The Preliminary Report requests review of three questions: 1.) Has there been an acceptable amount of existing data collected by the contractor (Western Kentucky University Center for Cave and Karst Studies (CCKS)), 2.) Are the interpretations of these data reasonable, and 3.) Are proposed assessments and studies of the hydrogeology adequate to demonstrate that groundwatersheds recharging Mammoth Cave National Park will not be effected by recharge from the proposed KTT site?

These questions overlap the eight specific questions by Dr. Nick Crawford as stated in his cover letter for the Preliminary Report dated March 26, 2002. The park's response is organized as answers to the eight questions plus additional general comments.

Question 1: Do you consider the "existing data" itself to be valid and accurate?

Yes, we consider the existing data to be valid and accurate, but only if reviewed and interpreted within their inherent limitations. For example, water levels used to prepare the potentiometric surface were collected over several years at general base-flow conditions and located vertically upon a 7.5-minute topographic map (contour intervals of 10 or 20 feet). These data are as accurate as the methods of the time allowed and any interpretation of these data must be done in proper context, which is a general, regional potentiometric surface rather than a water table in specific, local areas.

Question 2: Are there other existing data that we should consider?

Yes, although the CCKS compiled all known hydrogeologic records there was little discussion of geomorphic, water quality or biological data. The report makes mention of water quality data collected by the Kentucky Division of Water but lacks a statistical analysis. Biological inventories are also mentioned but data and its interpretation are not given. The Kentucky Division of Water has collected various water quality data over the years at Graham Springs (Wilkin's Bluehole), Waterworks Spring and Poorhouse Spring. It is our understanding that a group was contracted to perform a biological survey of the trunk cave streams downstream of the proposed KTT site in 2001, however their specific findings were not presented.

Question 3: Do you consider the interpretation of the existing data as presented in this report to be valid and accurate?

No, although portions of the existing data were interpreted accurately, there are several instances where interpretation extends beyond what can be supported and reasonably deduced from the data. The well float-recorder records presented in Figure 13 are a credible presentation of these7 data - at least for base-flow conditions. The base-flow potentiometric surface should be very accurate as all wells used in this figure were leveled to USGS benchmarks. The interpretation of the flood-level potentiometric surface is misleading. During flood events the potentiometric surface (water table) of a karst aquifer is extremely dynamic, both temporally and spatially. The flood surface on Figure 13 is based on peak levels from each well following the storm, and thus are not the water levels of these wells at any given moment in time. As local gradient reversals can be very short-lived, they will not be revealed in these temporally-coarse float records. One must also be cautioned using all wells on this transect equally. The Hagan and Sanders wells, unlike the other wells, are domestic water supply wells of unknown construction. The Hagen well does not directly intersect the main conduit (and likely the Sanders well does not either), and therefore may not be comparable to wells that are drilled into the cave stream. The Sanders well did not have a float record for the May 1984 flood used to estimate the flood-flow potentiometric surface. Also, since there is no well near the approximated basin divide, it is subjective to extrapolate the potentiometric surface in this important area. Although there are no existing data that could be used to more accurately depict the potentiometric slope, a depiction of confidence intervals in displaying flood-level water table should be attempted or at least stated. As a final point to potentiometric interpretation, we must be mindful that Figure 13 represents a composite of linear transects across two adjacent basins, and not an aerial depiction of regional flood dynamics.

Geological data (stratigraphy and structure) are interpreted in an accurate manner with the exception of the discussion of sinkhole occurrence on pages 24-27. The report down plays the fact that the entire KTT site is located upon the Sinkhole Plain. Although the sinks are shallower within the proposed site, they are nonetheless sinkholes, as this entire area is internally drained. Upon close examination of the relationship shown between base-level groundwater divides and the structure of the Bon Ayre Anticline, we see that the crest on the anticline is as much as 2.5 miles west of the groundwater basin divide. Additionally, only three dye-traces bracket this structural feature and at least one trace completely crosses the nose of the anticline. Using structural maps, which are based upon a geologic contact hundreds of feet below the surface (Chattanooga Shale), assumes that no changes in stratigraphic thickness occur within the section. Additionally, if cave conduits are ultimately controlled by geologic structure, then all cave passages should be orientated down the geologic dip. They of course are not.

