geological Engine (Kamchatka, Russia)

1941–2017 period of the Valley of Geysers monitoring (Kamchatka, Kronotsky Reserve) reveals a very dynamic geyser behavior under natural state conditions: significant changes of IBE (interval between eruptions) and power of eruptions, chloride and other chemicalcomponents,andpreeruptionbottomtemperature.Nevertheless,thetotaldeepthermalwaterdischargeremainsrelativelystable;thusallofthechangesarecausedbyredistributionofthethermaldischargeduetogiantlandslideofJune3,2007,mudflowofJan.3,2014,andothereventsofgeothermalcaprockerosionandwaterinjectionintothegeothermalreservoir.Insomecases,waterchemistryandisotopedatapointtolocalmeteoricwaterinfluxintothegeothermalreservoirandgeysersconduits.TOUGHREACTV.3modelingofVelikangeyserchemicalhistoryconfirms20%dilutionofdeeprechargewaterandCO 2 components after 2014. Temperature logging in geysers Velikan (1994, 2007, 2015, 2016, and 2017) and Bolshoy (2015, 2016, and 2017) conduits shows preeruption temperatures below boiling at corresponding hydrostatic pressure, which means partial pressure of CO 2 creates gas-lift upflow conditions in geyser conduits. Velikan geyser IBE history explained in terms of gradual CO 2 recharge decline (1941–2013), followed by CO 2 recharge significant dilution after the mudflow of Jan. 3, 2014, also reshaped geyser conduit and diminished its power.


Introduction
Geysers Valley is a unique site in Kamchatka where magnificent hydrothermal features are expressed in the form of numerous geysers, boiling springs, and mudpots with the total rate of ∼250-300 kg/s of chloride thermal waters discharged in the Geysernaya river, mostly within the area of 1.0 km to 0.2 km along the Geysernaya river downstream basin. Discovered by T. Ustinova in 1941, this "kingdom of geysers" attracted a number of studies, which focused their work on different aspects of geysers functionality, geological setting, recharge/discharge hydrogeological conditions, heat sources, and geochemistry of hydrothermal system as a whole [1][2][3][4][5][6][7][8]. One of the significant results of these studies is the conclusion that cycling CO 2 recharge is the main driver of Velikan geyser activity [1]. A comprehensive review of geysers phenomena can be also found in the recent paper of Hurwitz and Manga [9].
In spite of a relatively calm period of 1941-2007, when geysers activity changed gradually, two catastrophic events (landslide on June 3, 2007, and clastic mudflow on January 3, 2014) significantly reordered discharge conditions ( Figure 1). A number of important geysers were buried by clastic rocks (Pervenets, Troynoy) or sank in Podprudnoe Lake (Maly, Bolshoy, and Conus) after landslide on June 3, 2007. While some of them were lucky to reappear again (Bolshoy, Pervenets), the next disaster mudflow on Jan. 3, 2014, severely damaged Velikan geyser (this was the most impressive one in Geysers Valley). Velikan geyser conduit was completely filled by clastic rocks and although it released significant part of them by 2016, geysers functionality was not recovered in a full. This mudflow also created Podprudnoe Lake 2 upstream of the Geysernaya river, which might additionally recharge cold water in geyser hydrothermal system. It is worth noting that this natural story is much more dynamic as compared to industrial exploitation histories of ; 3: basalt, andesite, and dacite lavas and pyroclastics ( Q 3 1-2 ); 4: low permeability units of caldera lake deposits (Q 3 4 ), which are complicated by a dyke complex (Q 3 3 ust); 5: assumed thermal fluid-conducting faults; 6: Uzon-Geysernaya caldera boundary; 7: uplifted area that is associated with the contours of the active magma reservoir [10]; 8: geysers and hot springs (for numeration, see Table 6 in [1]); 9: Podprudnoe Lake and Podprudnoe Lake 2 dumb by mudflows; 10: catastrophic landslide-mudflow on 3.06.2007; 11: landslide-mudflow on 3.01.2014; 12: Geysernaya river flow rate measurement points: a: Podprudnoe Lake exit; b: Geysernaya river mouth. Grid scale: 500 m. AB: grey dotted line of cross section shown in Figure 19.
the Pauzhetsky (200-250 kg/s since 1966) and Mutnovsky (450-500 kg/s since 1999) geothermal fields in Kamchatka, where thermal losses accounted for two small geysers and one hot spring.
This paper aims to analyze cycling (IBE, interval between eruptions), chemistry, and available geysers conduit temperature logging data during the historical period of 1941-2017 to understand issues, which caused geysers functionality change.

