Discussion of Faults in Houston Area
Environmental Aspects of Faulting
The consequences of active faulting are many and, almost without exception costly and undesirable. Visible fault damage was an aid to mapping throughout our work, but no special study of the impact of faulting was attempted. Nevertheless, some generalities, based on observations and measurements along nearly 300 km of active faults, can be made.
Rigid structures such as highways, runways, and building foundations, where built astride active faults, commonly display the consequences of differential vertical movement and horizontal extension. Breakage occurs as the sinking ground surface on the downthrown side no longer supports the structural load. Maximum damage normally occurs on and immediately adjacent to the fault scarp, within a zone that is typically 2–3 m wide, but which may vary in width from less than 1 m to more than 10 m. To either side of this zone, some limited fault-related failure may be seen in brittle material, but the most visible effect of fault movement here is tilting. The amount of tilt progressively decreases away from the fault scarp and is, for most faults, minor at distances of 20–30 m from the scarp. Surveying data, however, suggest that detectable amounts of tilt may be found as much as 70 m away from some faults. The effects of faulting are not identical on both sides of a fault; typically the downthrown block shows more damage, and the zones of breakage and tilting are wider, than for corresponding zones on the upthrown side.
Common effects of faulting beneath buildings include foundation failure (and attendant periodic infiltration of water), tilting of floors, distortion of window and door frames, leaking roofs, and extensive damage to facing materials and interior wails. The road network in the Houston area has also been damaged by faulting at hundreds of locations, some of which have necessitated repair at intervals as frequent as every 2 years. Airports have fared little better; the faulted runways at both William P. Hobby Airport and at Ellington Air Force Base (fig. 1) require nearly constant maintenance.
It should be noted that dip-slip movement* along any nonvertical fault must generate both vertical offset and horizontal separation at the surface. Most faults in the shallow subsurface are inclined at angles of 60°–75° relative to a horizontal plane (Van Siclen, 1967), which indicates that the value for the horizontal component of movement ranges from about one-fourth to one-half the value of the vertical movement (fig. 4). Repair of damaged structures, or the design of new structures that must bridge an active fault, should therefore take both components of movement into account.
Fault-related damage to residential, commercial, and industrial facilities, and to the road net is commonly obvious, but perhaps of no greater importance than more subtle effects of faulting. Buried pipelines, for example, must experience the same amount of distortion as do streets and buildings at the surface, but damage becomes obvious only upon sudden failure of the pipe. Clogging of sewer lines by sediment and contamination of water supplies are known consequences that necessitate costly repairs. Rupture of an oil or high-pressure gas line could have more serious consequences, but such pipes have considerable strength and we presently are aware of only one break that is almost certainly a result of fault movement (the September 1978 rupture of an 8-inch o.d. crude-oil pipeline that c the Long Point fault near Spring Creek)2. Other breaks, however, may not have been attributed to faulting owing to lack of prior knowledge that a fault exists.
Drainage ditches in Houston necessarily operate on very low gradients. We have observed gradients altered by faulting; locally, fault movements have decreased or reversed flow along sections of numerous ditches in the Houston area, an effect of obvious importance to flood-control programs. Homeowners on the downthrown sides of faults in areas of impaired drainage are increasingly susceptible to flooding, as their homes lose more elevation with every increment of fault movement. In such areas it is possible to suffer fault-related damage, caused by flooding, hundreds of meters from the nearest fault. Gravity-fed sewer lines may likewise experience reversed flow, leading to backup of storm drains and ponding of sewage at its sources. This effect has been noted at several locations, particularly along the Eureka Heights and Long Point faults.
Faults in the Houston area, though numerous, need not significantly impair future development of the region. With prior knowledge of fault locations and responsible planning of engineered structures, many problems evident in existing areas can be avoided in future construction.
Rates of Fault Displacement in the Houston Metropolitan Area
The first and, to our knowledge, the only attempt to study short-term variations in fault movement was that of Reid (1973), who, in 1971, installed instruments on three faults (Long Point, Piney Point East, and Eureka Heights) to measure both vertical and horizontal components of displacement. To paraphrase Reid (1973, p. 87), the first 14 months of observation revealed the following:
(1) The annual rate of vertical separation ranged from 6.7 mm/yr at a monitored site on the Eureka Heights fault to 34 mm/yr at a site on the Long Point fault.
(2) The vertical separation was accomplished by individual creep events that ranged from 0.09 mm in one hour to 3.33 mm in 4 days.
