Sinkholes in Evaporite Rocks

Joseph D Martinez, Kenneth S Johnson, James T Neal. American Scientist. Volume 86, Issue 1. Jan/Feb 1998.

Sinkholes characterize karst topography the world over. A mainly natural phenomenon, their development in carbonate rocks such as limestones of Florida and the Yucatan Peninsula of Mexico is well known (see “Karst Lands,” American Scientist, September-October 1995). However, the fact that sinkholes also form in or above highly soluble evaporite rocks, such as gypsum and salt, is far from common knowledge. And these sinks can be at least as dramatic and troublesome as those in carbonates.

Evaporite deposits form when various salts precipitate from evaporating water, mainly seawater. The principal evaporite rocks include gypsum (or anhydrite, its anhydrous form) and salt (halite), although potash (sylvite) and other rarer salts also are locally important. Evaporites have the highest solubility of common rocks; water that is unsaturated with respect to gypsum (CaSO42H20) or salt (NaCl) rapidly dissolves them and carries them off in solution. Indeed, gypsum and salt are, respectively, about 150 and 7,500 times more soluble than limestone. Such high solubilities enable subsurface dissolution channels and sinkholes to form in a matter of only days, weeks or years, and catastrophic collapse can result. Evaporite rocks underlie about 35-40 percent of the United States and are found in 32 of the 48 contiguous states. They are also found in Canada and Mexico, and are widespread on other continents.

Unlike carbonate sinkholes, which resuit mostly from natural processes, those in or above evaporite rocks can form as the result of either natural processes or human activities. A number of striking natural sinks exist in gypsum beds in Texas, New Mexico, Oklahoma and Nova Scotia. They also show up in gypsum diapirs-domes or arches of rock ruptured by the more plastic gypsum pushing upward from below. Excellent examples of these can be found in the Magdalen Islands in the Gulf of St. Lawrence, Canada, and in the Sierra del Fraile, in northeastern Mexico. Natural karst features, including major sinkholes and subsidence troughs, also occur over bedded salt in Texas, New Mexico and Arizona. The mining of rock salt, or the drilling of boreholes into or through rock salt, have accidentally created a number of large, man-made sinkholes in Louisiana, Texas, Kansas, Michigan and New York. Regardless of their origin, sinkholes in evaporites pose significant hazards and are costly to society. Only through a better understanding of the processes by which sinkholes form and a recognition of the regions in which they are likely to form can we minimize such risks.

Natural Sinkholes

Natural sinkholes in evaporite rocks develop by the same processes that form sinkholes in carbonates (limestone and dolomite), except that they develop much more rapidly. Water percolates over or through gypsum or salt and dissolves the highly soluble rock; typically, this leads to the formation of sinkholes, caves, natural bridges, disappearing streams and springs. There are four basic requirements for evaporite sinkholes to develop: a deposit of gypsum or salt; water, unsaturated with CaSO4 (calcium sulfate) or NaCl; an outlet for escape of dissolving water; and energy to cause water to flow through the system. Once a through-flowing passage forms in the evaporite rock, enlargement results from further dissolution and from abrasion by water-borne particles transported through the cavity.

Sinkholes and other karst features are known at least locally (and sometimes quite extensively) in almost all areas underlain by evaporites. The most widespread and pronounced examples of both gypsum and salt karst in North America are in the Permian basin (including parts of Texas, New Mexico and Oklahoma), but we shall also describe naturally occurring sinkholes in the Holbrook basin of northeast Arizona; Nova Scotia and the Magdalen Islands of eastern Canada; and the Sierra del Fraile of northern Mexico.

The Permian Basin

A large region of the southwestern United States is underlain by salt, gypsum and other evaporites of Permian age (about 290 to 230 million years old) that were deposited along with red beds (red-colored shales and sandstones) and carbonate rocks. During the Permian Period, a shallow inland sea extended from West Texas into northwestern Kansas-perhaps the equivalent of the modern-day Caspian Sea and the Gulf of Kara-Bogaz. The Permian basin contains one of the largest and thickest (typically 500-1,500 meters thick) accumulations of evaporites in the world. Gypsum sinkholes and other karst features are widespread in much of the basin, and subsurface salt beds were extensively dissolved in the geologic past to form many collapse sinks now filled with younger sediments.

