Pathways for Contaminant Transport
Los Alamos National Laboratory (LANL) carried out its Hydrogeologic Workplan activities from 1998 through 2004 to better characterize potential pathways for contaminant transport. The purpose of the workplan was to develop the basis for a sitewide groundwater monitoring plan—to be effective, monitoring wells must intercept the contaminant pathways. As noted earlier in this report, the committee’s study came at the juncture between completion of the workplan activities and development of the sitewide monitoring plan. The committee was asked to review the Interim Facility-wide Groundwater Monitoring Plan that LANL issued during the study period.
This chapter first summarizes LANL’s current understanding of hydrogeologic pathways that may transport contaminants from the sources described in Chapter 3 into the regional groundwater. After this summary, the chapter addresses two sets of questions in the committee’s task statement:
Does the laboratory’s interim groundwater monitoring plan1 follow good scientific practices? Is it adequate to provide for the early identification and response to potential environmental impacts from the laboratory?
Is the scope of groundwater monitoring at the laboratory sufficient to provide data needed for remediation decision making? If not, what data gaps remain, and how can they be filled?
The committee found the short answers to item 1 are a qualified yes and no, respectively. While the interim groundwater monitoring plan generally follows good scientific practices, there are opportunities for improving it. The plan is not adequate to provide early identification of potential contaminant migration with high confidence because LANL’s understanding of pathways for contaminant transport, especially inter-watershed pathways, is not yet adequate to support such confidence. The committee’s short answer to item 2 is a qualified no. Gaps remaining in LANL’s pathway conceptualizations and in the scope of groundwater monitoring at the laboratory are discussed in this chapter.
CONTAMINANT PATHWAYS AND MONITORING
Understanding pathways for aqueous transport of contaminants is necessary for determining the location and mass of contaminants at a given time, predicting their migration throughout the site’s hydrogeologic system, and estimating if and when there might be impacts on regional groundwater. Toward developing a monitoring program, LANL’s understanding of pathways is essential for:
Planning the locations of wells to sample the alluvial groundwater, perched-intermediate groundwater, and the regional aquifer so that the wells are most likely to intercept a contaminant plume;
Determining the well-sampling frequency and types of analyses needed; and
Providing a rationale or model for interpreting the sample results.
As summarized in Chapter 2 and described in this chapter, LANL has developed a broad understanding of the main features of the hydrogeologic environment beneath the Pajarito Plateau (Broxton and Vaniman, 2005; Newman and Robinson, 2005; Newman and Birdsell, 2006; Robinson, 2006; Vaniman, 2006). LANL (2005a, referred
to as the Synthesis Report) is a comprehensive summary of the geologic and hydrologic properties of the site and potentially affected regions nearby. This chapter provides the committee’s perspective and assessment of LANL’s current state of understanding of pathways and ways to build on this understanding to establish a scientifically sound groundwater monitoring program.
VADOSE ZONE FLOW PATHWAYS
LANL has concentrated its efforts on understanding vadose zone pathways that its scientists believe have the greatest potential to impact the regional aquifer in the near term. In addition to presentations to the committee and the Synthesis Report, LANL scientists have published details about the site’s vadose zone pathways in a special edition of the Vadose Zone Journal (2005). Such peer-reviewed publication is the standard of sound science and illustrates the quality of scientific effort LANL has brought to bear on understanding these pathways.
The stratigraphic units of primary interest for vadose zone flow are the Bandelier Tuff, Cerros del Rio basalt, and Puye Formation; see Color Plate 2. Water content distributions in the unsaturated zone (Vaniman et al., 2005), major ion and contaminant transport measurements, numerical models, field measurements at an instrumented site in basalt (Stauffer and Stone, 2005), and field injection tests in the Bandelier Tuff (Robinson et al., 2005a) form the basis for the LANL flow and transport conceptualizations.
Birdsell et al. (2005) summarize LANL’s understanding of vadose zone flow in terms of conceptual models for the four hydrologic regions at the site:
Dry and disturbed mesas, and
Wet canyons, believed to be the origin of most current groundwater contamination, have received by far the greatest amount of study. Other pathways assumed to present lesser or longer-term threats to regional groundwater have received less attention.
Wet Canyon Conceptual Model
Wet canyons are either naturally wet with their headwaters in the mountains (e.g., Cañon de Valle, Los Alamos, and Pueblo Canyons) or anthropogenically wet by discharges from cooling towers or wastewater treatment plants (e.g., Mortandad Canyon, Sandia Canyon).2 The wet canyon conceptual model is the one most developed by the LANL staff, and as such the wet canyons are the focus of most of the groundwater modeling and monitoring efforts. In the committee’s workshop, LANL scientists expressed a consistently high level of confidence in the wet canyon conceptualization.
Mortandad Canyon has been extensively studied, and, in large part, these studies form the basis for the conceptualization applied to all wet canyons at LANL; see Color Plate 7. Mortandad Canyon starts on the dry plateau but is considered a wet canyon because of anthropogenic discharges into the canyon. The radioactive liquid waste treatment facility at Technical Area-50 (TA-50) released treated effluent in excess of 107 L/yr via a small side canyon emptying into the larger Mortandad Canyon. The discharge volume and contaminant mass in the effluent are well documented and, thus, have proved useful for validation of the wet canyon conceptualization.
A key component of the wet canyon conceptual model is relatively large surface water flow volumes, whether natural or anthropogenic. In Mortandad Canyon, treated wastewater effluent is discharged into the canyon, where it mixes with uncontaminated surface water runoff from other locations. The non-sorbing contaminants3 are assumed to be well mixed with the water. To a first approximation, LANL considers this mixture to be a uniform source (“line source”) of water and contaminants to the deeper unsaturated zones (LANL, 2005c). While the assumption of a uniform line source to the deeper zones is a reasonable approximation for its intended purpose, other conceptualizations could include more complicated flowpaths through the intermediate zone.
