Salty Matters

The Blog is written by me, John Warren. Once every three or four weeks or so I will post an article or two on an evaporite topic that has piqued my interest. On the Saltwork Publications webpage (under "the Works") there is a growing library of pdfs and epubs based on these blogs. These articles on the website have much higher resolution extractable graphics in than in the blog. There is also a link to this set of pdfs and epubs on the home page (

Salt Dissolution (4 of 5): Anthropogenically-enhanced geohazards

John Warren - Thursday, November 30, 2017



As we saw in the previous article the dissolution and collapse of nearsurface and at-surface salt is a natural and ongoing process. When salt bodies experiencing natural dissolution and alteration are penetrated by drilling or parts of the salt mass are extracted in a poorly supervised fashion, the resulting disturbance can speed up natural solution and collapse, sometimes with unexpected environmental consequences (Figure 1; Table 1). To minimise the likelihood of unexpected environmental consequences tied to enhanced rates of salt dissolution in the vicinity of engineered structures the same rule applies as is applied to safe salt mining, namely "Stay in the salt" (Warren, 2016) In this context, here is a quote from Zuber et al. (2000) in a paper dealing with flooding and collapse events in Polish salt mines.

“Catastrophic inflows to salt mines, though quite frequent, are seldom described in the literature, and consequently students of mining and mine managers remain, to a high degree, ignorant in this respect. Contrary to common opinion, inflows are seldom caused by unavoidable forces of nature. Though some errors were unavoidable in the past, modern geophysical methods are, most probably, quite sufficient to solve the majority of problems (e.g., to determine a close presence of the salt boundary). Detailed study of the recent catastrophic floods, which happened in Polish salt mines, shows that they usually occur, or have strong negative impacts, due to human errors. Most probably similar human errors caused catastrophic inflows to salt mines in other countries. It seems that a knowledge of the real history of catastrophes, better education of mine engineers and the application of modern geophysical methods could lead to the reduction of floods in salt mines.”

This article, the 4th of 5 in Salty Matters, focuses on the anthropogenic enhancement of salt dissolution, however, it discusses only a few of the many documented examples of problems that result from enhanced dissolution brought about by human activity. For a more comprehensive documentation of relevant case histories and an expanded discussion that includes mine collapse, brinefield subsidence and collapse and industrial accidents associated with salt cavity storage, the interested reader should refer to Chapters 7 and 13 in Warren (2016).

As wells as mine floods and mine losses, poorly monitored then abandoned brinefield and solution cavities can be areas with major environmental problems especially where old extraction wells are undocumented and unmonitored. Like areas of natural solution collapse; they can become zones of catastrophic ground failure. This is especially problematic if located near cities or towns (Tables 1, 2). Some of the most outstanding examples of how not to solution mine a resource and how not to control ground subsidence effects are to be found in east European countries that are still trying to deal with the environmental outcomes of being former satellite states of the Soviet Union. But caving problems have been tied also to mines and boreholes in the United States and Canada, where once again human error, greed or ignorance has created many of these problem structures (Table 1 and Figure 1).

Ocnele Mari Brinefield, Romania

SALROM, a government-owned company, solution mines Badenian (Miocene) halite in the Valcea Prefecture of Romania (Figure 2). Production from Field 2 was shut down on March 5, 1991, after significant earth vibrations were noted by SALROM workers. A subsequent sonar survey showed that poorly monitored brinefield leaching between 1971 and 1991 had created a gigantic merged cavern as salt pillars separating adjoining caverns were inadvertently dissolved (Figure 2); the upper parts of the captive boreholes 363, 364, 365, 366, 367 and 369 had merged into a common cavern. The cavern was filled with some 4.5 million cubic metres of brine, was less than a hundred metres below the landsurface, was more than 350 metres across and was overlain by loosely consolidated sandy marls (von Tryller, 2002; Zamfirescu et al., 2003). The cavern was overlain by a bowl of subsidence and, even though mining operations in the region of the cavern completely stopped in 1993, the ground above continued to subside to a maximum of 2.2 metres prior to the September 2001 collapse. In the period 1993 - 2001 the cavern continued to enlarge and the northern part of the cavity expanded by some 25-35 m (see inset in Figure 3 showing sonar surveys 1995 -2001).

The 1993 sonic survey of the cavern showed it was so large and shallow that its roof must ultimately collapse, predicting when was the unknown. The fear was that if it collapsed catastrophically, it could release a flood of at least a million cubic metres of brine. Brine could be released either over several hours (least damaging scenario) or instantly forming a wave of escaping water several metres high that would flood the nearby Sarat River valley for many kilometres downstream. There were 22 homes on top of the cavern that would be immediately affected; also at risk were the hundreds and perhaps thousands of residents in the area of the saltworks and river valley below, as well as the local ecology and the civil/industrial infrastructure. Ongoing brine delivery from other nearby operational SALROM caverns to the large Oltchem and Govora chemical plants would also be disrupted. When collapse did ultimately occur (twice in the period 2001 - 2004), each episode took place over a number of hours and so a catastrophic wave of water did not eventuate.

According to Solution Mining Research Institute, there was no good engineering solution to prevent the Ocnele cavern roof from collapsing. Prior to collapse, brine in the cavern exerted pressure against the roof helping to hold it up; removing even a small amount of brine would remove hydraulic pressure that could possibly trigger a catastrophic collapse. Nor was it practical to fill the cavern with sand and industrial wastes as suggested by the Romanian government. It would take too long and it was unsafe to place men and equipment on top of the still expanding cavern. Even if it could be done, only the areas directly below the injection wells would be filled. The best solution was to construct a dam as close to the cavern as possible, and then perhaps trigger a controlled collapse by pumping out the cavern brine.

A partial collapse of the cavern roof atop well 377 occurred on 12 September of 2001, the ensuing brine flood killed a child and injured an older man. The collapse began at 7 pm with brine spilling out of wells 365 and 367. Southward of well 377 a collapse cone some 10 metres across started to form and fill with brine. The cone continued to expand and fill with water until its southern lip was breached. Water spilled out of the cone and down the hill slope. The flow rate of the expelled flow reached a maximum of 17 m3/sec at 3 am on September 13th, with flow continuously exceeding 10 m3/sec for a period of more than 6 hours around that time. By 7 am flow was down to 4-5 m3/sec and by 7 pm, 24 hours after the flow began the flow rate was 0.4-0.5 m3/sec. Some 24 hours after the onset, roof collapse had formed a water-filled lake with an area of 2.4 ha (Figures 2, 3; Zamfirescu et al., 2003).