The Delaware basin of West Texas and southeast New Mexico, in the southwest part of the Permian basin, contains evaporites of the Late Permian Castile, Salado and Rustler Formations. Outcrops of these three formations constitute the most extensive examples of gypsum sinkholes and karst in the nation. The area called the Gypsum Plain comprises about 2,600 square kilometers of outcropping gypsum of the Castile and Salado Formations, and additional gypsum outcrops are present just to the east in the Rustler Hills and in Reeves County, Texas. Delawarebasin gypsum deposits contain abundant sinkholes, caves, closed depressions, collapse sinks and underground drainage. Much of the area has been affected by subsurface dissolution of some of the salt layers, and most of the outcrops consist of massive beds of gypsum. Aside from the Gypsum Plain, other principal areas of gypsum sinkholes and karst are Nash Draw, Burton Flat and the Pecos River Valley. Sinkholes, a few meters to 100 meters across, are active collapse features in all four areas, and generally they are related to shallow, underground caverns less than 100 meters deep. One sinkhole, formed during a storm in 1918, collapsed suddenly to form a gaping hole about 25 meters across and 20 meters deep. In that one night, nearly 23,000 cubic meters of soil and bedrock disappeared into the cavern. Sinkholes are so numerous and dangerous in some areas that they are fenced off to prevent cowboys and cattle from accidentally falling in.

Along the western flank of the Permian basin, in eastern New Mexico, gypsum crops out extensively along parts of the Pecos River Valley. Various gypsum and carbonate units in the Permian Artesia Group and San Andres Formation contain a large number of sinkholes, caves and other karst features in the Vaughn-Roswell area.

Bottomless Lakes State Park, located about 20 kilometers east of Roswell, has numerous sinkholes, most of which are circular, steep walled or vertical, 50-100 meters across and 30-60 meters deep. They are aligned, almost like a string of pearls, along the east side of the Pecos River floodplain. Outcropping rocks include gypsum, red beds and dolomite of the Artesia Group, whereas underlying rocks are gypsum, salt, dolomite and red beds of the Artesia Group and San Andres Formation. Groundwater has dissolved both salt and gypsum in the shallow subsurface, forming underground channels and caverns; the caverns have caved in, resulting in steepwalled sinks now partly submerged by the lakes. Water in the lakes is crystal clear and generally brackish to saline (6,000-23,000 parts per million total dissolved solids), attesting to its passage through subsurface layers of gypsum and salt. Although the lakes typically are S30 meters deep, dark-green moss on their bottoms gives the impression of great depth. Of course, the lakes are not “bottomless,” because, as one local wag remarked, “If the lake had no bottom, wouldn’t the water run out?”

Another major gypsum-sink area lies along the eastern flank of the Permian basin, in western Oklahoma and northcentral Texas, where the principal gypsum units are the Permian Blaine and Cloud Chief Formations. Sinkholes in the extensive gypsum outcrops typically measure 1-30 meters in diameter and 3-25 meters deep, and in many areas their abundance restricts land use to the grazing of a few cattle.

Among the more important gypsum-karst features of the region are two well-known caves and a major salt-water spring, all developed in the Blaine Formation. The D. C. Jester Cave of southwestern Oklahoma has a 2,413-meter-long main passage. But with the inclusion of side passages, the total length is 10,065 meters, making it the longest reported gypsum cave in the western world. The cave, located about 25 kilometers northwest of Mangum, has many sinkhole passageways that open to the surface. Another cave, Alabaster Cavern, just south of Freedom in northwestern Oklahoma, has been developed as a tourist attraction. It has a main passage about 700 meters long, a maximum width of 18 meters and a maximum height of 15 meters; many sinkholes in the area connect to Alabaster Cavern and other nearby smaller caves.

