APPENDIX N VADOSE ZONE FLOW AND TRANSPORT The description of the movement of groundwater and solutes through the vadose zone from the ground surface to the water table of the underlying, unconfined aquifer is a major element in estimation of impacts on groundwater quality and human health for this Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington (TC & WM EIS). At the Hanford Site, past operations, current practices, and proposed future activities will affect groundwater conditions for long periods of time. For this reason, the assessment of potential impacts relies on mathematical modeling of vadose zone processes rather than monitoring or measurement of conditions. The scope of vadose zone analysis for this TC & WM EIS is large, including contributions from tanks and ancillary equipment at 18 high-level radioactive waste tank farms, six sets of cribs and trenches (ditches) immediately associated with tank farm activities, proposed new Integrated Disposal Facilities for radioactive and hazardous waste, and closure of the Fast Flux Test Facility. In addition, approximately 380 facilities not included in the scope of decisions of this TC & WM EIS are analyzed for their contribution to cumulative impacts. The primary objective of vadose zone analysis is to estimate the rates and magnitudes of movement to the unconfined aquifer of water and solutes introduced with natural recharge, planned liquid discharges, leaks, spills, and disposals. The estimates of release to the vadose zone described in Appendix M and transport through the unconfined aquifer described in Appendix O interface closely with the vadose zone analysis described in this appendix. A secondary objective of vadose zone analysis is to provide an understanding of the influence of the proposed Black Rock Reservoir on future Hanford Site (Hanford) hydrologic conditions. Estimates of human health impacts, based on integration of estimates of the rate of release (see Appendix M) and the rate of transport through the vadose (Appendix N) and saturated (see Appendix O) zones are presented in Appendix Q. Comparisons of impacts within and across alternatives are presented in Chapters 5 and 2, respectively. The balance of this appendix comprises a description of the technical approach to vadose zone analysis, a summarization of results for Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington (TC & WM EIS) alternatives, and a discussion of the sensitivity analysis of major parameters incorporated into the analysis of vadose zone processes. Although best available data and models are used to construct the analysis described in this appendix, uncertainty in results remains. This uncertainty derives from variability in natural conditions such as rates of precipitation and recharge and spatial heterogeneity of soil types, as well as lack of knowledge in areas such as the applicability of specific models to site-specific locations and conditions and the type of climate to be experienced in the future. N.1 TECHNICAL APPROACH TO VADOSE ZONE ANALYSIS The technical approach to vadose zone analysis involves selection of the upper-level framework for the analysis and implementation of framework specifics, including specification of the spatial extent of model study areas, characterization of geologic conditions, development of model grid configurations, and establishment of values of model parameters. The following sections describe these two elements of the vadose zone analysis. N.1.1 Upper-Level Approach The upper-level approach involves consideration of the boundaries of the model domain, the degree of integration of vadose and saturated zone analysis, and the establishment of initial and boundary conditions for the model. The required spatial extent of the model domain is that of the Hanford Site, approximately 1,450 square kilometers (560 square miles), with model depth ranging from 10 to 200 meters. The time frame for the analysis is from the start of site operations in calendar year 1944 out to 10,000 years in the future. The 10,000-year period of analysis is longer than the 1,000-year period N–1 Draft Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington recommended in the U.S. Department of Energy (DOE) guidance for performance assessment of low-level radioactive waste and deactivated high-level radioactive waste (HLW) facility closure (DOE Guide 435.1-1) but adequate to capture the longer vadose zone travel times of select constituents of potential concern for the TC & WM EIS analysis. The preferred approach to groundwater modeling is full integration of vadose and saturated zone analysis with transient location of the water table determined as part of the solution of a single model. Because of the large extent of the model domain, the small size of subareas of interest, and the long time period for analysis, the implementation of this approach is not practical with state-of-the-art computing capabilities. A second approach to integration of vadose and saturated zone analysis would be the specification of a single, large-scale saturated zone model; specification of smaller subarea models for integrated vadose and saturated zone analysis; and integration of the subarea models with the single, large-scale saturated zone model. Because of the rather small size of the required subareas and the time requirements for computation of transient water table locations for multiple subareas, this approach is also not practical at the current time. Specific issues that complicate integration of the MODFLOW (modular three-dimensional finite-difference groundwater flow model) saturated zone and STOMP (Subsurface Transport Over Multiple Phases) vadose zone–saturated zone models are the short duration of the transient period to be represented by the simulations and the spatial variation of the water table. The upper-level approach adopted for this TC & WM EIS groundwater analysis is the development of a single, large-scale saturated zone