According to the conceptual model (illustrated by Color Plate 7), surface water, shown as stream runoff, percolates through the alluvium until downward movement is slowed by less permeable Bandelier Tuff, maintaining shallow bodies of perched groundwater within the intermediate zone. Under portions of Pueblo, Los Alamos, Mortandad, and Sandia Canyons, intermediate-perched groundwater occurs in the lower part of the Bandelier Tuff and within the underlying Puye Formations and Cerros del Rio basalt. Two conceptualizations are hypothesized by LANL for infiltration from the canyon bottoms to the regional groundwater. In Mortandad Canyon, it is assumed that infiltration through the tuff units is by matrix flow.4 In contrast, near Otowi-1, at the confluence of Los Alamos and Pueblo Canyons, little or no tuff is present and rapid fracture flow is assumed through the basalts (Birdsell et al., 2005; Kwicklis et al., 2005).5
Some other important contaminants that tend to sorb onto solid materials, such as rocks and soils, are attenuated but are still transported down-canyon, albeit at lower concentrations; see Chapter 3.
Matrix flow refers to uniform flow through a porous medium, as opposed to non-uniform flow, for example through cracks or fractures in consolidated rock.
Flow through the fractured basalts can be both vertical and horizontal. In general, the interiors of thick basalts have a high percentage of high-angle (near-vertical) fractures related to columnar fracture patterns that formed when the basalt cooled at the time of its geologic origin. Fractures of all orientations, including high-angle and low-angle types, frequently occur near the upper and lower margins of the basalts. The basalts are commonly stacked in thick sequences containing a dozen or more flow units. The individual flow units are commonly separated subhorizontal zones of highly porous interflow breccia. Under unsaturated conditions, the rapid transport is thought to occur predominantly as gravity flow through the high-angle fractures and vertically across the interflow breccias. Near-saturated conditions may occur locally in regions with low effective porosities that allow the fractures to carry the groundwater and bypass lower-porosity regions within the basalt.
If surface water does not infiltrate through the alluvium, it will continue to carry contaminants down the canyons. Stormwater can remobilize considerable amounts of sediments and transport both mobile and sorbing species. The contaminants in the canyons are subject to transport by storm flow toward the Rio Grande. Surface runoff, which is an important pathway by which contaminants can be redistributed or transported offsite, is discussed later in this chapter.
Travel time of liquids from waste sources in the wet canyons to the regional groundwater is predicted to be relatively short (LANL, 2003; Nylander et al., 2003). The presence of anthropogenic contaminants in regional groundwater confirms that beneath wet canyons at least some vadose zone pathways have travel times on the order of a few decades (Birdsell et al., 2005; Robinson et al., 2005c). Data suggest vertical transport velocities of up to 9 m/yr (30 ft/yr) in Mortandad Canyon. Laboratory-derived contaminants (tritium, perchlorate) released in liquid effluents in Los Alamos and Pueblo Canyon have reached the regional aquifer and are present in Otowi-1.
In Sandia Canyon a sizeable wetland has flourished downstream of the cooling tower discharge. The sediments are retaining contaminants, such as some metals. The downstream end of the wetland contacts a visible fault, and it is likely that this wetland is providing an aqueous driver to encourage vertical movement of non-sorbing contaminants downward by the mechanisms and pathways described above.
The pathway conceptualization for the wet canyons is the most developed of the conceptualizations presented by LANL, and the interim groundwater monitoring plan relies heavily on this conceptualization. Wells have been sited to monitor the alluvium, and the perched intermediate zone is also monitored to provide early indication of potential regional aquifer contamination. However, the lateral extent and hydrogeologic continuity of intermediate-perched groundwater have not yet been established, and it is not clear where the contamination will impact the regional aquifer. This need for additional information relates directly to LANL’s plans for future site monitoring.
Dry Canyon Conceptual Model
In contrast to wet canyons, dry canyons have smaller catchment areas, infrequent surface flow, and limited or no saturated alluvial aquifers. Anthropogenic sources of water (if present at all) are considered to be small (Birdsell et al., 2005). Travel times from the surface to the regional aquifer are expected to be from hundreds to several thousands of years (Nylander et al., 2003; Birdsell et al., 2005) based primarily on the analysis of chloride data. This assumes that chloride, derived from atmospheric deposition, is concentrated in shallow vadose zone pore water by evapotranspiration (Walvoord and Scanlon, 2004). While relatively little scientific effort has been applied to dry canyons, the committee is in agreement with LANL that the dry canyon conceptualization is adequate under current dry, undisturbed conditions. However, disruptive events, such as the Cerro Grande fire and its aftermath of severe stormwater runoff, can lead to significant mobilization and redistribution of contaminants as noted in Sidebar 4.1 (also see Alvarez and Arends, 2000; LANL, 2005b).
Dry and Disturbed Mesas
Dry mesas are assumed to have annual net infiltration rates ranging from less than 1 mm/yr to 10 mm/yr, with travel times for contaminants migrating from the mesas to the regional aquifer—which lies some 300 meters (1000 feet) beneath the mesa tops in the central part of the plateau—on the order of several hundred to thousands of years (Newman, 1996; Newman et al., 1997; Birdsell et al., 2000; Nylander et al., 2003). The assumed infiltration rates are based on the conceptualization of dry mesas being generally composed of non-welded to moderately welded tuffs with low water content and, thus, matrix-dominated flow. These assumptions may not always be true if the mesa is disturbed, for example by human activities or other geophysical circumstances.
Birdsell et al. (2005) gave several examples that show focusing surface runoff on disturbed mesa tops can result in flux increases up to hundreds of millimeters per year. One example was focused runoff on Mesita del Buey caused by an asphalt pad. A second example is from Frijoles Mesa, where an elevated asphalt pad trapped surface water along its edge. The higher infiltration rates that occurred under these disturbed conditions were estimated to range from 60 to 388 mm/yr. Another potential water source on dry mesas mentioned in public meetings by the State of New Mexico and stakeholders is ponding of precipitation and runoff in disposal pits during the period of time that they remain open.
In the committee’s public meeting in May 2006, LANL presented the concept of a “breathing mesa.” According to this concept, changes in atmospheric pressure move air in
Contaminant Transport by Surface Water
Natural surface water in the Los Alamos area occurs primarily as short-lived stormwater or snowmelt runoff and in short ephemeral streams draining theuplands along the western portion of the Pajarito Plateau. Effluent from LANL and Los Alamos County operations also forms effluent-supported reaches. Natural and effluent-supported “base flow” conditions are the most important for downstream migration of aqueous-phase (dissolved) contaminants. LANL (2004b) describes investigations conducted in Los Alamos and Pueblo canyons of surface water processes that transport contaminants across the site. The report documents observations that historical transport of soluble constituents in surface-water base flow was rapid and may have at times reached the Rio Grande, especially during periods of extended snowmelt runoff or associated with high-volume and persistent effluent releases.