Geological and Structural Setting of the Study Area
2.1. Geological Setting. In this section we followed the description of Kiryukhin, 2016. The age of the Uzon-Geysernaya caldera ( Figure 1) was estimated to be 39,600 ± 1,000 years according to the radiocarbon dating of soil samples below the caldera-forming ignimbrites [6]. Uzon-Geysernaya precaldera deposits comprise dacite-andesite tuffs and lavas that are 40,000-140,000 years old ( Q 3 1-2 , Q 3 3 , and Q 3 3 ust). Initially, this caldera was an isolated hydrological basin, where volcanogenic and sedimentary lake deposits were formed (Q 3 4 ). These deposits, which have thicknesses up to 400 m near the caldera rim, are represented by layered pumice tuffs and minor breccias and conglomerates. Caldera lake deposits are overlain by 15,000-20,000-year-old rhyolitedacite lavas, which formed large domes and adjacent lava flows up to 100-150 m thick ( Q 3 4 and Q 3 4 ). Approximately 9,000 to 12,000 years ago, the southeastern wall of the caldera was eroded by the Shumnaya and Table 1: Principal production zones, Lower Geysers (1) and Upper Geysers (2), of the geysers geothermal field are defined as 2D clusters of geysers discharge zones. Note. The total number of geysers is 51; , , and are coordinates of the clusters centers. , and caldera rim dacite and rhyolite extrusions ( Q 3 3 ); 5: aquifer of basalts, andesite, dacite lavas, and pyroclastics ( Q 3 1-2 ); 6: aquifer ( Q 1-2 ) basalt lavas; 7: aquifer of Pliocene tuffs, basalts, and sandstones; and 8: basement that is composed of tertiary sedimentary basins ( Figure 1). Figure 2 shows cold water discharge at the contact of relatively low permeability units of caldera lake deposits (unit 2, below) and permeable rhyolite-dacite extrusions of Geysernaya Mt (unit 3, above) (Lavovy creek, southern slope of Geysernaya Mt).
Geyser production reservoirs occur in shallow layer-type structures: Lower Geysers reservoir of 0.22 km 2 is inclined to NWW with a dip angle of 7 ∘ ; Upper Geysers reservoir of 0.18 km 2 is inclined to the south with a dip angle of 7 ∘ (for details see Section 2.3).

Geysers Reservoir Production
Properties. This study applies the program Frac-Digger 2 (Russian reg. #2017618050) in order to define discreet plane-oriented clusters of production feed zones (assumed to be points, where geysers appear on the surface) and the corresponding plane parameters (dip angle, dip azimuth, and fracture planar area), by using the top elevations of geysers (the total number is 51 ( Table  6, [1])) as input data for this analysis. The method of 2D plane parameter estimation in Frac-Digger 2 is the same as that in Frac-Digger (see [13] for details of Frac-Digger), but the difference is in the cluster selection algorithm. The stochastic Monte-Carlo approach is used in Frac-Digger 2 to define the largest (the number of production zones included) plane-oriented clusters. Several assumptions were also made as follows: (1) the maximum distance between the feed zone and the approximation plane is less than 10 m. (2) The maximum horizontal distance between feed zones in a 2D cluster is less than 1 km. Table 1 shows the 2D production zone parameters defined suchwise.
HOBO U12-015 temperature loggers were used to measure the interval between the eruptions (IBE) of the Velikan and Bolshoy geysers starting in July 2007. The loggers, which were installed in the channels of water discharge out of geysers, recorded the temperature of water outflow every 5 min. The eruption time of the geysers was estimated according to the time of the absolute maximum temperature prior to its absolute minimum (in a cycle). The same temperature loggers attached to iron tubes were used for temperature logging in geyser conduits.
The samples collected from geyser conduits were analyzed in Central Chemical Lab Institute of Volcanology and Seismology (for Na, K, Ca, and Mg analysis spectrometer SOLAAR M and for SiO 2 and NH 4 spectrophotometer UVmini-2140 were used; HCO 3 and CO 3 were analyzed using titration, standard pH meter "0YK]Y," for pH determination). Geyser water isotope content ( D, 18 O) measurements were performed with a LGR IWA 35EP analyzer of water isotope composition in IVS FEB RAS since 2015.
Flow and chloride rates in Geysernaya river mouth measurements were conducted using standard hydrometric methods with water sampling for chloride determination. Since 2015 a Cl-tracer method for river flowrate determinations was also applied; for this purpose logger HOBO U24-001 (range 0-10,000 S/cm, the set recording interval of 10 s) was used. Since 2017 a portable Mainstream 400P flowmeter has been used too. Since 1969 a water level instrument "Valdai" was put in the downstream channel from Velikan geyser conduit [3]; thus geyser eruptions were fairly recorded by water level rise. While these measurements were made regularly in the summer only, significantly more IBE data were obtained until 1990. Next monitoring system was designed by V. A. Droznin and was based on electric circuit switch, on/off, depending on geyser discharge conditions. This system successfully operated during 1993-2003 time period, mostly in the summer. The last monitoring . River water elevation coupled with erosion of river bottom (that served as a silica caprock for geysers reservoir) may cause cold water breakthrough into production geysers reservoir.
Note also a small distance between river and Velikan geyser that is  1947  1949  1951  1953  1955  1957  1959  1961  1963  1965  1967  1969  1971  1973  1975  1977  1979  1981  1983  1985  1987  1989  1991  1993  1995  1997  1999  2001  2003  2005  2007  2009   it reached ∼5.3 m. In 2016 a merely stable IBE of ∼40 min was achieved, but the style of Velikan geyser eruptions completely changed from piston-type eruption to long-term chaotic boiling with rare 2-2.5 m fountain events followed by a water level drop-down. Geyser (28, Figures 1 and 9). The methods of IBE monitoring of Bolshoy geyser were the same as for Velikan geyser, as described above. There is no clear trend in Bolshoy geyser IBE plot (Figure 8(b)); it ranges from 60 to 140 min in the time period of 1967-1989, later on in a more narrow range from 85 to 115 min during 1991-2007. Landslide on June 3, 2007, created Podprudnoe Lake, which put Bolshoy under water for a few months, but in Sept. 2007 Bolshoy geyser appeared on the surface and started to cycle again. Bolshoy geyser functionality strictly depended on the water level in Podprudnoe Lake: in flood time the geyser disappears below water and its conduit was used as cold water injector, while in low water seasons Bolshoy was regularly cycling with IBE range of 50-70 min (that is 1.5 times shorter, as compared to IBE before June 3, 2007).

