Additional Evidence of Karst
The strongest direct evidence of a saturated karst system that will influence flows from WIPP was the penetration of the five caverns below the sinkhole at WIPP-33, located about 2600 ft. west of the LWA. Those caverns totaled 29.5 ft. in height, occurring in the Dewey Lake Red Beds, and in the Forty-Niner and Magenta units of the Rustler (Sandia Labs, 1981, Phillips and Snow, 1998). Because the WIPP-33 sinkhole is the most westerly of four in a chain aligned eastwards 2450 ft. to within 1000 ft. of the LWA, it can be surmised that an extensive conduit underlies the chain. A similar chain of sinkholes is aligned E-W just north of the LWA boundary. WIPP-14 there revealed 74 ft. of residual clay, much gypsum, but no caves, perhaps because the hole was drilled eccentric to the sinkhole formed over collapsed caverns. At both sites, trenching and augering done by Phillips (1987) revealed conical depressions in the near-surface Mescalero Caliche, culminating at a chimney, proving that they are sinkholes, not wind-deflation features. Phillips' investigation of numerous depressions on and near the LWA, many filled with sand, showed that there are several true sinkholes on site, including some that have been observed (e.g., during September, 1984) to conduct storm waters rapidly into the underlying formations. Coincident with several surface depressions in the LWA, Barrows (1983 and 1985) discovered anomalous gravity lows probably due to gypsification of large volumes of anhydrite, or to karst voids. Phillips (1987) showed that the gravity anomalies coincide with solution chimneys through, and downwarping of, the Mescalero Caliche horizon.
Throughout the Rustler formation, there are many instances, on and off-site, of drill holes that encountered dissolution products, including breccia or rock fragments displaced from higher strata, or residual clays derived from the dissolved salt beds and eroded from adjacent clastics. Phillips (1997) and Hill (1999) have summarized the drill data indicative of karst, such as those at WIPP-13, 14, 19, 33, and H-3, and the anomalous drawdowns, likewise consistent with karst. Gypsum replacements of former anhydrite beds have been formed by freshwater recharge (Snyder, 1985). Westward freshening of Rustler waters can best be reconciled with westerly flow and vertical recharge. Furthermore, the youthful ages of some waters sampled from the Rustler can be attributed to rapid, local infiltration (Hill, 1999, pp. 54,55).
DOE has relied on advice contradicting the argument that clays and breccias in the Rustler evaporites are residual materials consequent to salt dissolution. Several geologists, notably Ferrall and Gibbons (1980), Barrows (1982), Chaturvedi and Channel (1985), Lowenstein (1987), Phillips (1997) and Snow (1998) believe that these are residual materials deposited underground at channeled horizons where salt or gypsum formerly occurred in the Rustler. As Figure 3 shows, the formation thickens progressively eastward, consistent with the thicknesses of salt removed by dissolution that has progressed eastward from the edge of the basin. Conversely, the thickness changes could be due, as Holt and Powers (1988) allege, to non-deposition of the salts, but as Lowenstein points out, there are no other stratigraphic or lithologic changes of the Rustler across the site. Clays and channel fillings found by Holt and Powers (1986) in the walls of the Exhaust Shaft. have been attributed by them (1988), and other DOE contractors and consultants, to alluvial fillings of surface channels contemporaneous with (Permian) deposition. This issue is crucial to the controversy because if residuum has replaced portions of the salt formerly in the Rustler, the dissolution process has been pervasive and perhaps continual, and occasional open conduits are an inescapable corollary, even if they are locally obstructed by residuum.
Snow (1998) has described similar detrital deposits partially filling modern dissolution channels in salt over the flooding K-2 potash mine in the Prairie Formation, Saskatchewan, but in beds 3000 feet below the surface. The modern Prairie channels have fill features, such as graded and foreset bedding and particle imbrication that are consistent with both environments, surface and subsurface. Permian salt sequences at the Canadian site are overlain by thick alluvial redbeds of clay and sandy clay, in which slickensides (internal fractures) have all orientations. These are presumably due to shrinkage and swelling by surficial desiccation and wetting, analogous to slickensides observable in swelling clay soils anywhere. But thin lagoonal clay beds imbedded within the Canadian salt sequence have desiccation shrinkage polygons, not slickensides, except where horizontal slickensides occur in places where differential mine subsidence has caused shear. If the Rustler clays were syndepositional with the anhydrites, the clays would display omni-directional arrangements of slickensides, not just coplanar with the beds. The one feature of the Rustler clays logged at the Exhaust Shaft. inconsistent with surface channel deposition is thus the presence of horizontal slickensides. At WIPP, they must have formed millions of years aft.er deposition, as underlying salt beds were locally removed by dissolution, causing deflections. Observed breccias that include rotated blocks of higher-level anhydrites imbedded in clay are incompatible with the low-energy channel environment typical of the evaporitic lagoon. For those reasons, it is evident that Holt and Powers have erred in deducing that the salt beds are missing from the Rustler because of non-deposition. Salt dissolution is, and has long been, active at the WIPP site. If doubt remains, the conservative alternative should be favored.
