This section describes the current three-dimensional conceptual model of the Hanford Site unconfined aquifer system, which is the uppermost aquifer across most of the Site. The conceptual model was constructed from data on the ground-water flow system and inferences that can be drawn from the data. For example, water level measurements in wells are a part of the available data and directions of ground-water flow can be inferred from these data. The conceptual model was also developed from information on the hydrogeologic structure of the aquifer, spatial distributions of hydraulic and transport properties, aquifer boundary conditions, and the distribution and movement of contaminants. Separate annual reports document the observed movement of chemical and radiological contaminant plumes in Hanford Site ground water and the measured elevations of the water table across the Site (for example, see Dresel et al. 1995). Earlier status reports on development of the three-dimensional conceptual model are presented in Thorne and Chamness (1992), and Thorne et al. (1993, 1994).
The Hanford Site and adjacent areas north and east of the Columbia River lie within the Pasco Basin, a structural depression that has accumulated a relatively thick sequence of fluvial, lacustrine, and glaciofluvial sediments. Figure 2.1 shows the surface geology and major structural features of the area. Hanford Site geology and hydrology have been studied extensively for approximately 50 years. Detailed summaries are provided in DOE (1988), Delaney et al. (1991), Lindsey et al. (1992), Lindsey (1995), and Cushing (1995). Consequently, the general geology and hydrology of the Site is only briefly covered here.
The Pasco Basin and nearby anticlines and synclines initially developed in the underlying Columbia River Basalt Group, a sequence of continental flood basalts covering more than 160,000 km2 (DOE 1988). These basalt flows erupted as fluid, molten lava during the late Tertiary Period. The most recent, laterally extensive basalt flow underlying the Hanford Site is the Elephant Mountain Member of the Saddle Mountains Basalt Formation, although the even younger Ice Harbor Member is found in the southern part of the Site (DOE 1988). Sandwiched between various basalt flows are sedimentary interbeds collectively called the Ellensburg Formation. The Ellensburg Formation includes fluvial and lacustrine sediments consisting of mud, sand, and gravel which, along with the porous basalt flow tops and bottoms, form confined basalt aquifers across the basin. The Rattlesnake Ridge Interbed is the uppermost laterally extensive interbed and confined basalt aquifer of the Ellensburg Formation (Spane and Vermeul 1994).
Overlying the basalt within the Pasco Basin are fluvial and lacustrine sediments of the Ringold Formation (Newcomb and Strand 1953; DOE 1988; Lindsey et al. 1992). The ancestral Columbia River and its tributaries flowed into the Pasco Basin, depositing coarse-grained sediments in the migrating river channels and fine-grained sediments (silt and clay) in the overbank flood deposits. On at least two occasions, these river channels were blocked, forming a lake in the Pasco Basin and depositing extensive layers of fine-grained sediments within the Ringold Formation. The Plio-Pleistocene unit, consisting of a paleosol/calcrete and/or basaltic sidestream sediments, and the early "Palouse" soil, an eolian sand and silt deposit, overlie the Ringold Formation, but are present only in the western portion of the Pasco Basin. The uppermost sedimentary unit covering much of the Hanford Site is the Hanford formation, a complex series of coarse- and fine-grained sediments deposited by cataclysmic floods (called the Missoula floods) during the last ice age. For the most part, the fine-grained sediments are found near the margins of the basin and in areas protected from the main flood currents, which deposited the coarse-grained sediments. Capping the Hanford formation in many areas is a thin veneer of eolian sands and/or recent fluvial deposits.
The fine-grained layers of the Ringold Formation have a much lower permeability than the coarse-grained layers, forming aquitards. However, these aquitards are usually not continuous across the Hanford Site, allowing interflow between different parts of the suprabasalt aquifer on a sitewide scale. Consequently, the suprabasalt aquifer is considered one entity commonly referred to as the "Hanford unconfined aquifer system."
As the post-basalt sediments were being deposited, the Pasco Basin continued to undergo structural deformation (DOE 1988). The basin continued to subside, and the ridges continued to rise. This process caused sedimentary units to be thickest in the center of the basin and thin or, in places, pinch out along the anticlines. Hanford formation sediments directly overlie the basalt in a few places where the Ringold Formation either was never deposited or was eroded away by the ancestral Columbia River and its tributaries prior to the Missoula floods. Missoula floodwaters further eroded sediment and basalt in some areas. Two known vertical faults, the Cold Creek and May Junction faults, were also developing as the older Ringold sediments were being deposited. Faulting is thought to have occurred until middle Ringold time, with a maximum vertical offset of 150 m; there is no evidence of activity on these faults since that time (Johnson et al. 1993). The Cold Creek fault is known to affect hydraulic heads in the confined basalt aquifers; however, it is not clear if the unconfined aquifer is also affected.
The unconfined aquifer and a sequence of confined aquifers lie beneath most of the Hanford Site. The unconfined aquifer is generally located in the unconsolidated to semiconsolidated Ringold and Hanford formation sediments that overlie the basalt bedrock. Where it is below the water table, the coarse-grained Hanford formation makes up the most permeable zones of the unconfined aquifer system. The basalt confined aquifers are composed of the brecciated tops of basalt flows and sedimentary interbeds located between basalt flows of the Columbia River Basalt Group.
The saturated thickness of the unconfined aquifer on the Hanford Site is greater than 61 m in some areas but pinches out along the flanks of the basalt ridges. Depth to the water table ranges from less than 0.3 m near the Columbia River to more than 106 m near the 200 Areas. Perched water-table conditions have been encountered in sediments above the unconfined aquifer in the 200-West Area (Airhart 1990; Last and Rohay 1993) and in irrigated offsite areas east of the Columbia River (Brown 1979).
Ground water in the unconfined aquifer at Hanford generally flows from recharge areas in the elevated regions near the western boundary of the Hanford Site toward the Columbia River. The Columbia River is a discharge zone for the unconfined aquifer on both sides of the Columbia River. The Yakima River lies southwest of the Hanford Site and is generally regarded as a source of recharge to the unconfined aquifer in the southern part of the Site and in the Richland area. Areal recharge from precipitation falling on the Hanford Site is highly variable both spatially and temporally depending on climate, soil type, and vegetation.
The unconfined aquifer system at the Hanford Site is bounded by the Columbia River on the north and east, and by the Yakima River and basalt ridges on the south and west. These physical flow system boundaries have been defined as either prescribed head or no-flow boundaries in the numerical model. At the Cold Creek and Dry Creek Valleys, the unconfined aquifer extends westward beyond the boundary of the Hanford Site ground-water flow model. Boundaries have been defined across these valleys. An arbitrary boundary has also been defined between Umtanum Ridge and the Columbia River in the northwest corner of the Site. The upper boundary of the unconfined aquifer is usually the water table, which changes position over time. However, the aquifer may be locally confined by fine-grained sediments in a few areas. Flow through the upper boundary includes both natural areal recharge from precipitation and local recharge from liquid waste disposal, irrigation, and artificial recharge activities. Discharge through wells is minor and has not been included in the numerical model. At the Richland city well field there is actually a net recharge because of the input of Columbia River water to recharge basins. The bottom of the unconfined aquifer system is generally defined as the top of basalt. Any recharge or discharge through this boundary is a result of interflow with the underlying confined aquifer system.