For contaminants that are sorbed to sediment, stormwater and snowmelt runoff are the dominant mechanisms for migration along the canyons. The sediment and sorbed contaminants are entrained in surface-water flow via erosion of channel bed and bank sediments. A pronounced increase in stormwater runoff occurred following the May 2000 Cerro Grande fire due to the burning and widespread elimination of the thick organic layer (duff) on the forest floor. The increase in runoff caused erosion and deposition of a large amount of sediment derived from the burned uplands as well as within canyons on Laboratory property. This hydrologic perturbation is now largely diminished (LANL, 2004c).
SOURCE: Danny Katzman, LANL.
and out of the mesas. Drying is attributed to convective air circulation within the mesas. For liquid phase transport, the hydraulic conductivity of an unsaturated soil is a strong function of water content. Drying of a soil due to the mesa’s “breathing” would result in lower unsaturated hydraulic conductivity and a reduced downward migration rate of contaminants compared to wetter soil. Capillary suction also serves to draw water from wetter to drier soils if the liquid source is persistent. These same conditions will enhance vapor transport of volatile species (Stauffer et al., 2005). The importance of vapor transport in deep vadose profiles varies and strongly depends on soil texture and water content. Because the water contents in the dry mesas are low, vapor transport may be significant.
LANL’s conceptualization of contaminant transport from dry mesas is not as well developed as that for wet canyons. Given the large inventory of wastes disposed of on the mesas, assumptions that underpin the view that contaminants will be relatively immobile need more field and laboratory confirmation. Vapor transport deserves greater study. Wastes disposed of near surface on the mesas can be affected by disruptive events that might occur either by human activities or through natural causes. They can also be affected by anthropogenic activities that lead to ponding or focused runoff.
Mountain-front mesas are classified as naturally wet mesas, with greater precipitation, runoff, and infiltration than the dry mesas. The wet mountain-front mesas have numerous springs, which are rare in the dry mesas or the eastern part of the plateau except where the regional aquifer discharges near the Rio Grande. Mountain-front mesas are likely the dominant recharge zone for the plateau. The upper tuff units along the mountain front are often moderately to strongly welded, resulting in fracturing and minor faulting. Thus, fracturing appears to control spring locations and contaminant distributions in the subsurface near outfalls and wastewater lagoons.
LANL’s conceptual model for contaminant movement on mountain-front mesas includes rapid movement along fractures, but assumes most of the mass is transported through the soil matrix. Very rapid vadose zone flow and transport were shown to occur during a bromide tracer test performed in 1997 in a former high-explosives outfall pond at TA-16, although the majority of the contaminant mass remains close to TA-16 (LANL, 1998b).
While it is generally not feasible to directly monitor fracture flow on a routine basis, additional sampling of the matrix could be used to confirm that the expected mass of contaminants in the matrix can be accounted for. Natural tracers such as chloride and bromide and radioactive species (especially those associated with atmospheric testing of nuclear weapons such as tritium and Cl-36) have been used to identify rapid transport in fractures or faults at some sites. For example, bomb-pulse Cl-36 was used to identify rapid transport in parts of the Department of Energy’s (DOE’s) proposed Yucca Mountain repository for spent fuels and high-level radioactive waste (Wolfsberg et al., 2000).
Potentially Fast Vadose Zone Pathways
From most locations on the LANL site, water percolates slowly through the porous and permeable matrix of subunits of the Bandelier Tuff. The exceptions are fractures, perched water, and the combination of both that can lead to fast pathways. LANL scientists have identified several sites where fracture flow is evident. The complex geology of the site also suggests that funnel flow or perched flow may be important processes in redirecting the groundwater.
When water percolates through the unsaturated soil matrix, predictions of flow and transport are based on Darcy’s law and Richards’ equation and, generally, slow transport is predicted from the typically low values of recharge and
unsaturated hydraulic conductivity in the vadose zones of arid and semi-arid regions. However, there is evidence at LANL that preferential pathways are occurring in the vadose zone. Preferential flow may transport water and contaminants horizontally beyond the watershed in which they were discharged or transport them vertically to the aquifer far sooner than might be predicted based on bulk-media properties and Richards’ equation.
Preferential flow in the vadose zone refers to flow that is locally concentrated with fluxes higher than predicted by Richards’ equation for unsaturated matrix flow. Preferential flow paths include macropore flow resulting from soil fissures, cracks, and fractures (e.g., unstable flow, and funnel flow) (NRC, 2001). Because the term funnel flow is often associated with redirection of unsaturated flow by capillary barriers, one can distinguish between funnel flow and preferential flow resulting from perching of water on finer geologic strata. The complex geology of the site also suggests that funnel flow or perched flow may be important processes in redirecting flow.
Fractures in the Shallow Vadose Zone
Fractures are macropores that are obvious potential pathways for flow and transport but the presence of fractures, in itself, does not imply that they are always active as transport pathways. Near the surface, the characteristics of the source affect the tendency of fractures to transport contaminants away from the source. If the source zone is ponded water and the fractures are exposed at the surface, fracture flow would be expected to occur. If, on the other hand, fractures do not reach the surface or the source is not ponded, fracture flow may not occur.
Fracture flow can occur as film flow, which exhibits behavior not expected in capillary flow (NRC, 2001). Intermittent flow in fractures can also influence the travel depth of a contaminant from the surface. Fractures may increase the depth that liquids penetrate during cyclic infiltration events (Soll and Birdsell, 1998). In this scenario, fractures would fill and liquid would flow to depth during times of heavy infiltration, followed by flow out of the fracture into the matrix afterward. This process then leaves a high water content in the matrix and less capillary drive from the fracture to the matrix. The next large infiltration event would substantially bypass the moist matrix and move deeper before imbibing into the matrix. LANL’s climate consists of high-intensity, seasonal thunderstorms, which could possibly cause this behavior.