A natural salt-water spring previously discharged about 7,000 liters per minute of high-salinity brine from a gypsum sinkhole located on the Red River floodplain just north of Estelline, in north-central Texas. The nearly vertical, brine-filled sinkhole beneath Estelline Spring has been explored by SCUBA divers; it is generally 8-15 meters across, extends down 38 meters underground through a series of partly dissolved and collapsed gypsum, dolomite and shale layers, and tapers to a small opening (0.6 by 1.0 meter) that extends down to an unknown additional depth and size. The surface flow of brine, supplied by dissolution of nearby subsurface layers of salt, was contained in 1963 by an earthen dike that imposes a hydrostatic head 1.5 meters above ground level.

In most parts of the Permian basin evidence exists of past dissolution of subsurface Permian salts, resulting in sinks, collapse structures and subsidence troughs now filled with post-Permian sediments. Fresh groundwater, recharged in nearby upland areas, percolates down through porous and permeable pathways until it encounters salt beds. Dissolution of the salt forms brines and also creates underground cavities into which the overlying rock settles or collapses. If a cavity is large enough or shallow enough, the collapse feature migrates upward, through successive stoping of the roof, and can reach the land surface as a sinkhole or subsidence trough. Very old sinks or troughs, those formed thousands or millions of years ago, typically are filled with younger, Cenozoic (about 66 million years old and younger) sediments deposited from streams and other surface flow carried into the depressions. These sediment-filled sinks and depressions are abundant above salt-dissolution fronts in the Delaware basin, in eastern New Mexico and across much of the Texas Panhandle.

Holbrook Basin of Arizona

The Holbrook basin of northeast Arizona, covering some 6,000 square kilometers, contains 300400 meters of Permian evaporite deposits consisting of salt, anhydrite and potash minerals in the Schnebly Hill Formation. The evaporites are covered by 200-750 meters of Permian and Triassic (about 230 to 185 million years old) age shales, sandstones and limestones of the Coconino and younger formations. These units were tilted during uplift of the Colorado Plateau, and this affected regional groundwater flow toward Colorado River tributaries. The intensified hydrologic regime increased dissolution of the evaporites along a northwardmigrating front, now a region with hundreds of sinkholes and other karst features. A tectonic flexure known as the Holbrook anticline (an arch-shaped deformation), extending northwesterly for more than 100 kilometers, has been modified by the dissolution, with as much as 100 meters of collapse and topographic displacement. The collapse features, sinkholes and reverse dip (the angle at which rock layers are inclined from horizontal) of the flexure directly overlie the salt-dissolution front, and this marks the location of two major collapse depressions known as The Sinks and Dry Lake Valley. Dry Lake Valley has periodically held water in a playa lake during the 20th century, but seldom for extended periods because new sinkholes actively form and drain the lake. Major drainage events took place in 1963,1979,1984 and 1995, with more than 50 new sinkholes forming during that period. The continuing appearance of new sinkholes testifies to the ongoing nature of dissolution in the underlying evaporites.

Other karst features in the Holbrook basin attributed to evaporite dissolution are pull-apart fissures, graben sinks (downward-dropped blocks), breccia pipes and plugs (zones filled with cemented angular rock fragments), and many small depressions with and without sinkholes. The presence of the more than 500 karst features in the Holbrook basin evokes practical concerns, even in such a sparsely populated region. Industrial development, transportation and human habitation are all constrained by the presence of such features.

Eastern Canada

Nova Scotia contains enormous deposits of Mississippian (about 360 to 330 million years old) gypsum overlain by extensive Pleistocene (from less than about 1.6 million to 10,000 years ago) glacial drift. Numerous sinkholes can be found in this unique terrain in Antigonish County. These range from small, conical-shaped empty holes to larger, water-filled sinks up to 60 meters in diameter and 20 meters or more in depth. Groundwater in gypsum terrain typically is enriched in sulfate, but, surprisingly, the water in these sinkholes is potable and used by both people and cattle.

The gypsum resources of the province have been mapped by Gordon Adams of the Nova Scotia Department of Natural Resources, who has cited the presence of numerous sinkholes. He reported in 1991 that karst features are exposed locally in some areas of surface mining for gypsum and that these gypsum sinks are filled with Pleistocene sediments. Livestock and machinery have also been known to fall into some sinks, and in at least one case a home was threatened by a newly developed sinkhole. In Antigonish County, a farmer described an incident in which a calf fell into a dry sinkhole and then climbed out on bales of hay thrown into the depression. In the Milford quarry of National Gypsum Ltd., one sinkhole contained glacial drift along with a nearly complete skeleton of a mastodon. Such discoveries help establish the chronology of sinkhole formation and burial.