The Columbia River flows along the northern and eastern boundaries of the modeled portion of the Hanford Site. Ground water in both the unconfined and confined aquifer systems generally flows toward the river, which is the major discharge area within the Pasco Basin.
The current modeling approach is to represent the Columbia River as a prescribed-head boundary over the depth of the river and as a no-flow boundary from the bottom of the river to the bottom of the aquifer. It is unlikely that ground water in the unconfined aquifer system flows across this boundary because the river is the regional discharge. However, flow across this boundary is possible if a locally confined permeable unit extends beneath the river and is affected by stresses such as pumping. Definitions of hydrogeologic units in the conceptual model are being extended across the river to allow for possible simulations of such a scenario or other scenarios where local flow may pass under the river before discharging to it.
Water levels in many wells near the Columbia River fluctuate in response to changes in river stage. The river stage generally rises and falls daily because releases from upstream dams can change local river levels by up to 3 m within a few hours. Seasonal changes of about the same magnitude are also observed. River stage fluctuations measured at the 300 Area are only about half the magnitude of those measured near the 100 Areas because of the effect of the pool behind McNary Dam, located downstream from the Hanford Site (Campbell et al. 1993). Changes in water-table elevation near the river result primarily from pressure waves transmitted through the unconfined aquifer. However, some water also moves into the aquifer from the river during high river stage resulting in "bank storage" effects. Hydrographs showing the influence of the river stage on the unconfined aquifer at various locations along the Columbia River are presented by Newcomb and Brown (1961), Jensen (1987), Liikala et al. (1988), Schalla et al. (1988), Fruland and Lundgren (1989), Luttrell et al. (1992), McMahon and Peterson (1992), and Campbell (1994). For a general sitewide model, daily and seasonal changes in the river stage resulting from releases from upstream dams can be ignored, and a time-averaged river stage can be used for the prescribed-head value at the river.
Measurements of actual ground-water flux to the Columbia River would be extremely valuable for model calibration. However, because of the large flow of the Columbia River compared to the contribution from ground water, measurements of the relatively small flow rate changes expected to occur along the Hanford Reach are not feasible with any known technology. Estimates of ground-water discharge to the Columbia River have been made in past studies. Luttrell et al. (1992) applied a flow net analysis to calculate discharge in the area of the old Hanford Townsite. They estimated 6.6 x 106 m3/yr discharge to about a 1-km length of the river. An earlier estimate of 2.7 x 106 m3/yr for the same area was based on the sitewide flow model (Prater et al. 1984). In comparison, the average annual Columbia River flow is about 1.06 x 1011 m3/yr.
The Yakima River borders the southeastern corner of the modeled area for a distance of approximately 25 km. This includes the western edge of the southern end of the Hanford Site and the western edge of the city of Richland. The Yakima River has usually been represented by a prescribed-head boundary in previous models (Jacobson and Freshley 1990). Because the water levels in the river are higher than the heads within the adjacent aquifer, the river is a potential source of recharge. The recharge rate is controlled by the hydraulic conductivity of sediments adjacent to the river and the head difference between the river and aquifer. However, the recharge at this boundary is highly uncertain because of a lack of wells and a corresponding lack of information concerning hydraulic properties and water-level elevations near the river. This causes uncertainty in the model predictions of ground-water flow within the area between the Yakima and Columbia Rivers, an area which is becoming increasingly important as commercial development immediately south of the Hanford Site boundary continues.
As part of a study of ground-water chemistry of the Pasco Basin (Ebbert et al. 1993), the U.S. Geological Survey found evidence that the Yakima River recharges the unconfined aquifer in the reach adjacent to the Hanford Site. This conclusion was based on a comparison of the chemical composition of river water, ground water from a well completed in the Saddle Mountains Basalt, and ground water from an offsite well completed in the unconfined aquifer (Ringold Formation) near the river.
To help define aquifer behavior in the vicinity of the Yakima River, river-stage monitoring has been conducted at a location just below Horn Rapids Dam. As reported in Thorne et al. (1993), water levels were continuously monitored at well 699-S24-19 for both the unconfined aquifer system and the basalt confined aquifer system. As shown in Figure 2.2, water levels at this well do not show a direct response to changes in river stage. However, the water level of the unconfined aquifer interval does respond to the filling of a canal (the Horn Rapids Ditch) between the well and the river.
The section of the Yakima River below Horn Rapids Dam flows through flood plain sediments that consist of moderately permeable stream channel deposits within fine-grained overbank and oxbow lake deposits. In this area, the unconfined aquifer may be somewhat isolated from the river by these fine-grained deposits near the river. Examination of drilling logs for private wells near the river shows that there is often fine-grained material near the water table, which sometimes acts as a locally confining unit. After water-bearing sediments are encountered, the water level in the well rises into the depth interval corresponding to the fine-grained material. The presence of low-permeability sediments near the river would also explain the lack of water-level response to the river stage at well 699-S24-19. However, because this well responds to filling of the canal, which is closer to the well, it is likely that the low-permeability sediments do not extend to the canal location. The lack of response could also be explained by recent silt deposits in the bed of the river.
The boundary of the model region crosses the Cold Creek Valley at the northwestern corner of the Hanford Site. This is an area where the model boundary does not coincide with a physical boundary of the unconfined aquifer flow system. The unconfined aquifer sediments extend into the valley and are a conduit for recharge to the Hanford Site aquifer system. Actual recharge quantities from Cold Creek Valley are not known. Jacobson and Freshley (1990) used a prescribed-flux boundary with an assumed recharge of about 9100 m3/d at the mouth of the Cold Creek Valley in two of the cases they ran for the inverse calibration model. The result in both cases was unrealistically high head values calculated by the model in the vicinity of Cold Creek Valley. Therefore, either the prescribed recharge at this boundary was too large or transmissivities in the area were set too low. Better results were obtained by Jacobson and Freshley (1990) when using a prescribed-head boundary. However, uncertainty in the transmissivity distribution remains, because it is not known if the recharge calculated by the model at this boundary, which depends on the hydraulic gradient across the boundary and the transmissivity of the adjacent model elements, is realistic.
A hydraulic test was conducted at well 699-43-104 during 1994. This test resulted in a relatively low transmissivity estimate of 25 m2/d and an equivalent hydraulic conductivity of about 2 m/d. However, these values may not be representative of the bulk of Cold Creek Valley sediments.
Flow-system boundaries are formed by the contact between the unconfined aquifer system and basalt. At places where basalt subcrops above the water table, this contact may form either a perimeter boundary or an island of basalt within the model area. The basalt contact also forms the lower boundary of the unconfined aquifer system except in some areas where a mud unit may underlie the aquifer directly over basalt.