A second roof collapse occurred on July 13, 2004. Realitatea Romaneasca, Romania, reported that ground collapse occurred at 9:30 pm July 13, 2004, possibly triggered by heavy, recent rains. The wellhead of bore 365 was destroyed in this collapse, and many other well heads in Field 2 had been destroyed by ongoing subsidence in the period 2001-2004. This collapse was the culmination of a series of collapse-related events that began on Monday July 11. It was planned that a purpose-built earthen dam would contain the brine flood. But when the roof fall occurred at 9:30 pm, the dam was breached 30 minutes later by some 250,000 m3 of salt water. Flow rate through the breach reached a maximum of 6 m3/sec. Collapse did not occur until just after local authorities had evacuated people from 50 homes in the area around the cavern. Most of the brine forced out by the roof collapse escaped into the Sarat (Olt) River. The government attempted to dilute the effects of this flush of salt water by releasing fresh water into the river from nearby dams. Fortunately, unlike the 2001 event, there was no loss of life.

Induced collapse, Gellenoncourt saltworks, France

Gellenoncourt is one of three sites that exploit halite beds of the Lower Keuper in the eastern part of the Paris Basin. The deposits extend from Cezanne in the west to Nancy with a length of 250 km and a width of 50 to 70 km. Salt production utilising solution mining techniques is focused on three sites along the Meurthe river, which all lie immediately to the east of Nancy.

On March 4, 1998, a sinkhole more than 50 m across and 40 m deep formed atop the SG4 and SG5 brinefield caverns in the Gellenoncourt saltworks near Lorraine, France (Figure 4; Buffet, 1998). It was an induced collapse designed to prevent a possible uncontrolled future ground collapse. The problem started in 1967 with the beginning of the exploitation of Triassic salt layers in the Keuper Fm. In total the Keuper is more than 150 m thick with five salt layers at its base, passing up into variegated and poorly consolidated marls and sandstones and capped by the Dolomite du Beaumont. The top of the salt is some 220 m below the surface and divided into five beds numbered 1 to 5 from top to bottom, with the solution mining program designed to leach beds 1 through 3 in the region of SG4 and SG5 caverns (Figure 5). Five wells, SG1 to SG5, were joined using hydrofracturing in February 1967. Theoretically, the process was designed to leave a substantial salt pillow atop all the cavities and so separate the solution cavities from the overlying marls (Figure 5; 1967-1971). The SG4 and SG5 caverns unexpectedly joined in 1971. Brine injection to these two wells was stopped, but crossflowing brines flowing to the producing SG1 well continued to excavate these two caverns. By 1982 the salt cushion in the roof of the SG4-SG5 cavern had completely dissolved away, placing the cavity roof in direct contact with the base of the marls.

From 1982 until October 1992 there was no further upward growth of the cavern roof. Then a 25 m-thick section of the variegated marls in the roof broke free and fell to the cavern floor, leaving a large section of the cavity roof in direct contact with the brittle Dolomite de Beaumont. This stiff dolomite layer prevented any further immediate collapse of the roof and consequent propagation of the cavity to the surface. But continued growth of the roof span beneath the dolomite would mean a later, larger, perhaps catastrophic collapse. In 1995 the operator tried to induce a controlled collapse by placing a submerged pump in the cavity and pumping out brine to create an exposed upper face. But this didn’t work. The next approach was to further enlarge the roof span by injecting 300,000 m3 of freshwater into the cavity. Collapse occurred on March 4, 1998, forming a 50 m wide crater. To protect the surrounding countryside from any brineflood damage a dam was constructed to capture any brine overflow, but in the actual event it was not needed.

Retsof Mine, New York State, USA

The Retsof Mine collapse is perhaps one of the best documented examples of an operational mine being lost to dissolution features and consequent flooding, yet even here the exact causes of the mine loss are still argued. Was it because of the intersection of the expanding mine workings with a natural water-saturated salt dissolution and fracture system, or the insection of the mine workings with brine filled cavities formed by wild brining operations in the 1800s, or was it a result of mine roof instabilities related to a change in room and pillar sizes?

At the time it was operational, the 24 km2 area of subsurface workings in the AKZO-Retsof salt mine made it the largest underground salt mine in the USA, and the second largest salt mine in the world (Figure 6, 7).

Operational history

Retsof Mine started operating in 1885 after completion of the 3.7× 4.9-meter wide, 303.5-meter-deep Shaft #1. The mine claimed an initial 5,460-metric-tons per day hoist capacity (Goodman et al., 2009). Early main haulage ways were driven east and west while production headings were driven north (updip) for salt-tramming ease. Room heights in the 6-meter-thick salt bed were 2 to 4 meters, with salt left in both the floor and roof. Four-meter-high rooms were worked in two benches. Rooms were 9.2 meters wide and separated by 9.2-meter pillars. At that time there was no timbering, the mine was dry, mine air temperature was 17°C, and the mine was largely gas-free.

By 1958, the Retsof Mine was connected to the former Sterling Mine for ventilation and emergency escape purposes [Figure 6a; Gowan et al., 1999]. By the late 1960s, the mine had advanced beneath the modern Genesee River and Valley (Figure 6b). By the early 1970s, the Retsof Mine operators had installed an underground surge bin, fed by a new conveyor system, and the old rail-haulage system was eliminated. Mainline conveyors led to yard or panel belts feeding each mining section, where a Stamler feeder-breaker crushed salt delivered by diesel-powered Joy shuttle cars. In the early 1980s, the shuttle cars were gradually replaced by load-haul-dump vehicles (LHDs). In 1969, Netherlands-based Akzo Corporation acquired International Salt Company and operated the Retsof mine until its abandonment due to flooding in 1995.

During April 1975, an explosion occurred in the original Sterling B Shaft during efforts to control water inflow into the Retsof Mine from this abandoned and partially collapsed shaft (Goodman et al. 2009). The leaky B Shaft had not been used or maintained for years. By 1975, International Salt was concerned that freshwater inflow from the B Shaft could pose a salt dissolution, collapse, and flooding risk to the then-connected Retsof Mine. Removal of a partial shaft blockage of timber and rock debris was attempted as a means of regaining airflow needed to safely access, rehabilitate, and grout off the water inflow to the shaft and mine below. A maintenance crew attempted to dislodge the shaft obstruction by pushing a large boulder into the shaft that was to drop down and knock through the debris. A methane explosion occurred upon impact. The upward force of the explosion killed four people on the surface near the shaft collar and injured others. On November 19, 1990, a roof fall resulted in two fatalities. Deformation and fracture of roof salt can occur because of a concentration of stresses; i.e., punching of the roof by stiff pillars. After the fatalities, the mine tested smaller, yielding pillars to alleviate roof falls (Figure 6b). Positive test results led to the adoption of a yield-pillar design.