The Magdalen Islands in the Gulf of St. Lawrence, Canada, are underlain by salt and gypsum diapirs; upward movement of the evaporite masses undoubtedly accounts for the existence of the islands. A salt mine currently operates on one of the islands. Gypsum diapirs crop out locally, and sinkholes have developed in at least one of the diapirs. One of us (Martinez) was shown one of these sinkholes by George Mueller, a consultant for SOQUEM (a crown corporation of Quebec) in 1973, while assisting him in exploration for salt. The gypsum was schistose (recrystallized and foliated), which appears to be characteristic of gypsum diapirs.

Northeastern Mexico

Gypsum and anhydrite are exposed in the cores of breached anticlines in parts of northeastern Mexico. These evaporites are found in the Sierra del Fraile folds of the Sierra Madre Oriental, near Monterey, Mexico, as well as in the Coahuila Marginal Folded Belt. They are part of the Minas Viejas Group, probably Jurassic in age (185 to 130 million years old). Early studies by James R. Wall (with the USGS at the time) and his colleagues called attention to the intrusive nature of the gypsum diapirs in Sierra del Fraile. Alfred E. Weidie, of the University of New Orleans, and one of us (Martinez) reexamined the geology of the area and also mapped the gypsum schistosity One location, Potrero Chico, displays a topographic high formed from a breached anticline that is a mass of intrusive gypsum with sinkholes. The field studies included Weidie descending into one sinkhole, 10-12 meters deep, by rope to measure the strike and dip of schistosity of the gypsum.

Human-Induced Sinkholes

Human activities can either induce or enhance karst processes in evaporite rocks. Typically, such activity involves either construction on, or the directing of water into or above, outcropping or shallow evaporite deposits, or the drilling of boreholes into or through subsurface salt deposits. Because human-induced sinkholes in gypsum areas are uncommon, we shall describe only those in salt.

Drilling boreholes into or through bedded salt or salt domes, or opening dry mines in either type of salt, can enable (either intentionally or inadvertently) unsaturated water to encounter and dissolve the salt. Such water remains unsaturated, and continues to dissolve salt, until it reaches about 300,000 parts per million (30 percent, by weight) NaCl, nearly 10 times the salt content of normal sea water. If the dissolution cavity is large enough and shallow enough, successive roof failures can cause the water-filled void to migrate upward; land subsidence or catastrophic collapse can result. Solution mining of salt, dry mining of salt and petroleum activities can all lead to local salt dissolution and sinkhole development.

Sinkholes over Solution Mines

Solution mining of a soluble mineral such as salt involves introducing unsaturated water into the subsurface salt mass, dissolving the salt to form a brine, recovering the brine and then extracting salt from the brine, usually by evaporation. Solution mining typically entails creating one or several large underground cavities that become filled with brine. Cavities typically are 10-100 meters in diameter and 10-600 meters high, both dimensions based largely on the thickness of the salt and the depth to the top of the cavity. Occasionally, the cavity becomes too large and the roof collapses. C. Richard Dunrud and Barbara B. Nevins, of the U.S. Geological Survey, reported 10 areas of solutionmining collapse in the United States, as of 1981. Most such collapses result from cavities formed 5075 years ago, before modern-day engineering safeguards were developed. Modern design and construction have virtually eliminated this problem. Four well-documented sinkholes resulting from solution mining are Cargill sink (Kansas), Grand Saline sink (Texas), Grosse Ile (Michigan) and Bayou Choctaw (Louisiana).