Some of the perimeter basalt contact boundaries (i.e., Rattlesnake Mountain) may be recharge boundaries because of the infiltration of precipitation runoff and spring discharge from the upper slopes. There is also a potential for interflow (recharge or discharge) between the basalt confined aquifer system and the unconfined aquifer system at the lower boundary. Over most of the Site, the amount of interflow is thought to be small because of the low hydraulic conductivity of the rock separating the two aquifer systems. However, areas of increased vertical flow have been previously identified in the Gable Mountain and Gable Butte area on the basis of chemistry data (Graham et al. 1984; Jensen 1987). Hydraulic head data for the uppermost confined basalt aquifer also indicates the potential for water to discharge from this aquifer upward into the unconfined system in the northeastern part of the Hanford Site (Spane and Raymond 1993; Spane and Webber 1995). Figure 2.3 shows a comparison of observed hydraulic heads for the two aquifer systems and delineates areas of upward and downward hydraulic gradient.
Another potential area of increased vertical flow between aquifers is in the vicinity of the Yakima River horn, where the river has incised the upper basalt confining layers. A recent investigation (WHC 1993) identified a bimodal distribution of chloride in the unconfined aquifer in this area. Some wells yield concentrations of less than 10 mg/L and other wells have greater than 20 mg/L. The lower concentration ground water is chemically similar to water from Rattlesnake Hills springs, suggesting that this ground water comes from subsurface discharge from the underlying basalts. The ground water with higher chloride concentrations may come from infiltration of surface flow, which is subject to greater evaporation.
Interflow between the unconfined and basalt confined aquifer systems is not accounted for by the current numerical model. The rate of ground-water movement between the confined and unconfined aquifer systems is difficult to quantify. Therefore, it is not known if ignoring this contribution has a significant effect on the accuracy of the ground-water flow model. Differences in ground-water chemistry and temperature offer two possible methods for identifying areas of enhanced interflow and possibly quantifying flow rates. The possible use of temperature logs for this purpose has been preliminarily investigated and results are presented in Thorne et al. (1994).
Natural recharge to the unconfined aquifer system occurs from infiltration of runoff from 1) elevated regions along the western boundary of the Site, 2) infiltration of spring water that originates from the basalt aquifer system, and 3) infiltration of precipitation falling across the Hanford Site. Some recharge also takes place along the Yakima River, in the southern end of the Site. Since the start of Hanford operations in the mid-1940s, the estimated recharge from these natural sources has been less than the artificial recharge from waste-water disposal facilities. However, during the past 5 years, most production activities on the Hanford Site have been curtailed resulting in a decrease in waste-water disposal. Currently the volume of artificial recharge is similar to the volume of natural estimated recharge (Fayer and Walters 1995).
The Columbia River is the principal discharge area for the unconfined aquifer system. A few wells produce water from the unconfined aquifer on the Hanford Site (Figure 2.4). However, the total volume produced is relatively small and is not expected to be a significant discharge component on the sitewide scale. The supply wells serving the 400 Area have the highest withdrawal rates, which average about 500 m3/d.
Natural areal recharge from precipitation falling on the Hanford Site is highly variable both spatially and temporally, ranging from near zero to more than 100 mm/yr depending on climate, vegetation, and soil texture (Gee et al. 1992; Fayer and Walters 1995). Areas with shrubs and fine-textured soils like silt loams tend to have low recharge rates, while areas with little vegetation and coarse-textured soils, such as dune sands, tend to have high recharge rates. Recharge is also generally higher near the basalt ridges because of greater precipitation and runoff. Past estimates of recharge have been summarized in earlier status reports (Thorne and Chamness 1992; Thorne et al. 1993). To support the three-dimensional model, a natural recharge map (Figure 2.5) was developed by Fayer and Walters (1995). The distributions of soil and vegetation types were mapped first. A recharge rate was then assigned to each combination on the basis of data from lysimeters, tracer studies, neutron probe measurements, and computer modeling. Estimated recharge rates for 1992 were found to range from 2.6 to 127 mm/yr and the total volume of natural recharge from precipitation over the Hanford Site was estimated at 8.47 x 106 m3/yr. This value is of the same order of magnitude as the artificial recharge to 200 Area waste disposal facilities during 1992 and is about half the volume of discharge to these facilities during 1979 (Fayer and Walters 1995).
The large volume of waste water discharged to disposal facilities (Figure 2.6) on the Hanford Site over the past 50 years has significantly affected the ground-water flow system. As shown in Figure 2.7, the volume of artificial recharge has decreased significantly during the past 10 years and is currently still decreasing (Barnett et al. 1995; Dresel et al. 1995). Until it was taken out of service in 1984, Gable Mountain Pond received the largest volume of discharge on the Hanford Site. Major ground-water mounds have occurred beneath B Pond, Gable Mountain Pond, and U Pond, and have affected sitewide ground-water flow patterns (Bierschenk 1959; Dresel et al. 1995). Waste water is no longer being discharged to U Pond and Gable Mountain Pond, which have been decommissioned and are now dry. Other smaller-volume recharge sources have existed until recently in the 100, 200, and 300 Areas and may affect ground-water flow on a local scale. Currently, the two major artificial recharge sources are B Pond and the 200 Area Treated Effluent Disposal Facility (TEDF). Effluent discharges to TEDF began in April 1995, and averaged from 545 to 817 m3/d during the first 2 months. Discharge volumes averaged about 1909 m3/d during June through September, and have been averaging from 2074 to 2290 m3/d since September. Eventually, discharge to B Pond is planned to be eliminated. After that time, all tritiated water will be disposed to the State-Approved Land Disposal Site (SALDS), and clean water will be disposed to TEDF. Additional information on waste-water discharge is available in the Hanford Site Ground Water Protection Management Plan (Barnett et al. 1995).
The city of Richland infiltration ponds, agricultural and lawn irrigation, and ground disposal of waste water at a potato-processing plant are other sources of artificial recharge that may affect ground-water flow in the north Richland area and in the southern part of the Hanford Site (Liikala 1994).
Understanding the lateral extent and relationships between the hydrogeologic units found in different parts of the Hanford Site is crucial to understanding the movement of ground-water contaminants and for constructing accurate contaminant transport models. For example, it is important to determine whether or not fine-grained units found in the eastern and western portions of the Site directly overlap one another in the central part of the basin to form a continuous aquitard.
The steps involved in developing the hydrogeologic framework were 1) identify the minimum number of distinct hydrogeologic units to adequately define the Hanford unconfined aquifer, 2) determine the geologic contacts between these layers at as many wells as possible across the Site, and 3) use the three-dimensional visualization software package EarthVision({3}) to transform this data set into two-dimensional grids (layers) for each of the units. The two-dimensional grids are combined into a three-dimensional model of the hydrogeologic structure when they are input to the three-dimensional ground-water flow model.