(3) Individual creep events were separated by periods of inactivity that lasted from 4 to 60 days. In addition, Reid found that different parts of the same fault moved at different rates, and only occasionally could discrete movement events at one site be correlated with observed events at other sites.
The interpretation of these and additional data gathered from Reid’s instruments since mid-1972 is complex and will not be discussed here. We point out only that fault movements are episodic on the scale of hours to days, at least for the three faults monitored by Reid.
Average rates of movement over periods of years can be determined “by dividing the vertical separation developed in a paved surface by the number of years since its surface was constructed or releveled” (Reid, 1973). This has been done by a number of workers (Gray, 1958; Weaver and Sheets, 1962; McClelland Engineers, 1966; Van Siclen, 1967; Reid, 1973). Most active faults surveyed showed average vertical displacements of between 0.3 and 2.3 cm/yr., with a maximum observed rate (Reid, 1973) of 10.3 cm/yr.
Long-term as well as short-term variations in fault movements have been documented, so the vertical separation from year to year at any particular location on a fault is likely to vary. Sections of some faults that have apparently been dormant for years may suddenly become active (illustrated dramatically by faulting at Goose Creek oil field; see account of Pratt and Johnson, 1926), and, conversely, other faults with a long history of damage to manmade structures may stop moving (Van Siclen, 1967). The same fault may be dormant at one location and active at another. There appears to be little assurance that identification of the most active segments of faults today will be of material assistance in estimating fault movements beyond the immediate future.
No special study of movement rates for any of the faults shown on maps 1 and 2 and figure 3 was attempted, but, on the basis of observed damage to structures of known age, vertical offsets of 1 cm/yr or more appear characteristic for much of the Long Point, Piney Point East, Eureka Heights, Memorial Park, Pecore West, and Pecore East faults.3 Movement rates have not been estimated for the Piney Point West fault, but it is active and has damaged structures built on it.
The Clodine fault is problematic, in that few manmade structures more than a few years old have been built across the fault. Opportunity to assess its rate of movement, if indeed it is active, is thus severely limited. The geologist must resort to old aerial photographs and old topographic maps for evidence of previous signs of the fault; he must inspect borrow ditches and irrigation canals for evidence of gradient changes caused by faulting; and he must examine every paved surface that crosses the fault, no matter how new, for signs of fault-related damage. The bits of evidence that we have gathered for the Clodine fault suggest that it is active, at a seemingly low but undetermined rate. We stress, however, that the various lines of evidence, even collectively, are far from diagnostic and present at best only a weak indication that the fault is active.
No manmade structures had been built on the Renn scarp at the time of our investigation; we therefore have no evidence by which to directly assess the possibility of movement.
Relation Between Subsurface Faults and Scarps at Land Surface
Numerous faults beneath the Houston area, at depths of 1,000–4,000 m, have been delineated through study of well logs and seismic lines. It is reasonable to assume, but difficult to prove, that some of these subsurface faults penetrate younger sediments at shallower depths, and that among them are some faults that have offset the present land surface to produce recognizable scarps. Unequivocal demonstrations of this relationship (for example, McClelland Engineers, 1966) are few in the published literature. Among unpublished studies, however, there is ample evidence that many faults recognized at land surface may be traced to appreciable depths; that is, that faults in the Houston area are not purely a surficial phenomenon, but are part of the overall geologic structure of the Texas Gulf Coast.
High-resolution shallow seismic lines provide extraordinarily detailed records of shallow geologic structures, from depths of about 30 to 400 m or more. This method has been used to directly confirm connections between scarps and shallow subsurface faults at several locations in the Houston area (J.B. Farr, oral commun., 1978; Western Geophysical, written commun., 1975). More recently, ship-mounted high-resolution seismic equipment has been used to investigate the shallow subsurface beneath Galveston Bay and adjacent waterways (S.J. Williams, U.S. Army Corps of Engineers, unpub. data; U.S. Geological Survey, Corpus Christi office, unpub. data, 1978). Among the numerous faults visible on these seismic sections are some that are coincident with inferred extensions of known scarps on land only 0.25 to 2 km away. For a limited number of scarps, then, a connection with a subsurface fault has been quite well established.
Careful, bed-to-bed correlation, using electric logs of closely spaced drill holes across known scarps, is a common method of detecting faults in sediments pierced by the drill, In this manner, subsurface faults have been verified beneath scarps at over 30 locations in the Houston area by D.C. Van Siclen (oral commun., 1978) and at several locations near Texas City, south of Houston, by A.J. Langford (oral commun., 1979).