Cargill sink formed on October 21, 1974, as a result of solution mining for salt by Cargill, Inc., near Hutchinson, Kansas. Robert E Walters, of Wichita, reported that the surface crater reached a diameter of 60 meters within four hours and then stabilized with a diameter of 90 meters and a maximum depth of about 15 meters; the volume of the crater was calculated to be about 70,000 cubic meters. Salt had been solution mined beneath this property since 1888. The Permian Hutchinson salt is about 105 meters thick at this site and occurs at a depth of about 130 meters. The salt is overlain by 110 meters of Permian shales, and these, in turn, are covered by about 20 meters of water-saturated, loose Quaternary (1.6 million years ago to present) sand. Because the locations of some of the earliest brine wells are unknown, and the early dissolution methods were not well controlled, the location and extent of some of the solution cavities on this property are unknown. The sink developed in an active brine field that included both operating and abandoned wells. Embraced within the sinkhole was a brine well that was drilled in 1908 and plugged and abandoned in 1929. The collapse also left the tracks of the Missouri-Pacific Railroad suspended 6 meters above the water that accumulated in the sinkhole.

Post-subsidence test drilling of the Cargill sink area, as described by Walters, showed that an elongated cavern had developed in the salt beneath the sink. The cavern parallels a line of brine-producing wells that were hydraulically connected. The span of the cavern roof is more than 400 meters long and less than 90 meters wide. Apparently the roof span exceeded the capacity of overlying shales to support the overburden. Therefore, failure of the roof caused collapse of successive overlying rock units until the uppermost rock layer finally dropped into the water-filled void. At this point, the water-saturated Quaternary sands flowed into the cavity, creating the surface sink. The sand flowed down and now fills a subsurface chimney that is about 30 meters in diameter and located below the center of the sinkhole.

Grand Saline sink developed in the city of Grand Saline, Texas, in 1976. The sink occurred at the site of a brine well that penetrated the top of the Grand Saline salt dome at a depth of 60 meters and had produced brine from 1924 through 1949. The sink eventually grew to a diameter in excess of 15 meters, and a total of 8,500 cubic meters of silt and clay was displaced into the underground cavity. A house was endangered by this collapse, and, to make matters worse, a pipeline broke and sewage appeared to flow into the hole.

On Grosse Ile, located in the Detroit River near Detroit, several sinkholes developed in 1971 as a result of 30 years of solution mining of the Silurian Salina Group salts at a depth of 325 meters. Dunrud and Nevins estimated that the area affected by subsidence was 37,000 square meters and the volume of subsidence was 1.2 million cubic meters. A report by Kenneth Landes and Thomas B. Piper, sponsored by the Solution Mining Research Institute shortly after sinkhole development, described and analyzed the effect on the environment of brine-cavity subsidence at Grosse Isle. Although a moderately large area was affected, the subsidence was contained and was on access-controlled lands; the authors expressed the opinion “that these sinkholes have produced no significant damage to the resources or the environment.”

In 1954, the Bayou Choctaw salt dome, just southwest of Baton Rouge, Louisiana, was the site of a large and deep collapse over a solution-mined brine cavern. A drilling rig that was producing brine at the time fell into the collapsing depression and was never seen again. The collapse created a large lake, about 250 meters in diameter, which is still present. The cavern-roof failure appeared to be caused by lack of a protective oil blanket over the brine in the cavity, a necessary precaution in modern brining operations, as described by Martinez in an earlier issue of American Scientist (September-October, 1991).

Sinkholes over Dry Salt Mines Sinkholes can form at the surface above underground mines of all kinds as a result of the collapse of overburden into subsurface voids, especially where the mines are shallow and overburden is poorly supported because of large roof spans. Dry (room-and-pillar) salt mines may create special problems in this regard, especially when groundwater finds pathways into the minedout voids. If a leak develops, and unsaturated water enters a dry salt mine, the salt can dissolve quite rapidly With the introduction of large volumes of water, mined-out rooms can fill with salt-dissolving water, and the facility then behaves like a solution mine. Uncontrolled dissolution can enlarge the rooms, remove salt pillars that support the roof, dissolve salt that remains in the roof, induce collapse of the overburden and thus create small to large sinkholes. Shallow salt mines are more likely to develop leaks, whereas salt mines more than 300 meters deep seldom have significant leaks; the higher lithostatic pressure at the greater depths reduces open fractures and rock permeability, and thus the capacity for water inflow. Three dry salt mines that have been invaded by water and have developed overlying sinkholes are at Jefferson Island (Louisiana), Weeks Island (Louisiana) and Retsof mine (New York). All are now closed.