Identification of Hydrogeologic Units The movement of ground water within the aquifer is controlled by the hydraulic conductivity, which is closely related to the sediment texture. Texture is a function of the grain-size distribution, sorting, and consolidation/cementation. Sediments were differentiated into either coarse or fine texture groups, then split into individual hydrogeologic units based on stratigraphic position, color, and distinctive markers such as ash horizons. Normally, identification of geologic units also uses depositional environment and relative time of deposition to define contacts between units. Because we are interested in the movement of ground water, the important geologic information is related to the movement of ground water. Figure 2.8 shows a comparison of a geologic stratigraphic column and the one developed here. The two are very similar, but it is important to clarify the difference. An example is the lower part of the upper Ringold as defined by Lindsey (1992), which in some places becomes progressively more sandy with depth. Where sand is the only (or overwhelmingly dominant) grain size, it was grouped with the underlying coarse-grained Unit 5. Although this may not conform to standard geologic classification, the sandy base of the upper Ringold is probably hydraulically connected with and hydrologically similar to Unit 5, with which it is grouped in this report. Generally, sands were grouped with sandy gravels, and silt was grouped with clay, assuming similar hydraulic conductivities. Nine distinct hydrogeologic units were identified above the top of basalt.
Others have previously identified similar units in studies focused on operational areas. Individual reports on the 100 Areas (Peterson 1992; Hartman and Lindsey 1993; Lindberg 1993a,b; Lindsey and Jaeger 1993), the 200 Areas (Connelly et al. 1992a,b), and the 300 Area (Swanson 1992) have been released in the past 5 years, but few geologic studies have addressed the regions of the Hanford Site lying between these operational areas. The nine hydrogeologic units identified for this conceptual model are similar to those in the reports above, with some differences in the location of unit contacts in places as discussed above and shown in Figure 2.8.
Determination of Geologic Contacts Data from 426 wells across the Hanford Site have been used to define hydrogeologic units based on textural composition. Top of basalt was identified in an additional 150 wells. The areal distribution of these wells is shown in Figure 2.9. Data used in defining hydrogeologic units included well logs, downhole geophysical logs, particle size analyses, calcium carbonate content, and geologic interpretations from other reports. Once the distribution of each of the hydrogeologic units was understood, a line showing the estimated extent of each unit was generated, i.e., lines showing where each unit reaches zero thickness. These "extent lines" were made as accurate as possible on the west and south sides of the Hanford Site, where the units pinch out on the basalt highs near the edge of the model. To the north and east, however, the numerical model extends only as far as the Columbia River and the basalt highs are much further away than that. Consequently, in those areas the extent lines were drawn to some arbitrary distance beyond the river to create an appropriate thickness for each unit at the edge of the model beneath the river. The gridding program interpolated the data set beyond the actual lateral extent of the unit and the model boundary was used to truncate the interpolated two-dimensional grids, where necessary.
The uppermost basalt flow was used as the bottom of the hydrogeologic framework because it forms the base of the Hanford unconfined aquifer. Figure 2.10 shows the top of basalt elevation. Unit 9 lies directly above basalt at the bottom of the unconfined aquifer system. This unit consists of fluvial sand and gravel and generally correlates to Lindsey's Unit A (basal Ringold). Unit 9 is found in the deeper parts of the basin, pinching out on (or eroded from) the limbs of the basalt anticlines. Figure 2.11 shows an isopach map of Unit 9 within the Hanford Site, and Figure 2.12 shows the distribution of wells where this unit was identified. In most places, Unit 9 is overlain by Unit 8. Unit 8 is equivalent to Lindsey's Lower Mud Sequence (the lower Ringold and part of the basal Ringold) and forms an aquitard across much of the Site. The mud in this unit is often described as blue or green, sticky clay, and frequently includes a white "ash" that may correspond to the ash in the lower Ringold in Bjornstad (1984). As shown in the isopach map (Figure 2.13), Unit 8 is relatively extensive across the Site. Figure 2.14 shows the locations of wells where this unit was identified.
Units 7 and 6 have more complex relationships and are more difficult to classify. During the time these units were deposited, the river channel apparently shifted position more often, depositing a complex pattern of overbank and mainstream deposits. To simplify the conceptual model, Unit 7 is defined as the coarse-grained sediments immediately overlying Unit 8. Figure 2.15 shows an isopach map of this unit. The locations of wells where Unit 7 is identified are shown in Figure 2.16. Unit 6 is defined as the sequence of mostly fine-grained sediments with some interbedded coarse-grained sediments overlying Unit 7 and underlying Unit 5. Figure 2.17 shows an isopach map of Unit 6. The locations of wells where Unit 6 is identified are shown in Figure 2.18. Unit 7 generally corresponds to Lindsey's Units B and D, while Unit 6 corresponds to Unit C and the unnamed mud layers.
Where coarse-grained Unit 7 is not present, Units 6 and 8 cannot usually be distinguished. In these cases, the fine-grained sediments are usually grouped into Unit 8. Likewise, where fine-grained Unit 6 is not present, Units 5 and 7 cannot be distinguished and the coarse-grained sediments are grouped into Unit 5. Unit 5 corresponds to the fluvial, coarse-grained sediments of Lindsey's Unit E (middle Ringold) (see Figure 2.8). This unit is quite thick in the western portion of the Site where Units 6 and 7 are not recognized. In many parts of the Site, the water table is presently found in Unit 5. Figure 2.19 shows an isopach map of Unit 5. The locations of wells where the unit is identified are shown in Figure 2.20.
Overlying Unit 5 is Unit 4, a fine-grained fluvial and lacustrine unit that corresponds to Lindsey's Upper Ringold Unit. Unit 4 has been eroded from large portions of the Site. Figure 2.21 shows an isopach map of Unit 4. The locations of wells where Unit 4 is identified are shown in Figure 2.22. In the eastern part of the area north of Gable Mountain, distinction between the fine-grained Unit 6 and the probable base of Unit 4 cannot be made, and the sediments are all grouped into Unit 6.
Units 2 and 3 correspond to the early "Palouse" soil and the Plio-Pleistocene unit, respectively. Unit 3 is a buried soil horizon containing caliche and side-stream basaltic gravels. It is only recognized in the western part of the basin (Figure 2.23). The locations of wells where Unit 3 was identified are shown in Figure 2.24. The caliche developed on the top of the eroded Ringold sediments and has a low hydraulic conductivity, while the side-stream gravels have a high conductivity. There is only one small area south of the 200-West Area where Unit 3, as the side-stream gravels, intersects the water table. Unit 2 is a small pocket of fine-grained sediments that have been interpreted as eolian silt. Figure 2.25 shows an isopach map of Unit 2. The locations of wells where Unit 2 was identified are shown in Figure 2.26. Unit 2 does not intersect the water table.
Unit 1 is the Hanford formation, which is generally a high permeability sand and gravel unit that covers most of the Hanford Site. In most areas where Unit 1 is below the water table, the sediments are gravels or coarse sands. The finer-grained sand- and silt-dominated facies are mostly above the water table within the boundaries of the Hanford Site. For this study, the surficial sand dunes have been included with Unit 1. Figure 2.27 shows an isopach map of Unit 1. The locations of wells where Unit 1 is identified are shown in Figure 2.28.