In summary, a connection between a scarp at land surface and a subsurface fault has been verified for more than 40 faults for which sufficiently detailed information is available for the shallow subsurface. Moreover, the subsurface fault is invariably a growth fault* (D.C. Van Siclen, oral commun., 1978), whose definitive characteristic, that of increasing displacement with depth, is generally considered as evidence of a long history of fault movement. For some 40 faults, then, we have evidence that they are natural geologic features, that their histories of movement span tens of thousands to millions of years, and that their associated scarps reflect only the most recent displacements on faults that were active long before the present land surface of the Houston area was formed.
Similarities between surface and subsurface fault patterns and the seemingly close association of many faults mapped at the surface with known subsurface geologic structures, such as salt domes (Pratt and Johnson, 1926; Weaver and Sheets, 1962; McClelland Engineers, 1966; Van Siclen, 1967; Verbeek and Clanton, 1978) provide indirect evidence that the above conclusions may be extended to many other faults as well. It is instructive, for example, to compare maps of faults in southeastern Houston (fig. 1) and northern and western Houston (fig. 7) to each other and to subsurface fault maps of the same areas (McClelland Engineers, 1966). The following facts can be noted:
(1) Faults in southeastern Houston (fig. 1) are characteristically more numerous and more varied in trend than faults in northern and western Houston (fig. 7).
(2) Faults in southeastern Houston are confined to well-defined belts of high fault density. Furthermore, high concentrations of faults are closely associated with underlying salt domes.4 In northern and western Houston, where salt domes are few, faults tend to occur either singly or in pairs (Eureka Heights—Memorial Park; Long Point—Piney Point) that have complementary senses of displacement; there is little tendency to cluster within recognizable belts of high fault density.
(3) A map of subsurface faults, at depths of 1,500–3,000 m (prepared by McClelland Engineers, 1966), mimics the general style of faults mapped at ground level—dense arrays of diversely trending faults characterize the southeastern Houston area, particularly in the vicinity of salt domes, whereas only long, gently curvilinear faults trending east-northeast cut the subsurface sediments of northern and western Houston.
The fact that similar patterns, and variations in patterns, can be recognized in surface and subsurface fault maps has been noted by other authors (McClelland Engineers, 1966; Van Siclen, 1967) and has been cited by them as evidence that scarps at land surface connect with recognized subsurface faults. This evidence must be used cautiously, for the large vertical distances (1,500–3,000 m) between surface and subsurface fault maps render one-to-one correlations between surface and subsurface faults doubtful except under favorable circumstances. Nevertheless, even a cautious interpretation of the evidence would suggest that many, if not most, fault scarps in the Houston area connect downward with growth faults. The evidence does not permit the assumption that all faults in the Houston area may be so regarded, nor does it, by itself, shed much light on the causes of current fault movement.
These general remarks must be extended with caution to faults shown on maps 1 and 2, for we have little substantive evidence concerning subsurface structures associated with most of the mapped scarps. Reid (1973) traced the Long Point and Eureka Heights faults to depths of nearly 3,000 m by use of electric logs, but little subsurface information has been reported for the other faults. All of the faults shown on maps 1 and 2, however, appear to be part of a larger group of east-northeast trending faults, some of which have been studied recently by Van Siclen (1978) in the Addicks and Barker dam areas, north of and adjoining map 1. Of nine features tentatively identified as faults on the basis of surface evidence, and for which subsurface confirmation was sought, seven were identified by Van Siclen as growth faults.
History of Fau1ts in the Houston Area and Causes of Current Fault Movement
The conclusion that some, and probably most, observed fault scarps are the surface manifestations of natural geologic features leaves open the question of what factors promote current fault movements. Certainly natural processes must be considered, as was suggested by Reid (1973) in his discussion of movement on the Long Point and Eureka Heights faults. However, Reid also advocated large-scale pumpage of ground water as an additional large factor in reinitiating or accelerating fault movements. Kreitler (1977) extended this view a step further by emphasizing the importance of fluid withdrawal (ground water and, locally, petroleum) as the prime factor governing fault movement; he suggested that most faults in areas of little fluid production are either inactive or are moving so slowly that few recognizable scarps have formed.