The Retsof mine was lost to water flooding in 1994-1995.Before abandonment, the mine had been in operation since 1885, exploiting the Silurian Salina Salt and prior to shut down was producing a little over 3 million tons of halite each year. At that time it supplied more than 50% of the total volume of salt used to de-ice roads across the United States.

Geology and hydrology in the vicinity of Retsof Mine

The Genesee Valley sediments preserve evidence of several complex geologic processes that include; (1) tectonic uplift of Palaeozoic sedimentary rocks and subsequent fluvial down cutting, (2) waxing and waning glacial events that drove erosion of bedrock and the subsequent deposition of as much as 750ft of glacial sediments; and (3) ongoing erosion and deposition by postglacial streams (Figure 7a; Yager, 2001; Young and Burr, 2006). The Genesee Valley spans through western New York north to south from Avon, NY to Dansville, NY, including the Canaseraga Creek up through its mergence with Genesee River. A detailed section from Palaeozoic rocks and younger have been recorded in the Genesee River Valley (Figure 7a, b); however, detailed analysis of glacial sediments and till are still somewhat scarce. The B6 salt bed (Retsof Bed) of the Vernon Formation was the salt unit extracted at the Retsof Mine (Figures 7b). Several other salt layers exist in the Salina Group both above and below the B6. These salt layers include two horizons in Unit D at the base of the Syracuse Formation approximately 50 m (160 feet) above the B6 salt level.

Quaternary-age sediment in the Genesee Valley consists mostly of unconsolidated glacial sediments ranging up to 750 feet thick. These sediments encompass gravel, sand, silt and clays that were deposited mostly during the middle and late Wisconsin deglaciation and filled the lower parts of the pre-existing glacial scour valley. End moraines consisting of glacial debris were deposited in lobes to the south of the slowly retreating glacier. As the glacier had scoured through the valley, carving out bedrock and accumulating sediment, steep-sided valley walls were cut and pro-glacial lakes formed. The glacial lake sediments are dominated by muds, but also include large boulders and cobbles carried to the lake depressions by glacial ice. Fluvial sediment from the Genesee River and Canaseraga Creek also drained into these glacial lakes. A final pro-glacial lake formed as the Fowlerville end moraine was deposited. The Fowlerville end moraine extends approximately 4.5 to 8 miles north of the Retsof collapse site (Figure 7a). The various glacial lakes and moraines disrupted the normal flow fluvial patterns of most local drainages and creeks in the valley. Alluvium is the uppermost layer of the surface and is variable in thickness throughout the valley, but normally ranges about fifty feet thick and is still being deposited across the Genesee River Valley floodplain (Yager, 2001).

The aquifer system is hosted within the glacial valley-fill and consists of three main aquifers separated by two confining layers. It is underlain by water-bearing zones in fractured Palaeozoic bedrock (Yager, 2001). The glacial aquifers are bounded laterally by the bedrock valley walls. The uppermost aquifer consists of alluvial sediments 20 to 60 ft thick (unit 1 in Figure 7b); the middle aquifer consists of glaciofluvial sand and gravel less than 10 ft thick (unit 3 in Figure 7b); and the lower aquifer consists of glaciofluvial sand and gravel about 25 ft thick overlying the bedrock valley floor (unit 5 in Figure 7b). These aquifers are separated by aquitards dominated by muds and clays (Units 2 and 4 in Figure 7b).

The now abandoned Retsof Mine lies 550 to 600 ft below the eroded valley floor (Figure 7b). Hence, the upper and middle aquifers are separated by an upper confining layer of lacustrine sediments and till as much as 250 ft thick, and the two confined aquifers are separated by a lower confining layer of undifferentiated glaciolacustrine sediments as much as 250 ft thick. The principal water-bearing zone in the bedrock overlying the mine consists of fractured carbonates and sands near the contact between the Onondaga and Bertie Limestones. The fractured aquifer that occurs at this level in the stratigraphy supplied a significant volume of the water that ultimately flooded the Retsof Mine. The glacial aquifers are hydraulically connected at the edges of the confining layers and in subcrop zones, where water-bearing zones in the bedrock intersect a fractured and karstified bedrock surface.

Ground water within the valley generally flowed northward and updip before the mine collapse (Yager, 2001). The hydraulic head distribution in the confined aquifers under natural (precollapse) conditions is assumed to have been similar to that in the upper aquifer before the collapse, but water levels in the confined aquifers were probably above the water table beneath the valley floor. Much of the ground water reservoired along the fractured Onondaga/Bertie Limestone contact also flowed northward to escape at the Bertie Limestone subcrop, now located in the valley north of the Fowlerville Moraine (Figure 7c).

Water influx tied to changes in room and pillar mining?

In 1993, ceiling falls began to occur in rooms in the deepest part of the Retsof mine near its southern boundary (Figure 6a; Yager et al., 2009). In response, the mine owner, AkzoNobel Salt Incorporated (ANSI), turned to an innovative “yielding pillar” mining technique that utilised many narrow (20 feet × 20 feet) pillars rather than few wide ones in the mined section (Figure 6b). Geotechnical analyses indicated that the resulting configuration would allow the salt pillars to slowly yield and create a “stress envelope” in the surrounding bedrock to support the entire mined room.

Closure monitoring was conducted in the yield-pillar test panels and the two full-scale panels during mining to measure panel behaviour and to see if the new design mitigated the floor and roof problems being experienced in the large pillar area of the mine (van Sambeek et al., 2000). Monitoring initially indicated that room closure rates were slightly greater than expected, but had an overall character (trend) of steadily decreasing rates, which is consistent with stable conditions. This trend changed dramatically to a rapid and unstable closure rates in the final weeks leading up to the inflow. The change in trend was initially obscured by fluctuating closure rates because salt extraction was occurring between the two yield-pillar panels as the monitored abutment pillar was isolated. Whereas the closure rates were expected to decrease after this mining was complete, they did not; in fact, they increased. This change in panel character later interpreted to indicate that a pressure surcharge existed or developed over two of the yield-pillar panels prior to the in- flow (Gowan et al., 1999).