Funnel flow occurs in connection with contrasting stratigraphic layers or lenses that are discontinuous; see Figure 4.1. Funnel flow occurs when unsaturated flow is deflected by sloping coarser lenses that act as capillary barriers (Kung, 1993; Ju and Kung, 1997). Water and contaminants are redirected, resulting in preferential flow paths, local increases in water content and therefore local increases
in hydraulic conductivity, and higher downward flux of the percolating water. The pollution potential of sorbing contaminants is higher for funneled flow because percolation rates increase, the time available for degradation is reduced, and the soil matrix in contact with the contaminant is limited, reducing retardation of the contaminant.
Perched groundwater refers to a zone of saturation within an unsaturated region, typically occurring when downward percolation is slowed by a low-permeability barrier. Color Plate 8 illustrates the variety of perched water occurrences on the LANL site, and Color Plate 10 indicates locations of wells that have encountered perched water. Perched groundwater tends to occur more frequently beneath large, wet canyons (Pueblo, Los Alamos, Mortandad, Sandia) than beneath dry mesa tops (Robinson et al., 2005b).
In some cases, perching can slow downward transport if the low-permeability layer is extensive. However, it is as likely that perched water will move contaminants to the regional water table faster and in difficult-to-predict pathways. Perched water that spreads on finer horizontal or sloping layers may move contaminants beyond the boundaries of the watershed where they were originally discharged. The geologic units that lead to the formation of perched water are often discontinuous horizontally. Where a perching horizon ends, water pressures can build up above the perching layer, potentially resulting in preferential fingers moving into and through the underlying finer geologic unit, increasing transport rates to the regional aquifer, as illustrated in Figure 4.2.
When perching occurs on a fractured geologic formation at intermediate depths, the fractures can become active pathways for fast transport. Perched water may result in complex pathways through the intermediate zones between canyons. The recent discoveries of elevated chromium concentrations in wells R-28 in Mortandad Canyon and R-11 in Sandia Canyon (LANL, 2006d) indicate the possibility of lateral movement between canyons facilitated by perched-intermediate groundwater. Although there are several possible sources of the chromium found in well R-28, a likely source is the cooling tower discharge at TA-03 in Sandia Canyon, one canyon to the north, as shown in Color Plate 10. The plate also shows that there has been limited drilling in Sandia Canyon, so relatively little is known about the possible flow directions or perching that could occur there. This demonstrates the importance of identifying potential pathways between watersheds, which may include perched water, and of further investigating them in the intermediate zone. These potential canyon-to-canyon flow pathways are important for the design of a monitoring program.
Fast Contaminant Transport by Surface Water
Surface water is a fast pathway for contaminants released from liquid disposal outfalls to migrate downgradient and either infiltrate the alluvial groundwater at a location that may be distant from its origin or migrate off site toward the Rio Grande. Storms generate large volumes of runoff water that can transport soils and contaminants along the canyons as a result of soil erosion and runoff; see Sidebar 4.1.
Graf (1994) has shown that the plutonium concentrations measured in bedload sediments collected in streams east of the LANL boundary but above the confluence with Pueblo Canyon are two to three orders of magnitude higher (0.19-3.3 pCi/g) than the average background levels of river sediments in the regional river system (0.002 pCi/g). Surface water from Acid, Pueblo, DP, and upper Los Alamos Canyons drains into Pueblo Canyon. Graf calculated that about 10 percent of plutonium deposited in Los Alamos canyon has been transported into the Rio Grande. Yet Graf estimates that the contributions of plutonium from LANL sources to the total annual plutonium flux for the entire river system is small compared to global fallout. Approximately 10 percent of the total annual plutonium flux to the sediment is from LANL operations.
Slow Transport Pathways
Fast pathways and mobile contaminants are the focus of most transport studies because early detections in groundwater are presumed to provide early warning of future groundwater contamination. However, mass balances to discern contaminants currently in the vadose zone are at least as important in forewarning of deeper groundwater contamination, especially when considering long-term monitoring for site stewardship. Although mass balances are inherently difficult to perform for highly heterogeneous media with preferential flowpaths, they can serve to test the validity of adopted conceptual models even when carrying a broad margin of error.
Modeling studies by Robinson et al. (2005c) show that tritium is likely in the vadose zone en route from its source to the regional aquifer. Hexavalent chromium detected near several drinking water supply wells provides another example of the importance of a mass balance to address monitoring and remediation decisions. Estimates of the amount of chromium released range up to 328,000 pounds (LANL, 2006d), much of which is probably in the vadose zone. This situation could hold for many, if not most, other potential groundwater contaminants.
Estimating the mass of contaminants along the entire pathway from source to groundwater from available monitoring data is an important step in ensuring groundwater protection. Additional characterization and monitoring work are clearly indicated in situations where a substantial amount of a contaminant is known to have been released but cannot be accounted for using available data.
REGIONAL AQUIFER PATHWAY CONCEPTUALIZATIONS
The complexity of the regional aquifer is demonstrated by the difficulty in interpreting the results of two tests to measure changes in the level of the regional aquifer (water table) in response to pumping from water supply wells PM-2 and PM-4 (LANL, 2005a). These wells are shown near the center of Color Plate 10. A 25-day test was conducted at water supply well PM-2 at a constant discharge rate of about 4700 L/min (1250 gpm). A number of observation wells (R-wells) installed as part of the hydrogeologic work plan were monitored during the 25-day pumping period and for 25 days thereafter. A second long-term aquifer test was conducted at supply well PM-4 at a constant discharge rate of approximately 5700 L/min (1500 gpm). The pumping interval was 21 days. Water levels were monitored for the 21-day pumping period and an additional 21 days during recovery.
The data from these aquifer tests suggested two competing conceptual models (LANL, 2005a). First, the regional aquifer may be a leaky confined aquifer with leaky units located above a highly conductive layer that is about 260 meters (850 feet) thick.6 A second possible conceptualization is that the regional aquifer appears to behave like a leaky confined system because it contains interbedded layers of alternating high and low hydraulic conductivities that are sandwiched together into a high-yielding zone.