It would be inappropriate in this paper to discuss all aspects of this controversy. Only a few pertinent facts with which any hypothesis of faulting must be consistent are reviewed here:
(1) Some faults are undeniably natural geologic features—that is, they were not inadvertently created by man, even though man’s activities are possibly accelerating or reinitiating movement along them. Scarps due to prehistoric movement along the Long Point and Eureka Heights faults, for example, are readily visible on topographic maps based on surveys completed in 1915–16 (U.S. Geological Survey, Hillendahl sheet, 1915; Houston Heights sheet, 1922). The present land surface and uppermost sediments in the Houston area are geologically very young—their ages are measured in tens of thousands, not millions, of years—and knowledge that they were faulted by a natural process, before large-scale fluid extraction had affected much of the Houston area, suggests the probability that natural fault movement is locally continuing today.
(2) Although some fault scarps are visible on topographic maps of 60 years ago, it is equally important to realize that most are not. Many scarps mapped in the Houston today did not exist, or were present only in subdued form, in 1915–16. Most of the present scarp height for many faults, then, appears to have been produced through fault movements within only the last few decades.
(3) The visibility of faults on aerial photographs of undeveloped areas has increased dramatically since 1930, when large-scale comprehensive coverage was first obtained. The amount of increase is far greater than can be attributed to the superior quality of modern cameras, films, and processing; it is probably attributable instead to recent growth in scarp heights.
(4) If present rates of fault movement (commonly in the range of 0.5–2 cm per year; McClelland Engineers, 1966) were characteristic of the recent geologic past, we should, but do not, see many fault scarps tens of meters high. Present levels of fault activity evidently are far in excess of the prehistoric long-term average.
These facts collectively suggest that some limited faulting of the land surface in the Houston area took place in the recent geologic past, but that most significant offset of the land surface has occurred only within the last 60 years. Ascribing the apparent sharp increase in modern fault activity to fluid withdrawal is tempting, but this conclusion, however attractive, has not been proved. Critical data to test a possible relation between fluid withdrawal and fault-movements for a meaningful number of faults are not presently available. The limited information so far acquired, from tiltbeams* monitoring vertical movements since 1971 on the Long Point and Eureka Heights faults, showed an excellent correspondence between variations in fault displacement with water-level fluctuations in nearby wells for the first 2 years of record (Reid, 1973; Kreitler, 1977). This was considered by Kreitler (1977) as proof of man-induced faulting and as an implied demonstration of the absence of other significant controls. The following 5 years of record, however, showed no such correspondence, casting doubt on proposals that fluid withdrawal is the dominant control of fault movement, at least by the mechanism proposed by Kreitler (Gabrysch and Holzer, 1978). One alternative, that greatly increased rates of natural fault movement are at least partly responsible for the current high level of faulting in the Houston area, may at first seem an untenable suggestion. However, some faults show clear evidence (from well-log data) of episodic movement in the geologic past, and it is conceivable that the normally slow pace of natural fault displacement was punctuated by brief periods of greatly accelerated movements, at rates comparable to those observed today. We cannot discount the possibility that the Houston area is experiencing the effects of another such period.
To summarize, the cumulative weight of evidence would appear to favor fluid withdrawal as the dominant factor in contemporary fault movements, but nearly all of the supporting evidence is based on a temporal coincidence between accelerated fault movements and large-scale production of subsurface fluids. No cause-and-effect relationship has yet been demonstrated.
Knowledge of the effects of fluid withdrawal is pertinent both to plans for future development and utilization of water supplies for the Houston area, and to the oil industry for recovery of petroleum. However, the mechanism(s) whereby fluid withdrawal may affect faulting in the Houston area is not understood, and, if fluid withdrawal does affect fault movements, the relative importance of ground water versus petroleum production remains unassessed. Moreover, knowledge that some fault movements are taking place along ancient and naturally occurring planes of weakness is insufficient assurance that new faults are not being created by stresses associated with fluid withdrawal. Until details of cause-and-effect relationships are understood, prediction of fault movements in the Houston metropolitan area must remain speculative.
2 The October 1978 pipeline explosion near Pearland, which claimed seven lives, was unrelated to movement on any known fault. The nearest fault mapped by Verbeek and Clanton (1978) is about 0.4 km from, and parallel to, the pipeline.
3 Parts of the Long Point and Eureka Heights faults appear to be particularly active; some sections of the Long Point fault have averaged more than 2 cm/yr of vertical offset over the last 20 years.
4 The same is true of other salt domes (Hastings, Manvel, Blue Ridge, Clinton) in the Houston area (Verbeek and Clanton, unpub. data, 1975–79).