Loss of roof stability and flooding

In November 1993, strain measurements in a yielding-pillar area within the mine indicated a larger than expected deformation of salt near the eastern wall of room 2 Yard South (Figure 6a, b). Mining in the area was halted as ceiling falls of salt continued during the next four months. On March 12, 1994, a magnitude 3.6 seismic event, caused by a large roof collapse, was detected by seismometers more than 300 miles away. Mine workers attempted to enter room 2 Yard South but found it was blocked by a pile of rock rubble within the formerly mined room and that saline water entering via fractures in the mine roof. Over the next several weeks, Akzo made concerted attempts to save the mine by pumping water out and drilling around the collapse area to inject cement grout so as to stabilise the collapsed room and prevent a further inflow of water. Meanwhile, unstable shale layers overlying room 2 Yard South sagged and collapsed to form a 300-foot-diameter zone of rock rubble that slowly propagated upward through overlying layers of shale (Figures 6b, 8a, b). This column of rock rubble is referred to as a rubble chimney.

The propagating rubble chimney eventually reached a layer of carbonate (limestone) rock that was strong enough to temporarily resist further collapse, stopping further the rubble chimney’s upward progression. At this point, the flow of water into the mine stabilised at about 5,500 gallons per minute. Water entering the mine was saline and probably a mixture of saline water from the shale and a prominent fracture zone aquifer within the Onondaga and Bertie Limestones (Figures 7c). By the end of March 1994, tons of cement grout had been injected into the mine and the rubble chimney through nearly 30 boreholes drilled in the collapse area, but these efforts failed to stem the rate of water flowing into the mine and the inflow was becoming increasingly less saline.

On April 6, 1994, the limestone rock layer collapsed, and 550 feet of unconsolidated sediments in the Genesee River valley quickly slumped downward into the resulting cavity, forming a sinkhole at the land surface, more than 15 feet deep and several hundred feet across (Figure 8, 9a b). The collapse of the limestone rock was like pulling the plug in a bathtub—it allowed groundwater from a fresh-water aquifer at the base of the unconsolidated glacial sediments (the lower confined aquifer (Figure10) to drain downwards through the rubble chimney and into the mine. By mid-April, a second collapse occurred in an adjacent room (11 Yard West; Figure 6b). On May 25, a drilling crew working above room 11 Yard West felt tremors and removed their drill rig, and themselves, just before this second sinkhole formed at the land surface. This one had a surface expression that was more than 50 feet deep and several hundred feet across (Figures 6a, 8a, 9a,b). The discharge from the aquifer through both rubble chimneys increased the flow of water into the mine to about 18,000 gallons per minute.

Water began to fill the southern end of the mine and then spread steadily northward, dissolving the bases of the salt pillars that supported the mine ceiling (Figure 6a). As the pillars gave way, the southern part of the mine began to collapse, causing the land surface above it to subside. The greatest subsidence (more than 15 feet) was beneath the two sinkholes, which altered the channel of Beards Creek, allowing surface water to fill the sinkholes. The surface water did not flow downward to the mine, however, because hundreds of feet of fine-grained sediments underlie the Genesee River valley. The instability also forced the closure of the U.S. Route 20A bridge over Beards Creek; the southern end of the bridge eventually subsided by 11 feet (Figure 9c, d). The bed of the Genesee River 1 mile north of the collapse areas subsided by as much as 5 feet and altered the pattern of sediment scour and deposition along a 1.5-mile reach downstream of Beards Creek.

Events indicating loss of mine

The eventual loss of the Retsof Salt Mine occurred in stages, driven first by “out of salt” roof breaches, followed by ongoing salt dissolution of the water-encased salt pillars in the flooded mine. It began in the early morning hours of March 12, 1994, with a magnitude 3.6 earthquake. The quake was caused by the catastrophic breakdown of a small mine pillar and panel section some 340 meters below the surface and was accompanied by the surface collapse of an area atop the mine that was some 180 by 180 meters across and 10 meters deep. This all occurred at the southern end of the mine near the town of Cuylerville. A month later, on April 18, an adjacent mine room collapsed to form a second collapse crater (Figure 6a, b) The initial March 12 collapse in the mine was accompanied by an inrush of brine and gas (methane) and by a sustained intense inflow of water at rates in excess of 70 m3/min, via the overlying now fractured limestone back (Gowan and Trader, 2000).

In a little more than a month, the two steep-sided circular collapse features, some 100 meters apart, had indented the landscape above the two collapsed mine rooms (Figures 6, 8, 9). The northernmost collapse feature, which was more than 200 meters across, included a central area that was about 60 meters wide and had subsided about 6 to 10 meters. The southernmost feature, which was about 270 meters in diameter, included a central area that was about 200 meters wide and had subsided about 20 meters (Figure 6b). Fractures extending up from the broken mine back created hydraulic connections between aquifers, which previously had been isolated from each and so provided new high volume flow routes for rapid migration of perched groundwaters into the mine level.

Water flooded the mine at rates that eventually exceeded 60,000 litres per minute and could not be controlled by pumping or in-mine grouting. By January 1996 the entire mine was flooded. Associated aquifer drawdown caused inadequate water supply to a number of local wells in the months following the collapse; the fall in the water table as ground waters drained into the mine in effect meant some water wells went dry (Figure 8c; Tepper et al., 1997).

Aside from the loss of the mine and its effect on the local economy, other immediate adverse effects included abandonment of four homes, damage to other homes (some as much as 1.5 kilometers from the sinkholes), the loss of a major highway and bridge, loss of water wells and prohibition of public access to the collapse area (Figure 9). Land subsidence, possibly related to compaction induced by aquifer drainage to the mine, even occurred near the town of Mt. Morris some 3 miles south-west of the collapse area. Longer term adverse effects are mostly related to increasing salinization of the lower parts of the Genessee Valley aquifer system in the vicinity of the mine (Figure 10; Yager, 2013).


What caused the loss of the mine?

Post-mortem examination of closure data from the two failed mine panels has been interpreted as indicating an anomalous buildup of fluid pressure above the panels in the period leading up to their collapse (Gowan et al., 1999). The initial influx of brine and gas following the first collapse coincided with the relief of this excess pressure.