Lying beneath the gravels of the Hanford formation in the central portion of the Hanford Site are the sand and gravel deposits commonly called the "pre-Missoula gravels" (PSPL 1982). These sediments have been grouped with the Hanford formation (Unit 1) for the following reasons: 1) the pre-Missoula gravels cannot be readily distinguished from the Hanford formation in most driller's or geologist's logs, 2) there are no known hydraulic property data for the pre-Missoula gravels, although its properties probably lie between the younger Hanford gravel-dominated facies and older sandy gravel of Unit 5, and 3) the pre-Missoula gravels are above the water table except in some areas near the Hanford Townsite and near the solid waste landfill in the center of the Hanford Site. Therefore, they do not present a primary pathway for ground-water movement.
Transformation of Data Set into Grids EarthVision is a software package developed to assist geologists in interpreting geologic data and provide a three-dimensional visualization. The data set developed as discussed above was run through the EarthVision software to develop the two-dimensional grids for each unit. The software also allowed visual inspection of the ensuing grids and their modification as necessary. Data were recorded so that the real value for a contact was given if the unit was present. If the contact was uncertain, the well did not penetrate deeply enough, or the contacts had not been determined yet, the data point was flagged. A different flag was used when the unit did not occur at that well. Several problems occurred during this stage of model development which are discussed in more detail below.
One of the problems encountered was the wide variety in accuracy and/or descriptions of sediments recorded by drillers. A majority of the wells on the Site were logged only by the driller. Where there was doubt as to the veracity of the geologic contact for a particular well (e.g., a thick mud layer was missing in one well and present in all those nearby), the suspect contact(s) were flagged in such a way that the gridding software would ignore that well's value for only that contact. This technique allowed the use of other contacts picked from the wells that were believed to be most accurate. It also helped smooth the grids somewhat by removing data that looked like "fliers" from consideration in the gridding process.
Another difficulty occurred because geologic layers could be simulated as either grading smoothly from one texture to another, or as an abrupt textural change that occurs from sediments being deposited on an erosional surface. However, the software program could not accommodate both of these processes occurring in different places in the same hydrogeologic unit. The effects of this are most noticeable in the thickness distribution of an affected unit. A geologic unit pinches out to zero thickness on underlying units in a depositional environment, whereas it may reach zero thickness abruptly as a cliff if the unit had been eroded. In this conceptual model, the unit edges pinch out only. This causes some distortion of the real hydrogeologic thickness in places, but is thought to have minor effects on the overall ground-water modeling.
Finally, the program was allowed to interpolate between data points. This interpolation may not exactly match other interpretations based on an understanding of geologic processes. It was not possible to constrain the two-dimensional gridding process to make it exactly fit our current geologic understanding of the Site. However, spot checks of where the Ringold Formation intersects the water table, indicate the computer interpolation is not greatly different from other interpretations.
Hydraulic properties including both horizontal and vertical hydraulic conductivity (Kh and Kv), storativity (S), and specific yield (Sy) are key components of the conceptual model. To support three-dimensional numerical modeling, the distribution of each of these parameters must be specified for each hydrogeologic unit. Hydraulic conductivity controls the rate of water flow through a unit thickness of the aquifer at a given hydraulic gradient. Storativity and specific yield determine the change in water-table elevation that will occur in response to a change in the volume of water stored in the aquifer.
Hydraulic property data for the Hanford Site unconfined aquifer have been derived mainly from aquifer pumping tests and, in a few cases, from laboratory permeameter tests. These results have been documented in dozens of published and unpublished reports over the past 50 years. A summary of available data for the unconfined aquifer was provided in DOE (1988) and an updated summary was provided in Thorne and Newcomer (1992) together with an evaluation of selected pumping test analyses. Additional tests have been conducted both to support the three-dimensional model and to support other Hanford Site projects. Some of the recent tests are documented in status reports on the development of the three-dimensional conceptual model (Thorne and Chamness 1992; Thorne et al. 1993, 1994).
During 1995, a pumping test was conducted by the city of Richland on a new water supply well located near Wellsian Way in the southern part of Richland. Data were collected from a nearby observation well and analyzed to provide an additional measurement of hydraulic properties for the Hanford formation (Unit 1) in this area. The test analysis and results are described in detail in Appendix A.
Newcomb and Strand (1953) analyzed the growth of ground-water mounds beneath liquid disposal facilities in both the 200-West Area and 200-East Area between 1948 and 1953 to estimate hydraulic properties for these areas. Recent decreases in disposal volumes have caused a decrease in these mounds that has been analyzed to obtain additional hydraulic property information. Details of the analysis of the mound dissipation are provided in Appendix B.
Hydraulic conductivity values for sediments composing the unconfined aquifer system range from less than 10-4 m/d for some mud units to about 106 m/d for coarse gravel flood deposits. The sand and gravel facies of the Ringold Formation are about 10 to 100 times less permeable than the coarse sediments of the overlying Hanford formation (DOE 1988). The Ringold Formation also contains relatively extensive layers of fine grained, low permeability sediments such as silt or clay. *
Most pumping test analyses result in estimates of aquifer transmissivity (T), which, for a vertically homogeneous aquifer, is the product of Kh and aquifer thickness (b). However, for an aquifer composed of n layers having different hydraulic conductivities, transmissivity is given by
(2.1)
A listing of available transmissivity data obtained from pumping tests in the unconfined aquifer system is provided in Appendix C. Where possible, the thickness of the tested aquifer facies has been noted and used to calculate an equivalent Kh. Figure 2.29 shows the distribution of the tested wells across the Hanford Site and also indicates the main geologic unit tested. The data listed in Appendix C include 36 single well pumping tests and 3 multiple well pumping tests that pertain to the Hanford formation (Unit 1). Thirty-seven single well pumping tests and 12 multiple well pumping tests pertain to Ringold Formation sand and gravel units (Units 5, 7, and 9). An additional 32 single well pumping tests, 7 multiple well pumping tests, and 2 specific capacity tests are included for which the tested hydrogeologic unit has not been defined. The quality of these results is affected by both aquifer conditions and analysis procedures and varies widely (Thorne and Newcomer 1992). Slug tests have also been conducted at several Hanford Site wells. However, because many of the single well slug test results are considered inaccurate, they have not been listed in Appendix C or used to determine hydraulic properties for the conceptual model. Multiple well slug tests have been conducted at a few wells in conjunction with multiple well pumping tests. Because of vertical aquifer heterogeneity, and because most of the tested wells at Hanford partially penetrate the unconfined aquifer, it is sometimes difficult to determine the aquifer thickness that should be used in calculating hydraulic conductivity from the test results.
As discussed in earlier status reports (Thorne and Chamness 1992; Thorne et al. 1993), the current approach for the three-dimensional conceptual model is to assign an areal distribution of Kh to the most significant permeable units that form the upper part of the unconfined aquifer system. Most ground-water flow and contaminant transport takes place in this part of the aquifer system. Single values of Kh are assigned to mud-dominated units and to deeper permeable units.