The Quinlan and Ray map (1989) is used as a cornerstone throughout this report as evidence that the KTT site will not impact the park's groundwater. What are not mentioned are specific details and assumptions that are part of this map. While this is indeed an excellent map, we must be aware of its limitations. For example, under what flow conditions were the traces conducted that delineate the boundaries? How many traces are used to define the boundary? What is the confidence interval in the potentiometric surface? And it is vital to note that this map was compiled during the 1970's and 1980's, before researchers had access to modern and very sensitive scanning spectrofluorophotometers. These traces were qualified by holding the elutant from a passive dye-receptor up to a back light, or in latter days, through a filter fluorometer. There have been many repeat traces that have shown surprising results when using modern equipment at Mammoth Cave, other karst areas in Kentucky and the Ozarks. This map should be used as a starting point, or a general guide to regional hydrogeology, and not for local, site-specific conclusions.

Appendix IV outlines methodology and study design to assess the likely-hood of rainfall intensities and distributions that would cause a significant storm within the Graham Springs basin without affecting the Turnhole basin (thus possibly aiding to a potential inter-basinal spill-over). The appendix states how one would go about statistically predicting a particular rainfall event. It does not actually model the data nor make a prediction. The approach by Dr. Foster seems reasonable, however, it appears that the report (not Dr. Foster's Appendix IV) already makes the assumption (page 16) that an event of this magnitude and distribution is "hard to imagine" without the model being run. We trust that results form the statistical model will be included in the final report.

Question 4: Is the existing data sufficient to draw conclusions as to potential flow reversal and spillover?

No, due to the limitations of the data as stated in the preceding question. At this time, the existing data are too limited in scope and accuracy to conclude that a flow reversal or spillover does not exist.

On page 16 (Paragraph 4, Line 4) and page 17 (Paragraph 1, Line 8) there are statements stating that a flood crest in excess of 480' MSL on the Barren River in Bowling Green would be needed to trigger a flow reversal of the Graham Springs basin into Turnhole. A stage increase of this magnitude on the Barren River (although it could cause such a reversal) is not necessarily needed to cause flow to seek alternate routes into adjacent basins. This is due to the differences between basic geomorphology of a karst conduit system versus a surface stream. Indeed a free-surface stream would have to rise to a level above a divide to spill over into an adjacent basin. In a surface stream, a flood splays out over natural levees and onto a flood plain to accommodate increased flow. However, in a cave conduit, with its many flow constrictions, water rises until the conduit is pipe-full. With no levee or flood plain, increased flow must pass through a constant cross-sectional area. Water either increases its velocity, and/or is temporarily hydraulically dammed behind restrictions. We have found that in the Turnhole Spring basin, stage increases within the cave streams two to three times that of the Green River. That is, a flood event that would cause a 30-foot rise in the Green produces a 60 to 90 foot rise in Logsdon or Hawkins Rivers. Figure 13 shows that the May 1984 flood increased the Barren by approximately 35 feet, while the float gauge at Wolf Sink increased about 103 feet. This identifies a 100+ foot flood zone within the aquifer.

We must be mindful that the karst aquifer is not a simple collection of currently active conduits. These aquifers have developed over a few million years, leaving behind upper-level passages in preference to newer, lower, more favorable flow routes as base-level master streams incise their valleys. Under flood conditions we have demonstrated in several instances, stage rising up to the point that older, usually dry "inactive" passages resume flow, spilling over low-flow basin divides into adjacent basins. For example, we have demonstrated that a ten-foot rise in Logsdon/Hawkins Rivers (Turnhole Spring basin) will cause an overflow into the adjacent Echo River basin. The explored and surveyed portions of the Mammoth Cave System now overlay portions of six groundwater basins. This ever-changing rearrangement of watersheds is one of the factors that make Mammoth Cave the longest known cave in the world. Over the history of cave/karst aquifer development, water is constantly being "pirated" to new, more efficient flow routes.