Gowan and Trader (1999) argued for the existence of pre-collapse pressurised brine cavities and gas pools above the panels and related them to nineteenth-century solution mining operations. They document widespread natural gas and brine pools within Unit D of the Syracuse Formation approximately 160 ft above the mined horizon in the Retsof Mine. The satellite image also shows that collapse occurred in a pre-existing landscape low that defined the position of Beard Creek valley above the mine (Figure 6a). Brine accumulations likely formed in natural sinks, long before salt solution mining began in the valley. Salt in the shallow subsurface dissolved naturally, driven by the natural circulation and accumulation of meteoric waters along vertical discontinuities, which connected zones of dissolving salt to overlying fresh water aquifers (see Warren, 2016, Chapter 7 for a detailed documentation of this salt related hydrology and geomorphology).

Gowan and Trader (2003) argued that daylighting sinkholes had formed by the down-dropping of the bedrock and glacial sediments into pre-existing voids created by the dissolution of salt and the slaking of salt-bearing shale upon exposure to fresh water. It is likely that the extent of these brine filled voids was exacerbated by the “wild-brining” activities of salt solution miners in the 1800’s.

Nieto and Young (1998) argue that the transition to the yield pillar design was a contributing factor to the loss of mine roof integrity. Loss of mechanical integrity in the roof facilitated fracturing and the influx of water from anthropogenic “wild brine” cavities. The exact cause of the loss of roof integrity and subsequent mine flooding is still not clear. What is clear is that once the Retsof mine workings passed out of the salt mass, and into the adjacent non-salt strata, the likelihood of mine flooding greatly increased.

Even so, the loss of the Retsof salt mine to flooding was a total surprise to the operators (Van Sambeek et al., 2000). The mine had operated for 109 years with relatively minor and manageable incidents of structural instability, water inflow, and gas occurrences. A substantial database of geological information was also collected throughout the history of the mine. It was this relatively uneventful mine history and the rich technical database that provided support for pre-inflow opinions by mine staff that there was no significant potential for collapse and inundation of the mine. The Retsof collapse took place in a salt-glacial scour stratigraphy and hydrology near identical to that in the Cayuga Mine region.


Patience Lake Potash Mine flood

In the 1970s the Patience Lake potash mine operation, located on the eastern outskirts of Saskatoon, Canada, encountered open fractures tied to a natural collapse structure and was ultimately converted to a successful solution mining operation (Figure 11). Grouting managed to control the inflow and mining continued. Then, in January of 1986, the rate of water inflow began to increase dramatically from the same fractured interval (Figure 12; Gendzwill and Martin 1996).

At its worst, the fractures associated with the structure and cutting across the bedded ore zones were leaking 75 m3/min (680,000 bbl/day) of water into the mine. The water was traced back to the overlying Cretaceous Mannville and possibly the Duperow formations. Finally, in January 1987 the mine was abandoned. It took another six months for the mine to fill with water. Subsequent seismic shot over the offending structure suggested that the actual collapse wasn’t even penetrated; the mine had merely intersected a fracture within a marginal zone of partial collapse (Gendzwill and Martin 1996).

Part of the problem was that the water was undersaturated and quickly weakened pillars and supports, so compromising the structural integrity of the workings. The unexpected intersection of one simple fracture system resulted in the loss of a billion dollar conventional potash mine. Patience Lake mine now operates as a cryogenic solution mine by pumping warm KCl-rich brine from the flooded mine workings to the surface. Harvesting of the ponds takes place during winter after cryogenic precipitation of sylvite in at-surface potash ponds (Fig. 11).


Unlike the Patience Lake Mine flood, there was a similar episode of water inflow in the nearby Rocanville Potash Mine. But there a combination of grouting and bulkhead emplacement in succeeded in sealing off the inflow, thus saving the mine (see Warren 2016 for detail). Unlike Patience Lake, the brine from the breached structure in Rocanville was halite-saturated, so limiting the amount of dissolution damage in the mine walls. Different outcomes between the loss of the Patience Lake Mine and recovery from unexpected flooding in the Rocanville Mine likely reflects the difference between intersecting a natural brine-filled dissolution chimney that had made its way to the Cretaceous landsurface and is now overlain by a wide-draining set of aquifer sediments, versus crossing a blind dissolution chimney in a saline Devonian sediment surround that never broke out at the Cretaceous landsurface. Understanding the nature of the potential hydrological drainages and water source is a significant factor in controlling unexpected water during any salt mine expansion.

Lake Peigneur, Louisiana

Lake Peigneur is a natural water-filled solution doline that overlies the dissolving crest of the Jefferson Island Salt Dome Figure 13). The most recently risen part (salt spine) of the Jefferson Island stock crest, just west of the town of New Iberia, Louisiana, is now 250 m (800ft) higher than the adjacent flat-topped salt mass, which is also overlain by a cap rock. The boundary shear zone separating the spine from the less active portion of the crest contains a finer-grained “shale-rich” anomalous salt zone that had been penetrated in places by the former Jefferson Island mine workings. The known salt anomaly (BSZ) defined a limit to the extent of salt mining in the diapir, which was focused on extracting the purer salt within the Jefferson Island spine, in a mining scenario much like the fault shear anomaly, as mapped by Balk (1953), defined the extent of the workings at nearby Avery Island. The spine and its boundary “shear” zone are reflected in the topography of the Jefferson Island landscape, with a natural sub-circular solution lake, Lake Peigneur, created by the dissolving shallow crest of the most recently-active salt spine.

On November 20, 1980, one of the most spectacular sinkhole events associated with oilwell drilling occurred atop the Jefferson Island dome just west of New Iberia. Lake Peigneur disappeared as it drained into an underlying salt mine cavern and a collapse sinkhole, some 0.91 km2 in area, developed in the SE portion of the lake (Figure 13; Autin, 2002). In the 12 hours following the first intersection the underlying mine had flooded, and the lake was completely drained.

Drainage and collapse of the lake began when a Texaco oil rig, drilling from a pontoon in the lake, breached an unused section of the salt mine some 1000 feet (350 metres) below the lake floor (Figure 14a). Witnesses working below ground described how a wave of water instantly filled an old sump in the mine measuring some 200 ft across and 24 feet deep. The volume of floodwater engulfing the mine corridors couldn’t be drained by the available pumps. At the time of flooding the mine had four working levels and one projected future level. The shallowest was at 800 feet, it was the first mined level and had been exploited since 1922. The deepest part of the mine at the time of flooding was the approach rampways for a planned 1800 foot level. In 58 years of mine life, some 23-28 million m3 of salt had been extracted. Prompt reaction to the initial flood wave by mine staff allowed all 50 personnel, who were underground at the time, to escape without anything more than a few minor injuries.