Bolshoy
Mudflow on Jan. 3, 2014, brought some mud in Bolshoy geyser conduit, but the geyser was recovered to previous IBE soon. In recent years a tendency of IBE splitting with some "missed" (or less powerful) eruptions was observed (Figure 8(c)).   Table 2 (a) Chemical composition (in ppm) of hot springs and geysers in 2015. The samples collected by A. V. Kiryukhin directly from geyser conduits were analyzed in Central Chemical Laboratory of Institute of Volcanology and Seismology. Geyser and spring numbers (##) correspond to

Temperature Measurements in Velikan and Bolshoy Geysers Conduits
In Aug. 1994 temperature measurements in the conduit of Velikan geyser were performed by V. A. Droznin by using an original device, which was made in the Institute    of Volcanology and included thermistor VVf-3, controller KR1006VI1, memory chips 537RU10, and motherboard of ser. # 561. Accuracy of measurements of this device was anticipated at 0.5 ∘ C. Maximum depth of measurements in Velikan geyser conduit was 5 m [11]. Temperature records (1 record in 30 s) during one full cycle of geyser activity were obtained at that depth ( Figure 12(a)).
Since 2007, temperature measurements in geysers conduits were performed by using temperature loggers HOBO U12-015. In the case of geyser Velikan two T-loggers were   (Figure 12(b)). Maximum temperature before eruption was 105.5 ∘ C that is significantly lower (−4.5 ∘ C) than saturation temperatures at the given depth. Thus, saturation pressures at recorded preeruption temperatures are lower than hydrostatic pressures, and the difference is attributed to partial CO 2 pressures [1], which is estimated at 0.14 bar (Table 3). It is also worth noting that "preplay" events (10-12) before terminal eruptions, which expressed as intermediate boiling in geyser conduit, are synchronized with cyclic 1 ∘ C temperature drops ( CO2 positive changes up to 0.04 bar) (cycle 6 example in Figure 12 (b)). These preplay events (or small eruptions) of cycling activity were characterized at this time by IBE2 = 21 min.