The uppermost permeable unit for most of the model region is either Unit 1 or Unit 5. Units 7 and 9 represent deeper permeable units. The hydraulic conductivity of Unit 1 generally ranges from about 1 to 1,000,000 m/d and is much higher than any of the other units that compose the unconfined aquifer system. Therefore, where it is present below the water table, this unit usually provides the dominant flow path within the aquifer. Figure 2.30 outlines areas of the Hanford Site where the water table was estimated to lie within the Ringold Formation during 1993. Unit 1 consists of sand and gravel of the Hanford formation and the pre-Missoula gravel deposits. Extensive fine-grained facies of the Hanford formation are not found below the water table within the model region. In the vicinity of B Pond, the saturated portion of the Hanford formation is composed of muddy sandy gravels that probably represent the lower limit of hydraulic conductivity for Unit 1. Aquifer tests (Thorne et al. 1993) indicate that the minimum Kh is about 1 m/d and the minimum Kv is about 0.02 m/d for Unit 1. The maximum measured value of Kh for Unit 1 on the Hanford Site is about 10,000 m/d (Thorne and Newcomer 1992; DOE 1988). However, the maximum hydraulic conductivity that can be measured by an aquifer test is limited by the well efficiency and the flow rate that can be pumped with available equipment. The upper limit of Kh for coarse gravel flood deposits of Unit 1 is probably greater than 1,000,000 m/d based on inverse numerical modeling. Maximum Kv is unknown, but may approach the value for Kh in relatively clean gravel zones where stratified layers of finer grained material are not present.
Units 5, 7, and 9 are all within the Ringold Formation and consist of sand to muddy sandy gravel with varying degrees of consolidation and/or cementation. Unit 5 is the most widespread unit within the unconfined aquifer and is found below the water table across most of the model region. Hydraulic conductivities of Units 5, 7, and 9 determined from aquifer tests vary within the range of about 0.1 to 200 m/d. Because these units are hydrologically similar, they were grouped together in areas where the intervening mud units do not exist. A few aquifer tests suggest vertical anisotropy is in the range of 0.01 to 0.1. Therefore, the range of Kv is estimated at about 0.001 to 20 m/d.
Mud-dominated units within the unconfined aquifer system include Unit 4, also known as the upper Ringold fines; Unit 6, which is a composite of intercalated mud and sand and gravel layers; and Unit 8, which is an extensive lower Ringold mud unit. Hydraulic conductivity of these units is generally about 2 to 5 orders of magnitude less than that of the permeable sand and gravel units. Therefore, the mud units are essentially aquitards and are not expected to transmit significant quantities of water or contaminants in the horizontal direction. They are most significant in slowing the vertical migration of contaminants and influencing vertical head distributions. Therefore, the values of Kv assigned to mud units are probably more important than the assigned values of Kh.
Hydraulic test results for mud-dominated units are listed in Table 2.1. These few tests yielded hydraulic conductivity (K) values of 0.0003 to 0.09 m/d. Some of the results are from well tests and some are from laboratory tests. Because of a tendency to complete wells only in zones
|
Hanford Well Number |
Hydraulic Conducitivity (K) (m/d)
|
Hydrogeologic Unit
|
|
299-W7-9
|
0.09
|
Unit 4 (vadose)
|
|
699-20-39
|
<0.06
|
Unit 6
|
|
699-84-35A
|
0.03
|
Unit 6 |
|
699-41-40
|
0.0003
|
Unit 4
|
Storativity and specific yield can be calculated from multiple well pumping tests and multiple well slug interference tests (Spane 1993, 1994). Storativity and specific yield results from the relatively few multiple well tests conducted on the Hanford Site are listed in Table 2.2. The average specific yield from these tests was 0.15. However, some of these estimates are highly uncertain because of the effects of nonideal test conditions, such as partially penetrating wells and aquifer heterogeneity. Such conditions generally have a more significant effect on the determination of storage properties than on the determination of transmissivity. Moench (1994) demonstrated that these conditions can affect specific yield values calculated from type-curve analysis of aquifer pumping tests, and usually result in the calculated values being low.
Specific yield can also be calculated by measuring the change in saturated aquifer volume in response to the injection or withdrawal of a known volume of ground water. This method was applied to the decreasing ground-water mound that occurred beneath the 200-West Area between
| Well | Storativity (S) | Specific Yield (Sy) | Hydrogeologic Unit Tested |
| 199-K-10 | 0.00007 | 0.04 | 5 |
| 299-w10-13 | 0.009 | - | 5 |
| 699-S27-E9A | 0.013 | 0.37 | 1,5 |
| 699-S22-E9B | 0.005 | 0.02 | 1,5,7 |
| 699-S14-20C | 0.005 | 0.25 | Unknown |
| 699-26-35C | 0.0015 | - | 1 |
| 699-31-53B | - | 0.38 | 5 |
| 699-32-72 | - | 0.05 | 5 |
| 699-36-61A | - | 0.05 | 5 |
| 699-37-82A | 0.02 | 0.18 | 5 |
| 699-42-40C | 0.02 | - | Unknown |
| 699-42-42B | 0.00003 | - | Unknown |
| 699-43-89 | - | 0.05 | Unknown |
| 699-47-35C | 0.002 | 0.15 | 9 |
| 699-48-77C | 0.001 | - | Unknown |
| 699-55-50A | - | 0.2 | 1 |
| 699-62-43B | - | 00.6 | 1 |
Specific yield for Unit 1 is estimated to range from about 0.1 to 0.3 and is expected to be higher for coarse, well sorted gravel than for poorly sorted mixtures of sand and gravel. Storativity is estimated to range from 0.0001 to 0.0005. Specific yield is estimated to range from 0.05 to 0.2 for the generally poorly sorted sediments of Units 5, 7, and 9. Storativity is estimated to range from 0.0001 to 0.001 for these units.
Hydraulic head information is important for determining ground-water flow direction and velocity. Head measurements are also needed to establish initial conditions for ground-water flow modeling and for model calibration.
Water levels have been measured on at least an annual basis using a sitewide well network since the 1940s. More than 600 wells are currently measured each year to determine the hydraulic head distribution for the unconfined aquifer on the Hanford Site and adjacent areas. Results of the 1994 measurements are presented in Dresel et al. (1995). Additional water-level data for the North Richland area are provided in Liikala (1994). The annual water-level measurements provide an extensive database that can be used to define initial head conditions for numerical modeling and for a comparison of modeling runs with historical data.
Prior to the mid 1980s, hydraulic heads increased by more than 13 m in some areas of the Hanford Site in response to waste-water disposal activities. Before waste-water disposal operations began, the uppermost aquifer was almost entirely within the Ringold Formation, and the water table extended into the Hanford formation at only a few locations near the Columbia River (Newcomb and Strand 1953). However, waste-water discharges have caused the water-table elevation to rise into the Hanford formation in the vicinity of the 200-East Area and in a wider area near the Columbia River. Figure 2.30 outlines areas of the Hanford Site where the water table was estimated to lie within the Ringold Formation during 1993. Water levels have begun to decrease over most of the Hanford Site during the last several years because of decreases in waste-water discharge (Dresel et al. 1995).