The Jefferson Island salt dome, just west of New Iberia, Louisiana, was the site of a catastrophic sink that developed in 1980. This huge sinkhole presumably was triggered, if not caused, by the inadvertent penetration by an oil rig of a chamber of the Jefferson Island salt mine, some 400 meters below shallow Lake Peigneur. Lake water began leaking through the borehole into the underground room-and-pillar mine. The flow of water became a torrent; soon the entire lake was drained, and more water was drawn through a canal that connected the lake to the Gulf of Mexico. The $5 million drilling rig vanished into the giant sink, along with barges, a tugboat and 4 hectares of land. Although there were more than 50 workers in the mine and on the lake at the time of the accident, there was no loss of life-a miracle that was the result of well-planned evacuation and escape procedures established for just such an emergency. This multimillion-dollar incident (see American Scientist, September-October, 1991) has alerted operators of other Gulf Coast salt mines to the risk, however small, of this kind of accident.

The Jefferson Island incident raised special concerns about sinkhole problems that later were discovered at the Weeks Island salt dome, about 30 kilometers away, where 73 million barrels of crude oil have been stored since 1981. The crude oil was stored in a pre-existing room-and-pillar salt mine as part of the Strategic Petroleum Reserve of the U.S. Department of Energy (DOE). A sinkhole measuring 10 meters across and 10 meters deep was first observed in alluvium over the SPR mine at Weeks Island in May 1992, apparently having formed about a year earlier. This, the first of two sinkholes, gradually enlarged and deepened, concurrent with the increasing dissolution of salt over the mined storage area below. Beginning in 1994, and continuing to the present, the injection of saturated brine directly into the sinkhole throat, some 75 meters beneath the surface, essentially arrested further dissolution; this action bought time for DOE to prepare for the safe and orderly transfer of crude oil to another storage facility. The transfer was deemed advisable because of the threat of a sudden collapse and uncontrolled release of oil, however unlikely such an event may be. The second, much smaller sinkhole was first noticed early in 1995, but it did not constitute a serious threat to the mine at that time.

Recognition of sinkhole causative processes led DOE to a decision, in December 1994, to decommission the facility and relocate the oil to other storage facilities. Almost all of the oil was removed by 1997, except for a small amount left as a protective blanket at the top of the mine as it is backfilled with saturated brine; this process will take about one year to complete. Meanwhile, to protect the stability of the mine, injection of saturated brine into the sinkhole throat will continue. To provide added insurance during oil transfer, a “freeze curtain” was constructed in 1995 around the principal sinkhole by installing 54 wells that allowed freezing of the overburden and upper salt to a depth of 67 meters. This freeze wall is expected to prevent groundwater flow into the mine through the sinkhole until the mine is completely filled with brine. Dealing with this sinkhole has been costly. Mitigation and the removal and transfer of oil, including dismantling of infrastructure (pipelines, pumps etc.), cost a total of nearly $100 million; the freeze curtain itself cost nearly $10 million.

In hindsight, based on an earlier event, one might fault the initial decision to select this mine for oil storage, because a groundwater leak in an adjacent part of the mine in 1978 may have been a forewarning of events to come. Injection of cement grout into the flow path controlled the leak at that time, but it could just as easily have become uncontrollable and formed a sinkhole then, had the appropriate mitigation steps not been taken. In 1978, however, the technology needed to understand the mine conditions, predict future events and thus influence management decisions was not available.

A temblor in western New York, first thought to be a 3.6-magnitude earthquake, awakened local residents near the town of Cuylerville, about 50 kilometers south of Rochester, during the early-morning hours of March 12,1994. Concurrent with this sharp energy release and ground shaking, a major leak began flooding the world’s largest salt mine-the Retsof mine, a room-and-pillar mine first operated in 1885, which underlies some 26 square kilometers. The source of the temblor, according to a report by Lawrence Lundgren at the University of Rochester and Richard Young at the State University of New York, Geneseo, was a precipitous collapse of part of the mine’s roof 340 meters below the surface. Developed in Late Silurian (433 to 410 million years ago) Salina Group salts, rooms in the salt mine were about 3.5 meters high. The roof failure opened hydraulic pathways through newly created fractures in overlying rocks; groundwater percolated down through, and partly dissolved, a series of salt layers above the mine and eventually entered the salt mine itself.