These two conceptualizations lead to very different pictures of how contaminants in the regional aquifer might behave. If there is low connectivity between layers within the aquifer, the contaminants might remain near the top of the regional aquifer and most likely discharge in the springs near the Rio Grande. On the other hand, higher connectivity could result in the contaminants spreading vertically and more likely entering the deep screened intervals of the regional water supply wells.
LANL scientists are aware of the importance of the conceptual model and that the regional aquifer conceptualization will have important implications for the groundwater monitoring program. Even though planned three-dimensional model simulations to further examine aquifer heterogeneity should provide a better interpretation of the aquifer test data, additional hydrogeologic characterization of the regional aquifer is warranted. Geochemical information could also be used to corroborate the aquifer test data. Effective design of a groundwater monitoring system will require an accurate and complete conceptual model of the regional aquifer.
Numerical models combine information on geology, geochemistry, infiltration, regional groundwater fluxes, and waste discharges in a manner that quantifies understanding of the physical/chemical processes and interactions involved in the transport of contaminants. Information gained during the process of model development provides valuable insight on the validity of the conceptualization implemented in the numerical model. Though many “solutions” are possible, comparison of predicted results to actual measurements provides an estimate of the level of understanding of the flow and transport processes moving contaminants away from their initial disposal locations.
Central to the numerical model is the conceptual model. The numerical model quantifies the meaning of the conceptual model, indicates where refinement to the current conceptualization is necessary, and helps to identify where more information could most likely reduce the level of uncertainty in numerical estimates of future conditions.
Chapter 3 introduced the two types of uncertainty (parametric and conceptual) that must be dealt with in any attempt to understand the site’s geohydrology. Handling these uncertainties is one of the biggest challenges in numerical modeling. Explicit evaluations of uncertainty in relation to an important model output, such as estimated concentration or expected travel time, are the most difficult yet the most important elements of scientifically sound modeling practice.
Numerical modeling of the regional aquifer at LANL is fairly recent. The model FEHM, a finite element heat and mass transfer code used to model unsaturated and saturated flow and contaminant transport in porous and fractured media (Zyvoloski et al., 1997), was first applied to regional aquifer modeling in 1998, and a number of related models have been developed since then (LANL, 2005a). Key features of the LANL modeling work include expanded model boundaries to better incorporate regional flow and recharge locations, which, in turn, better accommodate the simulation of the aquifer system under the LANL site. Slightly earlier regional models used for water supply were developed by the U.S. Geological Survey. LANL (2005a) gives a summary of the numerical models used for site modeling. Some of these models have estimated travel times through the vadose zone. Vadose zone predictive model results are typically most sensitive to assumptions regarding infiltration and waste inventory. Alternative conceptual models and infiltration rates are considered.
The committee recognizes that the vadose zone is complex and the exact pathways from source zones to the regional groundwater are unpredictable. However, the more information that LANL can bring to bear on the vadose zone transport pathways, and on the spatial and temporal knowledge of contaminant waste sites, the better LANL can evaluate the effectiveness of a groundwater monitoring system and improve its design.
An example of a good start for the process is in Robinson et al. (2005c). Two- and three-dimensional vadose zone models were developed to incorporate Los Alamos Canyon, DP canyon, Well R-9, and facilities such as the Omega West Reactor. A variety of contaminants, mostly radionuclides, were suspected to have been released into the canyon with a primary source being the Omega West Reactor. The tritium model predicted that, for locations near well Otowi-4, most of the tritium is likely still present in the vadose zone with a small but non-zero concentration predicted to have reached the regional aquifer. Well R-7, located downstream of the tritium contaminants but further upstream of the Los Alamos-DP canyon confluence, was predicted to have no tritium arriving at the water table. The most rapid transport to the water table was predicted at R-9, where the peak concentration of tritium already reached the water table.
Model results show that, within this portion of the Pajarito Plateau, the regional aquifer is most at risk for contamination at locations near or below the confluence of Los Alamos and Pueblo Canyons. Model results showed further that, even for a non-sorbing contaminant such as tritium, the majority of the released mass is still in the vadose zone. A small fraction of the released mass has reached the water table, primarily in locations in the canyon with high infiltration rates or where the Bandelier Tuff is absent. An update to the regional aquifer model is provided by Keating et al. who state that “predicted flux through older basalts in the aquifer can vary by a factor of three … the true uncertainty of our predictions, including the impact of possible conceptual errors, is likely to be larger and is difficult to quantify” (Keating et al., 2005, p. 653).
The modeling by Robinson et al. (2005c) and Keating et al. (2005) demonstrates that a comprehensive understanding of vadose zone transport processes depends on integrating data from geologic, hydrologic, and site characterization studies with uncertainty analyses. More generally, these LANL scientists have demonstrated that modeling and site characterization studies are important to selecting well locations and sampling frequency as part of the design of an effective monitoring system.
In the August workshop, Vesselinov and Birdsell described a stepped coupled modeling approach that will be applied to the wet canyons. Point sources are simplified into uniformly distributed unit sources along alluvial canyon bottoms, consistent with LANL’s wet canyon conceptual model. Twenty-one potential source configurations have been studied so far, with the travel time through the intermediate zones assumed to be instantaneous. These types of modeling exercises have the potential to directly link the wet canyon conceptual model with the regional groundwater monitoring program.
Modeling at LANL is appropriately incorporating important features of the vadose and saturated zone: matrix flow, fracture flow, varying stratigraphy, and hydrogeologic properties. Important to this effort will be to maintain
a balance between the level of modeling sophistication already available and the understanding of the actual site hydrology. This will be particularly important in incorporation of uncertainty where it quite often happens that the non-modeled uncertainty (conceptual uncertainties about the actual site conditions not reflected in the model’s equations) can outweigh the uncertainty in parameters included in the model. Overlooking conceptual, non-modeled, uncertainties can lead to results that give an overly optimistic perception of the current state of knowledge about present and future groundwater contamination.
EVALUATION OF THE INTERIM GROUNDWATER MONITORING PLAN
The committee was asked to evaluate the interim groundwater monitoring plan developed by LANL. Specifically, two questions were posed in the committee’s task statement: Does the plan follow good scientific practices; and is it adequate to provide for the early identification and response to potential environmental impacts from the laboratory? As noted previously, the short answer to the first question is a qualified yes, while the answer to the second question is no.