Most of the wells in the current unconfined aquifer monitoring network are completed in the upper part of the aquifer, within 7 m of the water table. Most of the wells that were originally open to a greater depth interval were reconfigured in the early 1980s. Additional details concerning the reconfiguration of these wells are provided in a later section on contaminant distributions.
Three-dimensional modeling requires information on the vertical distribution of hydraulic head as well as the areal distribution. Therefore, a listing of selected wells currently completed in the deeper part of the unconfined aquifer and wells with individual piezometers open to different depth intervals was compiled and presented in an earlier conceptual model status report (Thorne et al. 1993). An updated version of this listing is provided in Appendix D. Water levels measured in some piezometer clusters are presented in Table 2.3. Figure 2.31 shows the location of these wells at the Hanford Site. Some of the measurements may be affected by well construction problems. Some of the wells containing several piezometer tubes placed to various depths in a perforated well casing were originally completed by backfilling around the piezometer tubes with sand. Because the sand may not have provided adequate isolation of the individual depth intervals, water level data from this type of piezometer completion are suspect. Other piezometers may be in communication with the well annulus or other isolated intervals.
To accurately model contaminant transport, parameters including effective porosity, dispersivity, and retardation coefficients must be specified. Longitudinal, transverse, and vertical dispersivity values are needed for a three-dimensional model. Retardation coefficients are not discussed here because they are specific to each contaminant species and may vary depending on geochemical conditions within the aquifer. Information of retardation coefficients for Hanford unconfined aquifer sediments is available in Ames and Serne (1991) and Kaplan and Serne (1995).
Porosity is defined as the volume of void space divided by the total volume of the soil or rock matrix that contains it. Effective porosity does not include void space that is isolated from ground-water flow and, therefore, may be smaller than the total porosity. The average velocity of a conservative contaminant (non-sorbing and non-decaying) as it moves through an aquifer is equal to the average linear velocity of the ground water, which is inversely proportional to the effective porosity of the aquifer matrix (Freeze and Cherry 1979). Porosity can be determined from
| Well | Piezo | DTW | Piezo | DTW | Piezo | DTW | Piezo | DTW | Piezo | DTW |
| Increasing Piezometer depth --------> | ||||||||||
| 199-H4-15C | S | 9.83 | R | 10.24 | Q | NM | P | NM | ||
| 299-W22-24 | T | 72.91 | S | 74.72 | R | 71.71 | Q | 73.36 | P | 74.79 |
| 299-E23-2 | O | 97.33 | Q | 97.32 | P | |||||
| 699-S12-29 | Q | 26.98 | P | 26.22 | ||||||
| 699-2-33B | Q | 39.72 | P | |||||||
| 699-10-E12 | Q | 11.46 | P | 22.47 | ||||||
| 699-14-28 | Q | 33.57 | P | 33.52 | ||||||
| 699-25-33B | Q | 39.14 | P | 39.11 | ||||||
| 699-37-82B | O | NM | S | 52.71 | R | 53.12 | Q | NM | P | NM |
| 699-67-51 | Q | 38.46 | P | 38.89 | ||||||
| 699-69-45 | O | 27.26 | R | 27.36 | Q | 29.54 | P | 29.12 | ||
Laboratory measurements of porosity are available for samples from only a few of the available Hanford Site wells. Recently, 15 samples were collected from 6 wells at the 100 H Area (Vermeul et al. 1995). Porosity ranged from 0.19 to 0.41 and averaged 0.33 for the Ringold Formation and 0.31 for the Hanford formation. Samples from five depth intervals within the Ringold Formation at the 200-West Area were reported by Newcomer et al. (1995). The average porosity ranged from 0.21 to 0.33 and averaged 0.27. Laboratory porosity measurements are often considered unreliable, especially for unconsolidated sediments, because of the difficulty in obtaining undisturbed samples.
A few tracer tests have been conducted within the unconfined aquifer. Bierschenk (1959) reported an effective porosity of 0.10 from a tracer test with fluorescein dye under natural gradient conditions. Single borehole dilution tests, which do not provide information on porosity, were conducted by Graham et al. (1984). An effective porosity of 0.25 was assumed to calculate average ground-water velocity from the measurements. Borehole dilution tests and a two-well tracer test were conducted in the 200-West Area (Newcomer et al. 1995) under natural gradient conditions. However, porosity could not be determined from the two-well tracer test because the gradient was not well defined.
Porosity can also be estimated from measurements of aquifer specific yield. Specific yield is defined as the volume of water released from a unit area of an unconfined aquifer per unit decline in hydraulic head. Specific yield and effective porosity are equivalent if drainage of the aquifer matrix is complete. However, in reality, the specific yield may be lower than the effective porosity because of water held in pore spaces of the drained aquifer matrix by surface tension or adsorptive forces (Moench 1994).
As discussed in Section 2.5.2, specific yield can be calculated from 1) multiple well aquifer tests, or 2) measurements of the volume of aquifer drained or saturated in response to removing or injecting a known volume of ground water. A variation on the second method is presented in Appendix B. The specific yield was calculated from the change in saturated aquifer volume associated with dissipation of the ground-water mound beneath the 200-West Area from 1985 to 1995. The result was a specific yield value of 0.17, which is higher than values calculated by Newcomb and Strand (1953) when they analyzed the growth of ground-water mounds beneath liquid disposal facilities in both the 200-West Area and 200-East Area between 1948 and 1953. Water levels beneath the 200-West Area had increased by an additional 5 to 10 m from 1953 to 1985. Therefore, the difference in porosity could be caused by a difference in the sediments saturated during the 1953 to 1985 period compared to those during 1985 to 1995. Specific yield results from the relatively few multiple well tests conducted on the Hanford Site unconfined aquifer are listed in Section 2.5.2. These results range from 0.01 to 0.37 and average 0.15. However, some of these estimates are highly uncertain because the effects of nonideal test conditions, such as partially penetrating wells and aquifer heterogeneity. Such conditions generally have a more significant effect on the determination of storage properties than on the determination of transmissivity. Moench (1994) demonstrated that these conditions can affect specific yield values calculated from type-curve analysis of aquifer pumping tests, and usually result in the calculated values being low.
Mud-dominated units generally have higher porosity than sand-and-gravel-dominated units. Davis (1969) compiled porosity values that indicate ranges of 0.35 to 0.5 for silts and 0.4 to 0.7 for clays, respectively. However, because of the low permeability of such sediments, the porosity assigned to mud units in the model is not expected to have a major impact on model results.