The rapid flush of lake water into the mine, probably augmented by the drainage of natural solution cavities in the caprock below the lake floor, meant landslides and mudflows developed along the perimeter of the sinkhole, and that the lake was enlarged by 28 ha. The surface entry hole in the floor of Lake Peigneur quickly grew into a half-mile-wide crater. Eyewitnesses all agreed that the lake drained like a giant unplugged bathtub—taking with it trees, two oil rigs (worth more than $5 million), eleven barges, a tug boat and most of the Live Oak botanical gardens. It almost took local fisherman Leonce Viator Jr. as well. He was out fishing with his nephew Timmy on his fourteen-foot aluminium boat when the disaster struck. The water drained from the lake so quickly that the boat got stuck in the mud and they were able to walk away! The drained lake didn’t stay dry for long, within two days it was refilled to its normal level by Gulf of Mexico waters flowing backward into the lake depression through a connecting bayou (Delcambre Canal, aka Carline Bayou). But, since parts of the lake bottom had slumped into the sinkhole during the collapse, the final water level in some sections was higher than before relative to previous land features. It left one former lakefront house aslant under 12 feet of water.

Of course, an anthropogenically induced disaster like this attracted the lawyers like flies to a dead dingo. On 21 November 1980, the day after the disaster, Diamond Crystal Salt filed a suit against Texaco for an unspecified amount of damage. On 25 November, Texaco filed a countersuit against Diamond Crystal. The Live Oak Gardens sued both Diamond Crystal and Texaco. Months later, the State of Louisiana was brought into the suit since the incident occurred on state land. One woman sued Texaco and Wilson Brothers (the drillers) for $1.45 million for injuries (bruised ribs and an injured back) received while escaping from the salt mine. Less than a week before the scheduled trial, an out-of-court settlement was reached between the major players. Due to human error, related to a triangulation mistake when siting the drilling barge, Texaco and Wilson Brothers agreed to pay $32 million to Diamond Crystal and $12.1 million to the Live Oak Gardens.

An ongoing environmental catastrophe that was anticipated by environmental groups at the time of the accident never materialized. The lake quickly returned to its natural freshwater state, and with it the wildlife was largely un-affected. Nine of the barges eventually popped back up like corks (the drilling rigs and tug were never to be seen again). The torrent of water helped dredge Delcambre Canal so that it was two to four feet deeper. And of course, the former 1 metre deep Lake Peigneur was now 400 metres deep in the vicinity of the borehole!

Interestingly, the filling of the mine workings with water drastically slowed the rate of land subsidence atop the mine (Figure 14b). Measurements had been carried out between 1973 and 1983, some 7 years before the accident and continued for 2 years afterward (Thoms and Gerhle, 1994). Slowing reflects the post-accident reduction in the total pressure exerted on the roof of the mine to half its pre-accident levels. Prior to the accident, there was no hydrostatic pressure to alleviate some of the lithostatic pressure exerted by the weight of the overburden and so land subsidence above the mine workings was relatively rapid.

Although this incident is not directly related to any aspect of the salt mining operation and no human lives were lost (although three dogs perished), it clearly illustrates the speed of potential leakage following a breach in a cavern roof in any shallow storage facilities filled with low-density fluids. It also illustrates the usual cause of such disasters – human error in the form of a lack of due diligence, a lack of forward planning and a lack of communication between various private and government authorities. It also illustrates that filling a solution cavity with water slows the rate of subsidence atop a large salt cavity and that waters after the disturbance will return quietly to a state of density stratification.

The incident had wider resource implications as it detrimentally affected the profitability of other salt mines in the Five Islands region (Autin, 2002). Even as the legal and political battles at Lake Peigneur subsided, safe mining operations at the nearby Belle Isle salt mine came into contention with public perceptions questioning the structural integrity of the salt dome roof. Horizontal stress on the mineshaft near the level where the Louann Salt contacts the overlying Pleistocene Prairie Complex had caused some mine shaft deterioration. Broad ground subsidence over the mine area was well documented and monitored, as was near continuous ground water leakage into the mine workings. The Peigneur disaster meant an increased perception of continued difficulty with mine operations and an increased risk of catastrophic collapse was considered a distinct possibility. In 1985, a controlled flooding of the Belle Isle salt mine was completed as part of a safe closure plan.

Subsidence over the nearby Avery Island salt mine (operated by Cargill Salt) has been documented since 1986. This is oldest operating salt mine in the United States and has been in operation since the American Civil War, and after the Lake Peigneur disaster the mine underwent a major reconstruction and safety workover. Mine management and landowners did not publicly disclose the technical details of rates of subsidence, but field observations revealed the nature of the subsidence process. Subsidence along the mine edge coincided with a topographic saddle above an anomalous salt zone located inside the mined salt area, ground water had seeped into the mine, and there were a number of soil gas anomalies associated with the mine. Small bead-shaped sinkholes were initially noticed in the area in 1986, then over several years, a broad area of bowl-shaped subsidence and areas of gully erosion formed (Autin, 2002). Reconstruction has now stabilised this situation. Much of the subsidence on Avery Island was a natural process that occurs atop any shallow salt structure. Dating of middens and human artifacts around salt solution induced water-filled depressions atop the dome shows dissolution-induced subsidence is a natural process that extends back well beyond the 3,000 years of human occupation documented on the island.

Compared to the other salt domes of the Five Islands, Cote Blanche Island has benefited from a safe, stable salt mine operation throughout the mine life (Autin, 2002). Reasons for this success to date are possibly; (i) mining operations have not been conducted as long at Cote Blanche Island as other nearby domes, (ii) the Cote Blanche salt dome may have better natural structural integrity than other islands, thus allowing for greater mine stability (although it too has anomalous zones, a salt overhang, and other structural complexities), and (iii) the salt is surrounded by more clayey (impervious) sediments than the other Five Islands, perhaps allowing for lower rates of crossflow and greater hydrologic stability.

Haoud Berkaoui oilfield, Algeria

Located in the Sahara, some 32 km southwest of Ouargla City, the Haoud Berkaoui oilfield is an area of subsidence where numerous exploration and development wells were completed in the 1970s. Of these, the OKN32 and OKN32BIS wells have collapsed into an expanding collapse doline. It surfaced in October 1986 when a crater, some 200 metres across and 75 metres deep, formed (Morisseau, 2000). Today the solution cavity continues to expand and is now some 230 by 600 metres across. Its outward progression is continuing at a rate around 1 metre per year. The collapse is centred on two oil wells drilled in the late 1970s. The problem began in 1978, when the OKN32 oil exploration well was abandoned because of stability problems in Triassic salt at a depth around 650 metres. The target was an Ordovician sandstone at a depth of 2500m. Because of the technical problems associated with significant caverns at the level of the salt, the well was abandoned without casing being set in the salt, probably facilitating the escape of artesian waters (Morisseau, 2000; Bouraoui et al., 2012).