The Interim Facility-wide Groundwater Monitoring Plan (LANL, 2006c) states that the purpose of monitoring is to:
Determine the fate and transport of known legacy-waste contaminants,
Detect new releases,
Determine efficacies of remedies, and
Validate proposed corrective measures.
The Interim Plan notes that groundwater monitoring at the site was started in 1945 by the U.S. Geological Survey. The first monitoring network consisted of water supply wells, observation wells, and springs. Early monitoring was primarily from shallow alluvial groundwater. Twenty-five deep wells into the regional aquifer and six intermediate-zone wells were added under the Hydrogeologic Workplan between 1998 and 2004 (LANL, 2005a).
The Interim Plan is intended to monitor the seven main watersheds on the site: Los Alamos, Sandia, Mortandad, Pajarito, Water/Cañon de Valle, Ancho/Chaquehui/Frijoles, and White Rock. The major canyons that define these watersheds are shown in Color Plates 9 and 10.
In the Interim Plan, a table for each watershed presents the rationale for each well in that watershed. The design of the interim monitoring network is stated to be “based on conceptual models of potential sources, hydrogeologic pathways, and receptors” (LANL, 2006c, pp. 1-2). The division of monitoring into the following four modes is consistent with LANL’s pathways conceptualizations:
Base flow—persistent surface water that is maintained by precipitation, snowmelt, effluent, and other sources;
Alluvial groundwater—water within the alluvium in the bottoms of the canyons;
Intermediate-perched groundwater—localized saturated zones within the vadose zone; and
Regional groundwater—the deep, laterally continuous groundwater beneath the Pajarito Plateau.
The Interim Plan is responsive to the Consent Order (see Chapter 2), which is the regulatory driver that the plan addresses, and which specifies much of the structure, choice of locations, and sampling frequency set forth in the Interim Plan. For example, Table XII-5 of the Consent Order includes a listing of wells that must be included in the Interim Plan. The Interim Plan also states that it was based in part on guidance for monitoring network design published by the Environmental Protection Agency, and in particular, on Office of Solid Waste and Emergency Response Directive No. 9355.4-28, “Guidance for Monitoring at Hazardous Waste Sites: Framework for Monitoring Plan Development and Implementation” (EPA, 2004b).
The committee found that the Interim Plan follows good scientific practice in several respects. The report includes discussion of potential contamination sources and the media being monitored under the plan, i.e., stream base flows, alluvial groundwater, intermediate-perched groundwater, and the regional groundwater aquifer. The choice of monitoring locations by LANL appears to have been made using the hydrogeologic approach (Minsker, 2003), based on the use of expert judgment for selection. The reasons for those choices are presented in the monitoring plan tables provided for each watershed. This is especially important when the choice of sampling locations or frequency differs from the locations or frequency specified in the Consent Order.
However, there are areas where the Interim Plan does not appear to follow good scientific practice. The most important of these is the focus on a watershed approach, where the monitoring plan for each watershed within LANL is developed and laid out individually in the Interim Plan. This structure, which is specified in the Consent Order, works quite well for monitoring surface base flows and alluvial groundwater that are confined to the canyons. However, it does not work well for the intermediate aquifers and even less for the regional aquifer. For example, in the discussion of the monitoring plan for Mortandad Canyon in Part 4 of the Interim Plan, the potential contaminant sources that are discussed are only those that fall within the Mortandad Canyon watershed.
As pointed out in the chromium workplan7 (LANL,
The chromium workplan was developed following the discovery of unexpectedly high levels of chromium in some wells in Mortandad Canyon; see Sidebar 3.3. The chromium workplan lays out further investigations to determine the extent of contamination for planning possible remediation actions.
2006d), the source of high concentrations of chromium recently found in Mortandad Canyon does not appear to be within that canyon, but from the use of chromium in large amounts as a corrosion inhibitor at power plants in Sandia and Los Alamos Canyons, one or two canyons to the north. This finding suggests that a canyon-based approach to development of monitoring plans for the intermediate and regional aquifers is not sound.
Minsker (2003) and EPA (2006) document quantitative methods for optimizing a monitoring network, which might be used by LANL and the New Mexico Environment Department (NMED) for improving future monitoring plans to (1) optimize the monitoring network and (2) better incorporate uncertainty into its design. Approaches that incorporate uncertainty are published (e.g., Neuman and Wierenga, 2003) and may also prove useful for application at LANL. The selection and application of any approach should be balanced by the level of knowledge and quality of data available. The main elements of an uncertainty analysis would involve the development and evaluation of alternative conceptual models for the transport of contaminants from identified sources to receptors. The alternative conceptual models might include differences in assumed transport pathways (i.e., alternative models of the hydrogeology and geochemistry), forcing conditions (e.g., input and boundary conditions), and numerical modeling approaches (Neuman and Wierenga, 2003).
For LANL, these alternative conceptual models might be used to address uncertainty in the source terms and the uncertainty in flowpaths from the sources to the regional aquifer. The alternative conceptual models can be evaluated by their ability to reproduce system behavior (e.g., contaminant plume concentrations) using calibration and inverse analysis. Predicted plumes resulting from those alternative conceptual models could then be used to evaluate the probability that the plumes would be intercepted by monitoring wells before moving off the LANL site or reaching a municipal well. Optimization approaches (e.g., Reed et al., 2000) could be used with alternative plume models to design the regional aquifer monitoring network to minimize the probability that a plume would be missed.
Plans for such an approach were identified in the Decision Analysis for Addressing Groundwater Contaminants from the Radioactive Liquid Waste Treatment Facility Released into Mortandad Canyon (LANL, 2005c). That analysis incorporated alternative conceptual models and uncertainty analysis. However, as the report points out, the current version of the decision analysis approach developed by LANL cannot be used for groundwater monitoring network design (LANL, 2005c, Section 5.2.2). The presentation on the LANL Decision Support Process (LDSP) by Chris EchoHawk at the committee’s August meeting suggested that one of the goals of the LDSP is to continue to develop the approach so that it could be used for monitoring network design (EchoHawk, 2006). The use of such an approach would require negotiation with NMED.