Thermal-Hydrodynamic-Chemical Modeling of Velikan Geyser
TOUGHREACT-ECO2N [15,16] and TOUGH2-EOS1 + tracer [17]  (2) time dependent CO 2 mass flow, 0.010 kg/s during 1941-2013, and then drop to 0.008 kg/s. The river water recharge source was assigned to be active since 2014 with mass flow rate of 0.2 kg/s and an enthalpy of 457 kJ/kg or 109 ∘ C (heated to reservoir temperature). Velikan geyser discharge was assigned in the model in terms well on deliverability (PI = 10 −9 m 3 , Pb = 1.39 bar) which corresponds to hydrostatic pressure bottom hole conditions in a geyser conduit; this approximation seems to be reasonable since 80% of Velikan geyser time cycle was spent in self discharge conditions   before terminal eruption (Figures 12(a) and 12(b)). Rock porosity 0.5 was assigned to activate water/rock interaction and modeling time 2000 years preceding the time interval of interest  was assigned in this model too.
Chemical properties of recharged fluids were assigned according to chemical analysis of samples taken from Velikan geyser (representing deep water) before 2007 and samples taken from Geysernaya river (representing river water) ( Table 5). The only corrections were applied to pH values in deep water (we used in the model pH = 7, instead of 8.4 measured in a sample, to account for degassing of samples) and in river water (we used in the model pH = 11.3, instead of 8.4 measured in a sample, since we envision a two-step process whereby river water having a pH of 8.4 reacts with volcanic glass to form secondary minerals before entering the geyser, causing the pH to rise to 11.3). Note that this possibility of pH rise in river water was confirmed using TOUGHREACT-ECO2N one-element lumped parameter modeling too.
The initial mineral composition was assigned as glass 3, and possible secondary minerals include calcite, gypsum, quartz, sio2(am), cristobalite-a, and mordenite (we also used a kaolinite for testing river water pH rise). Calcite and gypsum were assigned at equilibrium, while other minerals were assigned using kinetic rate law conditions. Figure 15 shows the results of TOUGHREACT-ECO2N runs and comparison with Velikan geyser observational Cl, HCO 3 , and preeruption bottom temperature data. The model reasonably matches observational data; thus it explains deep water component (Cl) decline and HCO 3 disappearing after 2014 (Figures 15(a) and 15 (b)) as a result of high pH river water influx, following geyser bottom hole temperature rise (Figure 15(c)) due to partial pressure of CO 2 drop. Of note, we used a TOUGH2-EOS1 + tracer model to get a better match between modeling and observations for chloride Cl; in this case Cl component was assigned in the deep water source as a time dependent linear decreasing function with chloride drop rate of 1.7 ppm/year during 1969-2014 and then 5.6 ppm/year after 2014 (Figure 15(a)).  35EP analyzer of water isotope composition in IVS FEB RAS (for data of 2010 see [18]; for data of 2011-2014 see [19]) which is close to local meteoric water, and there is no significant difference in water isotope composition along the Geysernaya river up to Podprudnoe Lake 2. Hence, D and 18 O data is difficult to use to identify which part of Geysernaya river valley acts as injection area into geothermal reservoir (Figure 7) or if this water came from Podprudnoe Lake 2 (created in 2014 after mudflow dam Geysernaya river 2.5 km upstream) ( Figure 16). (iii) Giggenbach geoindicators tool [12] clearly shows events of immaturation of thermal waters of Velikan geyser after 2010 and Bolshoy geyser after 2007, which possibly reflects the influx of river waters into geysers conduits ( Figure 17).

Discussion/Conclusions
(iv) There is no sign of the total deep water component discharge decrease during 1961-2017 ( Figure 18). That means chloride and CO 2 (deep magmatic components) changes are attributed mostly to mixing processes (due to river/meteoric waters influx into geothermal reservoir) and redistribution of discharge due to the change in surface conditions (giant landslide 2007 shapes, Podprudnoe Lake 1, Geysernaya river and Podprudnoe Lake 2 cold water injections events, selfsealing processes, etc.).
(v) Temperature logging in geysers Velikan (1994Velikan ( , 2007Velikan ( , 2015Velikan ( , 2016Velikan ( , and 2017 and Bolshoy (2015, 2016, and 2017) conduits shows preeruption conditions under temperatures below boiling at corresponding hydrostatic pressure that means partial pressure of CO 2 lowers boiling temperatures (vi) Velikan geyser IBE history 1941-2007 with an almost linear trend of IBE increase (+200 min per 65 years or +3 min/year, Figure 8(a)) and with almost unchanged conduit and shallow pool geometry conditions may be interpreted as a result of gradual CO 2 recharge decline, which should follow chloride decline (Figure 10(a)), as both are a deep magmatic components. Some shortage of IBE during 2007-2013 is explained by recharge rate increase due to reservoir pressure build-up after Podprudnoe Lake 1 entrance (caused by giant landslide of June 3, 2007). Once Velikan geyser conduit eroded on the top and was partially buried (as a result of Jan. 3, 2014, mudflow) and CO 2 recharge was by 20% diluted, then the power of its eruption significantly decreased. The open question is why its CO 2 influx cycles frequency increased twofold from ∼20 min to ∼40 min (cf. Figure 12 (b) and Figure 12(c)).
(vii) 1941-2016 period of monitoring in the Valley of Geysers (Kamchatka, Kronotsky Reserve) reveals a very dynamic geysers behavior in natural state conditions, in which a deep upflow CO 2 recharge plays a major role in geysers sustainability. Velikan geyser and Bolshoy geyser examples show that cycling CO 2 flux triggered gas-lifted geysers eruptions, while the power of eruptions is defined by partial CO 2 pressure. Deep magmatic CO 2 flux redistribution in geysers geothermal reservoir and local meteoric inflows were found to be crucial for geysers functionality (Figure 19). The next page in the geysers history significantly depends on future thermal hydrodynamic impact of the Podprudnoe Lake 2, created upstream of the Geysernaya river on Jan. 3, 2014.   Table 6 in [1]).