From hydrological testing of WIPP-area boreholes, there can be derived many indirect lines of evidence for a karstic origin of transmissibility measured in the Culebra dolomite. East of the LWA there is but one borehole (P-18), but it indicates water-tightness, presumably because gypsum fracture fillings remain complete. Westward across the LWA to Nash Draw, the increasing transmissibilities measured in the Culebra are consistent with salt bed removal and subsidence damage to the gypsum-filled fractures of the Culebra (Neill, et. al., 1998, p. 11), fracturing of the brittle anhydrites above and below the Culebra and with solution enlargements and karst conduits that coalesce downstream. Other hydrologic observations are consistent with karst, not matrix properties. These include anomalous drawdowns interpreted as high Culebra transmissibility between certain wells but not others (Phillips and Snow, 1997), or indicating anisotropy (Jones, 1992) related to structure (Barrows, 1982, Phillips and Snow, 1998). Lateral channel interconnections are also implied by equal Culebra heads at some adjacent wells, and vertical channels near some wells are indicated by equal heads in the Culebra and Magenta dolomites (Phillips and Snow, 1997).
Figure 8 (Chaturvedi and Channel, 1985, Plate 1) shows a local solution-enlargement of a vertical fracture in the Unnamed Lower (anhydrite) Member of the Rustler, exposed in the Construction and Salt Handling Shaft.. It proves that recharge has occurred across the Culebra, presumably to a solution channel at the top of the Salado. If not now plugged with residuum, a fracture of such dimensions could convey hundreds of gallons per minute at low gradients. It proves that karst has played an important role in groundwater movement directly over the repository, and must be presumed locally present. Conduits may be plugged at some levels, open and competent to transport escaping contaminated brine at higher levels and at later times.
To its detriment, DOE has shunned those concepts and avoided the necessary research, knowing that proof of karstic conduits would be fatal to the project. There is an affidavit (see Attachment A) from a former employee that Sandia hydrologists were forbidden to use the word 'karst', let alone investigate it. At one meeting attended in 1994 by all of the Sandia hydrologists, the writer personally raised for discussion the question of the significance of karst features. The senior hydrologist replied, "Oh, you mean the WIPP-33 problem", whereupon the subject was dropped. Because objective investigation has been stifled, the public should be skeptical of claims of suitable conditions for disposal.
When storm events recharge a mature karst aquifer, they commonly manifest a rapid rise of the water table. Yet for lack of such observations, water level transients are unknown in the Rustler Formation in the vicinity of WIPP. Not even the WIPP-33 hole has been converted to an observation well, an obvious choice for objective consideration of the implications of its deep, water-filled caverns. Finding that the Culebra dolomite was the only consistently conductive unit, DOE concluded early that it is the only significant aquifer, the Magenta dolomite having secondary importance. The spacings between partially opened or dissolved small-scale fractures in the Culebra are such that practically all test wells intersect those conduits, so the Culebra has appreciable permeability everywhere. Karst caverns are typically spaced hundreds to thousands of feet, so randomly placed holes have negligible probability of encountering them, or of measuring their conductivity. WIPP-33 was placed in a sinkhole, so does not reflect the sparse distribution of the network. Assuming continuous layer-cake hydrostratigraphy, the investigations done in support of Doe's certification application have largely neglected strata other than the Culebra, few wells having been completed at other levels. If open karst channels are most common near or above the water table, but not in the Culebra, they must be sought by drill holes targeting those levels. A slow recovery of water levels in the Culebra following drainage to the Air Intake Shaft. (Figure 6) does not rule out karst conduits at higher levels, unobserved either because they currently lie above the depressed water table, or simply because such conduits were not targeted for well completion. Hydrologic testing has done nothing to characterize the karst system clearly present in some places within the LWA, nor provided representative parameters for modeling. PA calculations were done for steady-state flow and transport through a continuous, low-transmissibility Culebra aquifer, when the karst conduit system, saturated in future at higher levels of the Rustler and Dewey Lake formations, may convey far greater, intermittent fluxes.
The salinity pattern from Culebra water samples (Phillips, 1997, Snow, 1998) is inconsistent with the current southerly gradient beneath the LWA, showing, rather, an irregular but progressive freshening westward, the likely primitive direction of flow in either matrix, fractures or solution conduits. As the water table is restored to higher levels, especially during wetter climates, it may saturate more karst channels, whose high conductivities and orientations will govern the flow directions and velocities, as will an altered gradient. Spring deposits of gypsum in NE1/4, Sec. 15, T22S, R30E, 240 to 280 ft. above the bottom of Nash Draw, that bear fossils of extinct Pleistocene mammals (Bachman, 1980) attest to higher water table levels during wetter climates, and confirm a westerly primitive flow direction. Anderson (2001) recognized that those gypsum spring deposits must reflect a dual hydrologic system. At low flows and water tables of the modern semi-arid climate (about 14 inches/year), the Culebra may be important, but at high flows and water tables of pluvial periods, likely to resume within 10,000 years, a karst system becomes more active. F. M. Phillips (1995) testified to the EPA that wetter climates of the past have increased recharge by at least an order of magnitude.
Many surface features within the LWA indicate karst. Closed topographic depressions have long been interpreted as sinks (Melton, 1934), especially if aligned. Stream channels are virtually absent at WIPP and the few short arroyos terminate at closed depressions. In some, storms produce ponds that disappear within a few days (Phillips, 1987, p. 86). Because wind-blown sand mantles much of the surface and fills most depressions, percolating rainwater is protected from evaporation. Drainage conduits (swallows) beneath the cover have been revealed by trenching (Phillips, 1987). Barrows (1983, 1985) found gravity anomalies within the LWA as well as beneath the WIPP-14 trend of nine sinkholes, close to the north limits of the LWA. Influent fresh water alters anhydrite to gypsum, producing these gravity lows, and the coincident surface depressions reflect its dissolution and the erosion of overlying Dewey Lake Beds.