About four weeks after the initial collapse in the Retsof mine, dissolution of salt pillars in the mine (and portions of overlying salt beds), along with downward slump or piping of thick, nearsurface unconsolidated sediments, caused subsidence of the land above the mine, resulting in two large sinkholes-the larger some 200 meters across and about 20 meters deep. The flooding of the mine, at a flow rate that eventually exceeded 60,000 liters per minute, could not be controlled by inmine grouting (the normal practice), and the entire mine was flooded; it is now closed and unlikely to reopen. Adverse impacts in the area include disruption of land, abandonment of four homes, damage to other homes (some as much as 1.5 kilometers from the sinkholes), loss of a major highway and bridge, loss of water wells and prohibition of public access to the collapse area. Although sinkholes above the Retsof mine caused major local disruption, the more significant, longer-term effects may lie in economic disruption to the community, along with possible degradation of groundwater supplies.

The combined economic loss resulting from these three human-induced sinkholes, to the mine operators and local communities surrounding Jefferson Island, Weeks Island and Retsof mine, amounts to several hundred million dollars, but the longer-term costs may be much higher. The Jefferson Island incident was attributed to accidental intrusion, but the Weeks Island and Retsof incidents were the result of natural processes superposed on human activities-mining of underground openings where the local mining effects and hydrogeology were not completely understood.

Sinkholes Related to Petroleum Activity Petroleum-industry activities can produce unintentional dissolution cavities through the drilling of exploration, production or disposal boreholes into, or through, subsurface salt units. Unintentional dissolution of the salt can create a cavity that is as large and shallow as those formed during solution-mining activities. And if the cavity becomes too large for the roof to be self-supporting, successive roof failures may cause the collapse to migrate upward and reach the land surface. The few collapse sinks related to petroleum activity involve boreholes drilled long ago, before development of proper engineering safeguards pertaining to drilling-mud design, casing placement and the use of salt-tolerant cements. Three well-documented subsidence/collapse sinks resulting from petroleum activities are the Wink sink (Texas), Panning sink (Kansas) and the Gorham oil field (Kansas).

The Wink sink, located near the town of Wink in Winkler County, Texas, formed on June 3,1980, and within 24 hours had expanded to a maximum width of 110 meters. Two days later, the maximum depth of the sinkhole was 34 meters and the volume was estimated at about 159,000 cubic meters, according to Robert W. Baumgardner of the Texas Bureau of Economic Geology. The collapse took place near the middle of the Hendrick field, a giant oil field that has been operating since 1926; one abandoned well that produced oil from 1928 to 1951 was incorporated within the sink itself, and a second oil well soon was plugged and abandoned because of its proximity to the sinkhole. It appears that the Wink sink resulted from an underground dissolution cavity that migrated upward by successive roof failures, thereby producing a collapse chimney filled with brecciated rock. The dissolution cavity had developed in salt beds of the Permian Salado Formation, which is about 260 meters thick and about 400-660 meters beneath the Wink sink.

The abandoned oil well embraced by the Wink sink probably was a pathway for water to come in contact with the Salado salt. The well was drilled in 1928, in all likelihood using a freshwater drilling fluid that enlarged or washed out the borehole within the salt sequence. Ineffective cement jobs and possible fractures in the cement lining may have opened pathways for water movement up or down the borehole outside the casing. Because the borehole undoubtedly enlarged during drilling in the Salado salts, the small amount of cement reportedly used to set the casing in the hole was enough to cement only the lower part of the hole-thus leaving most of the salt layers unsealed behind the casing. The casing probably was perforated by corrosion resulting from the coproduction of great quantities of brine with the oil. Furthermore, the use of explosives to realign the well while drilling in underlying formations not only fractured the rock and increased its permeability locally, but also may have fractured the cement lining farther up the borehole. In addition, the final removal of casing from the well in 1964 left an unlined borehole adjacent to several aquifers above the Salado salt, and fresh to brackish water drained down into the salt for a period of 16 years, until the Wink sink formed.