Even without a quantitative analysis of the sample locations in the intermediate and regional aquifers, the committee noted several modifications that could be made to the current monitoring network. Given the tendency for regional aquifer monitoring wells to be located in canyon bottoms, large portions of the intermediate and regional aquifers, namely, the portions beneath the mesas, are not monitored given the current monitoring plans and approach. This makes it far less likely that the current monitoring plan will provide early identification and response to potential environmental impacts from the Laboratory. Although the committee understands that there are strong economic and drilling incentives to locate regional monitoring wells in the canyons, and a number of additional monitoring locations could be placed in canyon bottoms that would contribute significantly to the existing network, eventually a way must be found to increase the area of the intermediate and regional aquifers that are monitored. This may require locating some deep monitoring wells on mesa tops, and/or the drilling of slant holes from canyon bottoms to monitor the regional aquifer beneath the mesas.
In looking at the regional monitoring network, the committee found that the southern portion of LANL is one area of the regional aquifer that is currently very sparsely monitored (see Color Plate 10). The committee assumes that this is mostly due to the general southward progression of the canyon investigation plans, and that the area will receive additional deep monitoring wells when the canyon investigation process advances to the southern canyons (Ancho, Chaquehui, and Frijoles Canyons). Another area that appears to be undersampled is the Pueblo de San Ildefonso to the east of LANL, which is generally downgradient from the site. Plans to install monitoring wells on Pueblo lands under the Memorandum of Understanding8 described in Section 3 of LANL (2006a) are a step in the right direction. Additional monitoring to ensure early detection of contaminant plumes beneath these Pueblo lands will likely be required.
There were other parts of the Interim Plan where the committee deemed that additional information is needed. One suggestion would be to broaden the overview of geology and hydrogeology in the main text of the Interim Plan (Section 1.10). The current overview is brief and does not include any graphics to orient the reader to the geology. A good example of what might be provided can be found in Section 5.B of LANL’s Environmental Surveillance Report (LANL, 2006h).
Regarding revisions of the monitoring plan, the section on integration (Section 1.6) states:
The Interim Plan will be updated annually to incorporate new information collected within a watershed. Locations, ana-
lytes, and sampling frequencies will be evaluated and updated appropriately to ensure adequate monitoring and avoid unreasonable budgetary expenditures. Information gained through characterization efforts, aquifer test results, optimization iteration models, and water quality data will be used to refine a long-term monitoring plan for each watershed.
However, no information is provided in the Interim Plan on how this aspect of the integration and revision of the monitoring plan is accomplished. A brief summary in Section 1.6 could describe the ways in which the information from related studies is used for updating the monitoring plan. More importantly, a discussion describing the changes that were actually made to the monitoring plan in response to investigations (e.g., additional drilling, sampling, aquifer testing) that were completed in the previous year could be included in revisions of the plan.
There is little to no information provided in Appendix A of the Interim Plan, or in the body of the plan, on pathways by which the contaminants are moving, which is a critical part of a conceptual model. Inclusion of graphics documenting the conceptual models would also be useful. For example, a cross section along each canyon (or the main canyon when multiple canyons are addressed) would help provide some perspective on the geology of the canyons; see for example Color Plate 7. The cross section could be used to highlight some of the potential flow paths.
GEOPHYSICAL METHODS FOR SITE CHARACTERIZATION AND MONITORING
Challenges in LANL’s groundwater protection program include understanding hydrogeological pathways in the vadose zone and monitoring large areas of the site, as described in this chapter. Well emplacements, as described in Chapter 5, are expensive and sample only limited areas around the borehole. Modern geophysical methods can at least supplement characterization and monitoring data obtained directly from well emplacements.
A previous National Academies study (NRC, 2005) described environmental monitoring at DOE sites as relying heavily on sampling and analyzing groundwater and noted that this practice provides data primarily for the individual locations that are sampled. Geophysical methods can provide continuous measurements in both time and space that can help fill gaps in understanding the subsurface hydrogeology between well locations and enable mapping of large subsurface areas. The report suggested that modern, non-invasive geophysical sensor techniques such as electromagnetic and electrical resistivity methods, seismic reflectivity, and ground-penetrating radar can substantially improve on direct sampling and lead to cost-effective long-term monitoring after site closure.
LANL’s presentations focused on well emplacements for characterization and monitoring; geophysical methods apart from borehole logging were not discussed. However, work at other DOE sites has shown that these methods are promising and improving rapidly, largely due to refined signal processing techniques and statistical methods for data analysis. An evaluation of geophysical technologies applicable to Hanford site characterization was recently completed (Fluor, 2006). Geophysical sensor technology developed at the Idaho National Laboratory is being used to monitor a waste storage area located at the Gilt Edge Mine Superfund site in South Dakota (Versteeg et al., 2004; Versteeg, 2005).
Development and greater use of geophysical methods are fertile opportunities for applying new science and technology to improve the effectiveness of LANL’s groundwater protection program and for increasing cooperation among DOE sites to address common site cleanup and remediation challenges.
FINDINGS AND RECOMMENDATIONS ON PATHWAYS
The committee found that the Laboratory’s current (i.e., interim) monitoring plan generally follows good scientific practices, but there are opportunities for improving it. The plan is not adequate to provide early identification of potential contaminant migration with high confidence because LANL’s understanding of pathways for contaminant transport, especially inter-watershed pathways, is not yet adequate to support such confidence.
The committee concurs with LANL’s approach, which is to characterize and understand potential pathways for contaminant transport in order to support the planning and implementation of a long-term sitewide monitoring program. The committee judged LANL’s current understanding of transport pathways adequate to begin this planning and implementation process. This current understanding can, and should, be improved to ensure groundwater protection in the coming decades and centuries.
The scientific framework used by LANL to categorize the main features of mesas and canyons important to understanding groundwater flow and transport processes is well reasoned and is commended by the committee. Conceptual models for vadose and groundwater flows currently go beyond simple conceptualizations of a qualitative nature. LANL scientists show a good understanding of the suite of possible conceptualizations for various scenarios, depending on source location, contaminant properties, contaminant loading, and source type. The committee encourages continuation of this line of investigation as it is an excellent example of the creativity required to address the Type B uncertainty described in Chapter 3. This framework represents an excellent start to establishing a sitewide monitoring plan that will provide early identification of contaminant migration, support remediation decisions, and eventually transition into long-term monitoring for stewardship.