As a solute plume moves through the aquifer it is dispersed by a combination of mechanical mixing and molecular diffusion. The three-dimensional transport of an ideal conservative solute affected only by advection and dispersion as it travels through an ideal isotropic, homogeneous and rigid porous media is given by the mass balance equation
(2.2)
The scale dependence of dispersivity is generally believed to result from spatial and temporal variations in the ground-water velocity field, which are caused by both spatial variations in hydraulic conductivity, and spatial and temporal variations in hydraulic gradient (Goode and Konikow 1990). Aquifer heterogeneity results in additional plume dispersion, beyond that seen in the laboratory, because the solute moves at different rates through different parts of the aquifer. Temporal variations in hydraulic gradient have been shown to have a strong effect on transverse dispersion (Goode and Konikow 1990; Rehfeldt and Gelhar 1992; van der Kamp et al. 1994).
Theoretically, the differences in flow paths that cause macrodispersion could be accounted for in a numerical flow model by a detailed delineation of the hydraulic conductivity throughout the model domain. However, this approach is impractical for large-scale problems because 1) such detailed and accurate information on hydraulic conductivity is not usually available, and 2) the number of elements required would be too large for most computers to handle. For example, Rehfeldt and Gelhar (1992) calculated that a minimum of 2 billion elements would be required to describe the hydraulic conductivity distribution for the Cape Cod sewage plume, which is 3500 m long, 1000 m wide, and 25 m thick. Because of these problems, the generally accepted approach for modeling dispersion is to use field scale macrodispersivity values.
Longitudinal dispersivity applied in a two-dimensional transport model accounts for dispersion caused by vertical variations in ground-water flow paths. In a three-dimensional model, some of these different flow paths are built into the model. Therefore, the correct longitudinal dispersivity for a three-dimensional model may be smaller than for the corresponding two-dimensional model.
Possible approaches for determining the appropriate dispersivity values for the Hanford sitewide three-dimensional model include
The third approach has generally been used in the past for determining dispersivity values for Hanford Site transport modeling. Law (1992) used values of AL = 43 m and AT = 12 m for a scale of 9500 m based on values compiled in Gelhar et al. (1985). An earlier model (WHC 1990) used values of 15 m and 1.5 m, respectively, for longitudinal and transverse dispersivity, which were also based on Gelhar et al. (1985).
Dispersivity values determined from field tests at 59 different sites were compiled by Gelhar et al. (1992). These include results from two investigations at the Hanford Site. The first was a 1950s tracer test that resulted in values of AL = 6 m and AL = 460 m for the Hanford and Ringold formations, respectively, as reported by Bierschenk (1959). Also included are values of AL = 30.5 m and AT = 18.3 m for a scale of 20,000 m. These were calculated from two-dimensional transport modeling of the 200-East tritium plume as reported in Ahlstrom et al. (1977). The dispersivity data in Gelhar et al. (1985) were classified according to quality of data and plotted to show the correlation between scale of observation and dispersivity values. Separate plots are presented for longitudinal, horizontal transverse, and vertical transverse dispersivity. Longitudinal and horizontal transverse dispersivity show a positive correlation with scale of observation. A correlation was not indicated for vertical transverse dispersivity. However, very few measured vertical dispersivity values were available.
Dispersivity is likely to vary across the Hanford Site depending on the degree of heterogeneity and the temporal variability of flow gradients. Ahlstrom et al. (1977) noted that the ratio of AT to AL calculated from their model of the Hanford Site was much higher than the ratio expected. They attributed the high ratio to heterogeneity. However, horizontal dispersion may have been enhanced by temporal variations in flow gradients caused by disposal practices. The flow paths for the tritium transport from the 200-East Area have gradually shifted from due east to a south-easterly direction, in response to waste-water discharges to B Pond and the 200-East Area. This shift in the flow path has enhanced the apparent dispersion of the tritium plume emanating from the 200-East Area.
To establish initial conditions for a transport model, information on both the areal and vertical distributions of contaminants within the unconfined aquifer is needed. Temporal data on the distribution of contaminants is also needed for model calibration.
Concentrations of both chemical and radiological contaminants are measured in hundreds of Hanford Site wells each year. Contaminant distributions measured during 1994 and information on sampling and analysis techniques are provided in Dresel et al. (1995). Like the hydraulic-head measurement network, the current sampling network is composed mainly of wells completed in the upper part of the unconfined aquifer system, generally less than 7 m below the water table. Most of the wells that were originally open to a greater depth interval were reconfigured in the early 1980s based on an investigation by Eddy et al. (1978), which showed that contaminant concentrations were highest near the top of the aquifer. The sampling network wells were, therefore, reconfigured so that ground water containing the highest concentrations was sampled and dilution with relatively uncontaminated water from deeper in the aquifer was avoided. Remediation of the sampling wells also eliminated the potential for enhanced vertical migration of contaminants through the well casing.
A limited number of wells and piezometers completed in the deeper part of the unconfined aquifer system are currently available. However, for some of the piezometer completions, the integrity of the seals between intervals is questionable. As mentioned in the last section, some wells containing several piezometers in a single perforated well casing were originally completed by backfilling around the piezometer tubes with sand. Because the sand may not have provided adequate isolation of the individual depth intervals, data obtained from these piezometers are suspect and should not be used. Most of the piezometers containing sand backfill were recompleted in the 1970s and 1980s by removing the sand and placing cement seals to isolate piezometer intervals. However, as noted in Appendix D, some piezometer wells still have apparent completion problems. Table 2.4 lists contaminant data for some deeper sample intervals within the unconfined aquifer system. All of the vertical contaminant data listed in Table 2.4 are from individual wells or piezometers with documented cement seals isolating the depth intervals and where head differences exist between different isolated intervals. The data indicate that small amounts of contaminants are found at depths of at least 50 m below the water table in some areas of the Hanford Site and their concentrations decrease with depth.
| Well | Date Sampled | Water Table Depth (m) | Sampled Interval Depth (m) | Tritium (pCi/L) | Nitrate (mg/L) |
| 299-W22-24 T | 7/11/95 | 73.1 | 89.0 - 96.6 | 16,900 | <0.02 |
| 299-W22-24R | 7/11/95 | 73.1 | 125.6 - 133.2 | 296 | 0.2 |
| 299-W22-24 Q | 7/11/95 | 73.1 | 144.8 - 151.5 | 194 | 0.5 |
| 699-14-E6 S | 6/14/95 | 28.3 | 88.4 - 92.1 | <10 | NA |
| 699-18-21 | 6/12/95 | 43.1 | 58.5 - 70.7 | 133,000 | NA |
| 699-24-01 S | 6/13/95 | 31.7 | 60.4 - 67.4 | <10 | NA |
| 699-24-01 R | 6/13/95 | 31.7 | 89.3 - 95.7 | <10 | NA |
| 699-24-01 Q | 6/13/95 | 31.7 | 96.6 - 108.2 | <10 | NA |
| 699-28-40 Q | 6/14/95 | 48.2 | 103.6 - 106.7 | 815 | NA |
| 699-31-11 | 6/12/95 | 28.7 | 56.7 - 73.8 | 156,000 | NA |
| 699-36-46 R | 7/11/95 | 91.7 | 109.7 - 116.4 | <10 | <0.02 |
| 699-36-46 R | 3/23/95 | 90.8 | 131.1 - 137.8 | <200 | 0.03 |