When it reached the 600m level, the well had already passed through 50 metres of anhydrite (220-270m depth) along with interbedded anhydrite clay and dolomite 270m -450 m depth). These are evaporite sediments that, in their undisturbed state, can act as aquicludes or aquitards to any access by unconfined phreatic groundwaters, although at such shallow depths the evaporite beds are likely also to be variably overprinted by active-phreatic dissolution processes. Prior to drilling it was thought that the Senonian halite extended continuously to a depth of 600 m in the well and in turn was underlain by 50 m of anhydrite (600-650m depth). Below the halite-anhydrite is an artesian aquifer (Albian) with a natural hydraulic head that is larger than the surface aquifer head by 2.5 MPa (Morisseau, 2000)

In 1979 a second well, OKN32BIS, was drilled located some 80 metres from the previous well and it successfully obtained its Ordovician target. But in March 1981, the lining of this second well broke, probably because of cavity collapse at a level around 550 metres (once again the regional level of salt) and the well was lost. Five years later, on October 1 1986, a large surface crater formed, centred on these two wells. The initial at-surface diameter was 200 metres and it was 75 metres deep, today it is even larger (Figure 15). Cavern diameter below the stope breakout at that time was estimated to be 300m and water flows to be around 2000 m3/hour.

Since the initial stope breakout, leaching has become progressively less effective and expansion rates have slowed (Morisseau, 2000). This is because cavern growth and water outflow flow are thought to take place preferentially near the centre of the collapse, which is now far from the collapsing cavern walls Dissolving salt may be salinising the crossflowing groundwaters, leading to undocumented, but possible, ongoing degradation of freshwater oases in the region. Continuing expansion is evidenced by the development of fresh centripetal cracks about the expanding collapse margin. Using MT-InSAR analysis, Bouraoui et al. (2012) documented ongoing subsidence near the crater, with an average subsidence of 4 mm per year (between 1992 and 2002). The zone of current zone of subsidence is centred on the OKN32 location and is slowly migrating north east.

As in the USA (see Table 2 and Warren, 2016, Chapter 13 for examples), the loss of these wells, in this case during their active life, emphasizes the need for caution when planning well abandonment in a salt bed, especially when it is highly likely that the salt is acting as a seal, or at least an aquitard, to a regional artesian system. The fact that the first well (OKN32) was lost during drilling argues that a natural breach or cavity was already present in the salt bed and perhaps was already stoping its way to the surface. It is also possible that the inappropriate completion and cementation of casing levels, prior to the well’s abandonment, may have accelerated cavity expansion. In hindsight, the loss of the second well some 5 years later was highly likely as was catastrophic cavity collapse 5 years after that; the OKN32BIS wellhead was situated only 80 metres from OKN32 and was dealing with the same cavern-ridden salt geology.


Regarding anthropogenically-enhanced salt karst, it is important to note that a salt mass used for storage has never failed catastrophically. Weak points tend to occur wherever “the outside has access to the inside,” so problems tend to be mostly where mine expansion breaches a salt edge (Warren, 2017). Likewise, almost all the problems related to well and cavity failure are more a matter of human error, either by negligence, or a lack of understanding by on-the-ground personnel. There is the same general rule of thumb when it comes to salt cavities and salt mines, and that is, “keep it in the salt!” Most failures and breaches occur when mining or solution leaching operations allow the cavity to contact the edge of the salt. There undersaturated water crossflows can exaggerate any uncontrolled dissolution problems. Often the salt edge is irregular due to natural dissolution and assumptions of flat or gently curved shapes to a salt edge are oversimplifications.


Autin, W. J., 2002, Landscape evolution of the Five Islands of south Louisiana: scientific policy and salt dome utilization and management: Geomorphology, v. 47, p. 227-244.

Bouraoui, S., Z. Cakir, R. Bougdal, and M. Meghraoui, 2012, MT-InSAR monitoring of ground deformation around the Haoud Berkaoui sinkhole (SE Algeria): Geophysical Research Abstracts, EGU General Assembly 2012, held 22-27 April, 2012 in Vienna, Austria, v. 14, EGU2012-3344.

Buffet, A., 1998, The collapse of Compagnie des Salins SG4 and SG5 drilling: Proc. S.M.R.I. Fall Meeting, Rome,, p. 79-105.

Gendzwill, D., and N. Martin, 1996, Flooding and loss of the Patience Lake potash mine: CIM Bulletin, v. 89, p. 62-73.

Goodman, W. M., D. B. Plumeau, J. O. Voigt, and D. J. Gnage, 2009, The History of Room and Pillar Salt Mines in New York State,” in S. Zuoliang, ed., Proceedings, 9th International Symposium on Salt, Beijing, China, September 4–6, 2009, v. 2, Gold Wall Press, Beijing, China, p. 1239–1248.

Gowan, S. W., and S. M. Trader, 1999, Mine failure associated with a pressurized brine horizon: Retsof Salt Mine, western New York: Environmental & Engineering Geoscience, v. 6, p. 57-70.

Gowan, S. W., and S. M. Trader, 2003, Mechanism of sinkhole formation in glacial sediments above Retsof Salt Mine, Western New York, in K. S. N. Johnson, J. T. , ed., Evaporite karst and engineering/environmental problems in the United States: Norman, Oklahoma Geological Survey Circular 109, p. 321-336.

Morisseau, J. M., 2000, Uncontrolled leaching of salt layer in an oil field in Algeria: Proc. S.M.R.I. Fall Meeting Technical Session, San Antonio, p. 330-333.

Nieto, A., and R. A. Young, 1998, Retsof Salt Mine Collapse and Aquifer Dewatering, Genesee Valley , Livingston County , NY, in J. Borchers, ed., Poland Symposium Volume: Land Subsidence, Spec. Pub. 8, Assoc. Engineering Geologists, p. 309-325.

Payment, K. A., 2000, Loss of the Retsof salt mine: legal analysis of liability issues, in R. M. Geertmann, ed., Proc. 8th World Salt Symp., Salt 2000, The Hague, v. 1: Amsterdam, Elsevier, p. 399-404.