Many times these new routes lead into adjacent basins. For example, in the book Karst Hydrology, Concepts from the Mammoth Cave Area (1989, White and White, eds.), there is a chapter discussing subsurface drainage and the evolution of groundwater basins. On page 89, the authors state that there is geomorphic evidence found in Crump Cave (Smiths Grove) that the entire eastern portion of the ancestral Graham Springs basin was pirated by the Turnhole Spring basin. The main trunk of Crump Cave (39 feet wide and at least 49 feet high) is only 88 feet above the base-flow potentiometric surface. This means that a very large, normally abandoned conduit system within the 100+ foot flood zone of the Graham Springs basin proximal to Smiths Grove exists. We know that this normally abandoned trunk of Crump Cave was active during the March 1997 flood event. How far westward does this ancient conduit or its tributaries extend? We believe that the probability of flow reversals over the length of the basal trunk conduit (from the proposed KTT site to the Turnhole Spring basin) is remote. However, until we better understand the geomorphic evolution of these two adjacent basins, we are concerned about the re-activation of higher, older conduits within the flood zone.

Question 5: Based on the existing data, do you believe there is a significant chance of flow reversal and spillover potentially causing groundwater contamination to Mammoth Cave? If yes, why? Please try to evaluate "how significant a risk" you believe it to be.

We believe that the possibility for a spillover or flow reversal from the proposed KTT site and the Turnhole Spring groundwater basin does exist. We also believe that although the probability for a spillover or flow reversal is rather low, additional data and research can be collected and performed which will better assess this possibility.

Question 6: If the existing data is insignificant to evaluate the risk of potentially contaminated groundwater flowing to MCNP, what other reasonable and necessary research is needed that is not already proposed by the CCKS? Please consider the fact that we cannot accurately measure the discharge of Graham Spring, Turnhole Spring or Poorhouse Spring during high discharge and that performing a dye trace during a low year probability flood is not realistic.

Although this preliminary report summarizes and interprets results of regional hydrogeologic investigations, little information was presented from site-specific studies. This void may be due to the lack of, or oversight of existing data. Future research may be focused in at least four main areas: Geomorphologic investigation, statistical evaluation of water quality data, cave aquatic inventory, and dye tracing.

It is highly unlikely that a gradient-reversal within the base-flow conduit system could transfer water from the proposed KTT site to the Turnhole Spring basin. A far more likely scenario would be the re-activation of former flow routes under flood flow conditions. Data show that the karst aquifer flood zone is in excess of 100 feet. As described in the book Karst Hydrology, Concepts from the Mammoth Cave Area (1989, White and White, eds.), much of the eastern portion of Graham Springs basin was once part of the Turnhole Spring basin. This interpretation is based upon geomorphic evidence found in Crump Cave (Smiths Grove). A thorough geomorphic investigation of known caves within this flood zone is needed to determine the aerial extent of overlap between the ancestral Turnhole Spring and Graham Springs basins.

A statistical analysis of existing water quality for Graham Springs should be done. This evaluation must include water quality records from other regional springs in South Central Kentucky, including heavily impacted (Lost River Rise), and public drinking water supply springs for an objective comparison. The report accurately states that the water of Graham Springs is far from pure. However, it must be viewed in proper context with water quality of other karst watersheds, both developed and pristine. For example, how does the water quality of Graham Springs compare with Lost River Rise? How does it compare to water quality of public drinking water supply springs at Auburn and Guthrie Kentucky? Are there any examples of large-scale industrial developments on karst that have not produced groundwater contamination issues?

It is unclear as to the degree of cave aquatic inventories conducted within the Graham Springs basin between the proposed KTT site and the Barren River. We realize that a biological inventory of a cave stream is not a routine task. Stream access and physical restraints limit inventoried stream stretches, and general conditions prove arduous to researchers. We also realize that without a thorough effort, questions will remain as to the completeness of the survey.

An obvious research area would be flood-flow dye tracing, regardless of the CKKC stating in Question 6 that it is not realistic. Such a study, if conducted under the proper flow conditions, would determine if the proposed KTT site does or does not impact the groundwaters of Mammoth Cave. The main problem with a flood-flow tracer study would be waiting for the proper flow condition to occur. An event comparable to the March 1997 or May 1984 would be needed.

Question 7: Do you consider the precautionary steps recommended relating to construction to be sufficient in reducing the risk of regolith collapse sinkholes? If not, what do you recommend.