All of the abovementioned activities, though consistent with regulations and standard industry practices during the life of the well, could have aided in conducting freshwater from shallow aquifers down the borehole to the salt beds. Outlets for the high-salinity brine thus formed included the porous and permeable strata underlying the Salado Formation, as well as possible preexisting dissolution channels within the Salado. Thus a dissolution cavity may have formed around the well, probably in the upper part of the salt sequence, and this cavity eventually may have become sufficiently large to permit collapse of the roof. Because of successive roof failures, the cavity then migrated upward until it finally reached the land surface and created the Wink sink. There have been no follow-up geophysical or drilling studies to determine the size, shape or exact position of the cavity or chimney beneath the Wink sink.

Panning sink, in Barton County, Kansas, was formed in 1959 by subsidence and collapse around a salt-water-disposal (SWD) well on the Panning lease. Walters reported that the sinkhole reached a diameter of 90 meters and was at least 18 meters deep. The suspect well, drilled originally as a producing oil well in 1938, penetrated 91 meters of Permian Hutchinson salt at a depth of 298 meters. Freshwater drilling fluids dissolved the salt in the borehole to an excessive diameter (1.4 meters), and this washed-out zone was not cemented behind the 15.2-centimeter-diameter casing. Conversion of the borehole to an SWD well from 1946 to 1958 caused a large quantity of unsaturated oil-field brines (with a total concentration about the same as seawater) to be pumped into the well and to inadvertently further dissolve the salt. A large cavern was formed, and with successive roof falls, the water-filled void migrated upward to cause surface subsidence, tilting of an oil-field derrick at the site and eventual collapse of surface sediments to form the sinkhole.

Gorham oil field, in Russell County, Kansas, is the site of slow subsidence of a major highway (Interstate 70) above salt-dissolution zones in the Permian Hutchinson salt. A series of oil wells, drilled on 4-hectare spacing in 1936 and 1937, penetrated 75 meters of salt at a depth of 390 meters, according to Walters. The wells are now plugged and abandoned, but some of them contain corroded casing that has been left in the boreholes above, within and below the salt unit. As a result, unsaturated water probably has flowed down some of the boreholes and dissolved large volumes of the salt. Subsidence of I-70 pavement has occurred at rates of less than 0.3 meters per year, but cumulative subsidence through 1987 in two sinks is about 4 meters, and in a third sink is about 0.3 meter; this has required rebuilding parts of I-70 in 1971 and again in 1986.

Conclusions

Although the processes that produce natural sinkholes generally are not under our control, we do understand their natural history in evaporites and can take steps to reduce or ameliorate their ill effects. It would be best to avoid living or building over evaporite-karst areas, but this is easier said than done. Anticipating sinkhole development requires local understanding of how they form and detailed geologic mapping, which are data and resources that are not always available. If life or property is threatened by sinkholes that form after construction of a facility, the alternatives are to design a safe engineering solution (if the sink is small and stabilized), or abandon the site.

Human-induced sinkholes are most likely to develop above salt deposits. They can be prevented in most cases by not permitting unsaturated water to flow into or through the salt. This can best be accomplished by a combination of the following: detailed geologic mapping of the surface and subsurface; assessing the hydrogeology of the area; designing engineering systems that prevent unintended penetration of mines or cavities in salt; proper design, construction and maintenance of solution mines and room-and-pillar mines in salt to prevent roof collapses; and proper casing or sealing of salt beds when boreholes are drilled into or through the salt. Most human-induced sinkholes form in operations developed more than 75 years ago, before modern-day engineering safeguards were developed; proper, modern design has virtually eliminated this problem in new facilities.

Ironically, once a sinkhole has reached a state of equilibrium and has filled with water, a lake of great beauty may be formed. Water-filled sinkholes provide many life-enhancing opportunities: wildlife habitat, fishing, birding, boating, swimming and other recreational activities. They may even provide potable water, as in Nova Scotia. To cast our eyes on a beautiful lake in hilly or wooded terrain is a joy indeed. But we best remember that pretty as they may be today, sinkholes may have caused a great deal of trouble at their birth.