Detailed Findings and Recommendations
Findings and recommendations to assist LANL in addressing remaining gaps in pathway conceptualizations and improving its monitoring plans are as follows:
The current conceptualization of the LANL flow system into alluvial, intermediate-perched, and regional components, along with their importance to understanding the flow system within and below wet canyons, is a major accomplishment by LANL scientists. However, there is a lack of understanding of the interconnectedness of pathways between basins. While there is a general understanding that perched waters are probably redirecting contaminants from areas directly below canyons, where they originally infiltrate, to submesa areas and to other nearby canyons, the detailed knowledge needed to predict subsurface flowpaths does not exist. Lack of understanding of these phenomena, coupled with rapid flow in the alluvium and apparent rapid flow facilitated by perched waters, was central to the surprise over detection of chromium near the water supply wells. An improved knowledge of these inter-watershed processes is needed to design an effective, early warning monitoring program.
Recommendation: LANL should add a sitewide perspective to its future groundwater monitoring plans. This perspective would include the following:
Design additional characterization, modeling, and geochemical investigations to better understand potential fast pathways between watersheds.
Increase the area of the regional aquifer that is monitored by sampling inter-canyon areas from mesas or using directional wells from canyon bottoms.
Provide additional monitoring locations in the southern area of the site and on Pueblo de San Ildefonso lands.
Develop more applications of geophysical techniques to supplement information provided by well drilling and sampling, especially for understanding vadose zone pathways.
As LANL’s site characterization and monitoring programs mature, well locations should be derived from a quantitative spatial analysis of monitoring well locations to identify areas with the greatest uncertainty in plume concentrations, using geostatistics or other methods, possibly coupled with flow and transport modeling.
Mathematical models are essential tools for both codifying current knowledge and identifying knowledge gaps. Although LANL is using a numerically sophisticated multiphase model for vadose and regional groundwater modeling, it is not yet possible to predict with confidence when, where, or if a contaminant might appear in the regional aquifer. This is due largely to an exceptionally complex vadose zone. Studies show that most of the mass of many contaminants is likely still in the vadose zone on the way down from the release location to the regional aquifer.
Recommendation: LANL should increase its efforts to develop and use quantitative methods to describe contaminant pathways through the vadose zone and into the regional aquifer, as follows:
Mathematical models that incorporate the uncertainties from alternative conceptual models should underpin plans for design and operation of the sitewide monitoring system. Characterization of the vadose zone begun under the Hydrogeologic Workplan should continue with emphasis on new results from characterization and monitoring being used to test and improve the mathematical models.
To support an evaluation of the effectiveness of the monitoring system to provide early warning of potential impacts on the regional aquifer, LANL should quantify, to the extent possible, the inventory and current location of the contaminants disposed of in the major waste sites.
Large waste disposal sites in the dry canyons and on dry mesas have not received as much attention as wet canyons and wet mesas because they presumably lack an aqueous driver to move contamination. The presumed dry locations have received minimal characterization with regards to the presence, strength, and potential impact of aqueous drivers. In some of these, surface disturbances have led to unexpected increased infiltration rates. LANL provided few data to justify assumptions about the relative immobility of wastes at these sites.
Recommendations: LANL should confirm the integrity (lack of surface disturbances or conditions leading to increased infiltration) of the major disposal sites in the dry canyons and mesas.
LANL should schedule regular subsurface surveillance beneath disposed wastes on dry mesas and in dry canyons.
LANL’s present conceptualizations of the regional aquifer lead to very different pictures of how contaminants in the aquifer might behave. If there is low connectivity between layers within the aquifer, the contaminants might remain near the top of the regional aquifer and most likely discharge in the springs near the Rio Grande. On the other hand, higher connectivity could result in the contaminants spreading vertically and more likely
entering the deep screened intervals of regional water supply wells.
Recommendation: LANL should continue efforts begun under the Hydrogeologic Workplan to characterize the regional aquifer. More large-scale pumping tests and improved analyses of the drawdown data are needed to establish a scientifically defensible conceptual model of the aquifer, i.e., leaky-confined, unconfined, or layered.
LANL’s efforts to understand the role of geochemistry in contaminant migration have not kept pace with efforts to understand hydrology. The committee found a lack of basic, site-specific geochemical data to support LANL’s assumptions about the relative immobility of important contaminants—especially radionuclides—along transport pathways and judged that LANL underestimated the value of both field and laboratory geochemical measurements.
Recommendation: LANL should increase its attention to geochemistry within the context of its site characterization work. LANL scientists should conduct more field and laboratory studies to measure basic geochemical parameters such as sorption coefficients with the goal of testing and verifying their conceptualizations of subsurface hydrogeochemical processes.
The following finding and recommendations reflect the committee’s evaluation of the Interim Facility-wide Groundwater Monitoring Plan (LANL 2006c), which was requested in the statement of task.9
The Hydrogeologic Workplan has been effective in improving characterization of the site’s hydrogeology. However, the knowledge gained through the workplan does not appear to have been used effectively in the development of the interim plan. The workplan is mentioned only in the introduction of the interim plan, and rationale for the siting of new wells in the interim plan is not grounded in the scientific understanding of the site evident in the Synthesis Report and other publications such as the Vadose Zone Journal (2005).
Recommendations: LANL should demonstrate better use of its current understanding of contaminant transport pathways in the design of its groundwater monitoring program. Tables in the monitoring plan that give the rationale for locating monitoring wells should at least provide a general linkage between the proposed locations and the site’s hydrology, or a section discussing the relation between well locations and pathway conceptualizations should be added.
LANL should take a sitewide approach to monitoring of the intermediate and regional aquifers. Furthermore, the interim plan should summarize (e.g., in Section 1.6) the ways in which the information from related studies will be used for updating the interim plan. The current description of the conceptual models (in Appendix A of the plan) is useful, but it should be improved. First and foremost would be a description of potential pathways, both surface and subsurface, that connect the sources (listed in Appendix A) with the groundwater that is being monitored.
LANL should examine the potential for approaches (Minsker, 2003; EPA, 2006) that both optimize the monitoring network and incorporate uncertainty into its design.