The question that should be asked is if the recommendations relating to construction is sufficient in eliminating, not reducing, the risk of regolith collapse sinkholes. While the simple reduction of risk may be sufficient in other landuses, in an industrial or airfield setting, there is no room for sinkhole collapse. In addition, the report states (Page 25, Paragraph 2, Line 3) that the site is located upon the Lost River Chert, but upon close inspection of Figure 3, it appears that the site actually extends to areas where the Lost River Chert is some 50 feet beneath the surface. In combination with the statement on Page 25, Paragraph 2, Line 8 that sinkhole collapse is more likely to occur in areas not underlain by the Lost River Chert discloses two facts: 1.) Part of the proposed site lies well above the Lost River Chert, and 2.) Sinkhole collapse upon the Lost River Chert does occur. Recommendations that dynamic-compaction and the use of synthetic liners will reduce regolith sinkhole collapses may be accurate. However, the degree of reduction cannot be stated, nor can the complete elimination of risk be guaranteed. Have these techniques been field-tested?

Question 8: Are our recommendations for synthetic liners, a storm water collection and treatment system, and spill collection system a reasonable approach for reducing risk of groundwater contamination? If not, what do you suggest as additions or alternatives?

The recommendations on liners, storm water collection, and treatment and spill collection are admirable, but not required by law to date. There are absolutely no guarantees that the developers of the KTT site will follow these recommendations and cost estimates for these measures are not included in the budget. The report cites the recent Hartland development as an example of how to build on karst with minimal impact (Page 8, Paragraph 3, Line 1). Although this development has taken many steps in storm water management, it is largely a golf course, a convention center and an upper-tier residential area and not an industrial park. While the use of synthetic liners may reduce the risk of sinkhole collapse, the notion of "soil treatment" of runoff (Page 9, Paragraph 1, Line 4) is not adequate in treatment of runoff from an industrial site. Percolation of groundwaters through the soil may initially "treat" runoff however, soil macro-pores quickly develop and channel water into the karst aquifer as efficiently as bedrock conduits. The karst of South Central Kentucky and Bowling Green has long histories of industrial contamination, costly environmental consequences and expensive clean-up efforts.

General Points of Discussion

Several references in the report tend to equate surface and subsurface streams (Page 10, Paragraph 2, Line 18). Although easy for the layperson to understand, this equation is erroneous and causes serious reasoning flaws in describing karst flow dynamics. Under base-flow conditions, the dendritic and convergent flow-patterns of cave streams can be similar to surface streams. However, cave streams behave much differently than surface streams under flood flow conditions, notably the re-activation of alternate flow paths during times of high stage. On Page 10 the analogy between surface and cave streams is described as a mountain ridge, with water flowing down one side or the other. As discussed below, although this is a convenient parallel, it cannot be validly applied in the assessment of karst conduit flow.

The report uses the terms "up gradient" and "down gradient" (Page 10, Paragraph 2, Line 9). Due to the intense flow dynamics of a karst aquifer we must be certain to qualify gradient directions relative to flow condition and local and regional scales. For example, during base-flow conditions, only the lower, youngest (basal) conduits are active in the gross horizontal transfer of water. Karst gradients under flood flow conditions cannot be thought of being a gentle slope over the length of the basin, but rather a very dynamic "surface" that changes abruptly throughout the conduit system. Under flood-flow conditions local gradient slopes rapidly change as great volumes of water are forced through passage constrictions. During these short-lived events (on the order of hours or days), local gradients may actually reverse as water is temporarily dammed behind a constriction. This does not, however, imply that water within these flooded basal conduits will reverse flow up-basin over great distances on a regional, basin-level scale.

We also must be mindful that "up gradient" does not equal "up hill". One may get the impression that these terms are equal by reading this report. Only in free-surface streams is this true. When a passage becomes pipe-full, the gradient is independent of what is considered to be up-hill or downhill - analogous to domestic plumbing. Water flowing in a confined conduit under pressure will rise to the level of head placed upon it (less frictional forces within the conduit). There are several statements in the report that tend to either confuse the issue, or perhaps make the idea of gradient-reversals appear far-fetched. For example, on pages 13-14, the report summarizes this "hypothesis" as groundwater flowing upstream from the KTT site. There are at least two problems with this statement: 1.) Flood-flow gradients may differ from base-flow gradients, 2.) The most likely mechanism for basin spillover is the resumption of flow within upper, older, and usually dry passages within the karst aquifer's flood zone that transfer groundwater between adjacent basins.