Tepper, D. H., W. H. Kappel, T. S. Miller, and J. H. WilliaMS, 1997, Hydrogeologic effects of flooding in the partially collapsed Retsof salt mine, Livingston County, New York: US Geol. Survey Open File Report, v. 97-47, p. 36-37.

Thoms, R. L., 2000, Subsidence and sinkhole development over salt caverns: An introduction to the technology of solution mining; Spring 2000 Technical Class, p. 127-141.

Thoms, R. L., and R. M. Gehle, 1994, Analysis of a Solidified Waste Disposal Cavern in Gulf Coast Salt Dome: SMRI Fall Mtg. (1994) 637.

Thoms, R. L., and R. M. Gehle, 2000b, Winnfield mine flooding and collapse event of 1965: Proc. S.M.R.I. Fall Meeting Technical Session, San Antonio, p. 262-274.

Van Sambeek, L. L., S. W. Gowan, and K. A. Payment, 2000, Loss of the Retsof Mine: Engineering Analysis: Proceedings, 8th World Salt Symposium, The Hague, The Netherlands, May 7–11, R. M. Geertman (ed.), Elsevier Science Publishers B.V., Amsterdam, The Netherlands, pp. 411–416.

Von Tryller, H., 2002, The Cavern Field No. 11 in Ocnele Mari - History, Present and Future: Solution Mining Research Institute Proceedings, Spring Meeting, 28 April 1 May, 2002, Banff, Canada, p. 10 pp.

Warren, J. K., 2016, Evaporites: A compendium (ISBN 978-3-319-13511-3): Berlin, Springer, 1854 p.

Warren, J. K., 2017, Salt usually seals, but sometimes leaks: Implications for mine and cavern stabilities in the short and long term: Earth-Science Reviews, v. 165, p. 302-341.

Yager, R. M., 2013, Environmental Consequences of the Retsof Salt Mine Roof Collapse, US Geological survey Open File Report 2013–1174, 10 p.

Yager, R. M., T. S. Miller, and W. M. Kappel, 2001, Simulated effects of 1994 salt-mine collapse on ground-water flow and land subsidence in a glacial aquifer system, Livingston County, New York: US Geological Survey Professional Paper, p. 1-80.

Yager, R. M., P. E. Misut, C. D. Langevin, and D. L. Parkhurst, 2009, Brine Migration from a Flooded Salt Mine in the Genesee Valley, Livingston County, New York: Geochemical Modeling and Simulation of Variable-Density Flow, USGS Professional Paper 1767, 59 p.

Young, R. A., and G. S. Burr, 2006, Middle Wisconsin glaciation in the Genesee Valley, NY: A stratigraphic record contemporaneous with Heinrich Event, H4: Geomorphology, v. 75, p. 226-247.

Zamfirescu, F., M. Mocuta, T. Constantinecu, E. Medves, and A. Danchiv, 2003, The main causes of a geomechanical accident of brine caverns at field II of Ocnele Mari - Romania: RMZ - Materials and Geoenvironment, v. 50, p. 431-434.

Zuber, A., J. Grabczak, and A. Garlicki, 2000, Catastrophic and dangerous inflows to salt mines in Poland as related to the origin of water determined by isotope methods: Environmental Geology, v. 39, p. 299-311.



Post has no comments.
Post a Comment

Captcha Image

Trackback Link
Post has no trackbacks.

Recent Posts


lithium brine magadiite hectorite intrasalt solikamsk 2 rockburst crocodile skin chert potash ore price vanished evaporite astrakanite circum-Atlantic Salt Basins sinjarite Dead Sea caves Badenian Lake Peigneur nuclear waste storage potash sedimentary copper mass die-back anomalous salt zones Dead Sea karst collapse evaporite extrasalt knistersalz MOP saline clay sinkhole Salar de Atacama Weeks Island salt mine deep meteoric potash intersalt potash ore Ethiopia stevensite salt tectonics Ure Terrace Neoproterozoic Oxygenation Event salt mine Dallol saltpan Stebnyk potash cryogenic salt gas in salt oil gusher Five Island salt dome trend alkaline lake evaporite-metal association epsomite Realmonte potash Precambrian evaporites salt periphery nitrogen hydrothermal karst Archean Kalush Potash Sulphate of potash Hyperarid Koppen climate halotolerant palygorskite High Magadi beds CO2 Jefferson Island salt mine Evaporite-source rock association mummifiction Danakhil Depression, Afar geohazard gem CaCl2 brine venice blowout methane African rift valley lakes auto-suture tachyhydrite Turkmenistan evaporite dissolution Hell Kettle Sumo Koeppen Climate salt karst lithium carbonate dihedral angle subsidence basin collapse doline Atlantis II Deep bedded potash salt leakage, dihedral angle, halite, halokinesis, salt flow, eolian transport recurring slope lines (RSL) DHAL Ripon Red Sea organic matter Deep perchlorate carnallitite silicified anhydrite nodules Mega-monsoon Muriate of potash zeolite hydrological indicator evaporite-hydrocarbon association doline salt suture Messinian black salt natural geohazard lunette NaSO4 salts base metal kainitite trona gas outburst Belle Isle salt mine source rock Zabuye Lake hydrogen SedEx Pangaea MVT deposit waste storage in salt cavity climate control on salt bischofite brine evolution halite lapis lazuli Zaragoza jadarite silica solubility Magdalen's Road sepiolite nacholite allo-suture cauliflower chert Catalayud mirabilite water in modern-day Mars Mesoproterozoic snake-skin chert lot's wife Lomagundi Event dark salt evaporite karst salt trade sodium silicate ancient climate Hadley cell: hydrohalite lithium battery Kara bogaz gol Platform evaporite capillary zone Warrawoona Group namakier meta-evaporite water on Mars anthropogenically enhanced salt dissolution Ingebright Lake mine stability Pilbara supercontinent marine brine H2S authigenic silica vadose zone Paleoproterozoic Oxygenation Event SOP Great Salt Lake hydrothermal potash Musley potash Karabogazgol salt ablation breccia basinwide evaporite chert Clayton Valley playa: well blowout halite-hosted cave dissolution collapse doline antarcticite salt seal Neoproterozoic North Pole flowing salt halokinetic seal capacity lazurite Lake Magadi Crescent potash freefight lake gassy salt Quaternary climate gypsum dune Stebnik Potash Proterozoic Deep seafloor hypersaline anoxic lake