model followed by the development of multiple small-scale vadose zone–only models that are coupled with the saturated zone model through equivalent specification of boundary conditions to provide a consistent, integrated analysis of transient groundwater conditions. The development, calibration, and implementation of the large-scale saturated zone model are described in Appendix L. Simulation of the vadose zone subareas is accomplished using the STOMP computer code (White and Oostrom 2000, 2006). The STOMP model uses an integrated-volume, finite-difference approach to solve nonlinear water and solute transport balances for the vadose zone. Features of the STOMP model used in the TC & WM EIS analysis include (1) a three-dimensional representation of geology, hydraulic properties, and grid geometry; (2) temporal and spatial variability of groundwater recharge at the ground surface; (3) temporal and spatial variability of water and solute injection at any horizontal location and vertical depth; and (4) water and solute output fluxes at specified surfaces. Three-dimensional representation was selected to incorporate spatial heterogeneity of geologic and recharge conditions and to explicitly simulate the complexity of travel time behavior due to lateral spreading and preferential flow reflecting local conditions. The relationships of moisture content and pressure and moisture content and hydraulic conductivity within the vadose zone were simulated using the van Genuchten and Mualem models (van Genuchten 1980; Mualem 1976). These models contain seven adjustable parameters: saturated moisture content, residual moisture content, saturated hydraulic conductivity for three spatial directions, and two additional empirical constants that are determined by comparison with site data. N.1.2 Vadose Zone Model Implementation More than 400 subarea models are required for analysis of TC & WM EIS alternative and cumulative analysis sites. Each of these is simulated as a rectangular box where the upper surface represents the ground surface and the lower surface, the water table. The thickness of this box, different for each subarea site, is established from the long-term steady state of the unconfined aquifer model. Boundary conditions for the upper surface at each site are a specified recharge determined by technical guidance (DOE 2005) and zero flux of solute. Boundary conditions for the lower surface are atmospheric pressure and a zero gradient of solute. Boundary conditions for each of the four sides of the box are zero flow of water and solute. N–2 Appendix N ▪ Vadose Zone Flow and Transport The pattern of the horizontal grid for each subarea was based on the aqueous discharge from the source. The grid pattern and model extents were incorporated to limit the effect the boundary conditions and node size had on the model conditions. All nodes within the source were equal in size and were bounded by the source site boundary. Grid sizes could increase or decrease by the harmonic rule, meaning node lengths could increase or decrease by one and a half the adjacent node length. Sources with no aqueous discharge could have node length no greater than 20 meters (66 feet) within the source site. The node size could increase by the harmonic rule to at least 120 meters (39 feet) from the source boundary. Sources with aqueous recharge were categorized as moderate (<1 meter [3 feet] per year) or heavy (>1 meter [3 feet] per year). Moderate recharge sites had a grid length of no larger than 5 meters (16 feet) within the source site. The maximum 5-meter (16-foot) grid length continued to 50 meters (164 feet) from the site boundary. The grid size increased by the harmonic rule to a distance 150 meters (492 feet) from the site boundary. The heavy recharge site had a grid length of no larger than 5 meters (16 feet) within the source site. The maximum 5-meter (16-foot) grid length continued to 50 meters (164 feet) from the site boundary. The grid size increased by the harmonic rule to a distance of 170 meters (558 feet) from the site boundary. Given these conditions, development of the model was completed by specifying values of hydraulic properties for 16 Hanford soil types and subarea-specific geology and grid dimensions. In summary, the process for the selection of hydraulic parameter values involved the matching of predicted to measured borehole moisture content profiles for all 16 soil types followed by the matching of randomly generated to observed unconfined aquifer conditions for 3 primary soil types. It also provided for consistency with values of saturated hydraulic conductivity determined in the calibration of the saturated zone model and with area-specific geology and grid size requirements. A flow diagram for the process is presented as Figure N–1, and greater detail is provided in the following paragraphs. A stepwise, iterative procedure was applied to determine area-specific geology and grid dimensions and to identify values of hydraulic properties that best match conditions observed at the site. In an initial step, values of vadose zone parameters were determined for the 16 soil types by matching moisture content profiles predicted using the van Genuchten relationship to moisture content profiles measured in 140 undisturbed vadose zone boreholes. Values of saturated hydraulic conductivity were restricted to ranges consistent with the calibrated saturated zone model. An example of the match between predicted and observed moisture contents for an undisturbed borehole in the 200-East Area is presented as Figure N–2. The blue dots in the figure represent moisture content determined by the neutron scattering method. The red line is the model fit to the borehole data. The horizontal gray lines represent soil contact changes. The soils represented in this figure are Hanford Gravel, Hanford Sand, Plio-Pleistocene Silt, and Plio-Pleistocene Gravel. At this stage, sensitivity analysis was performed for generic 200-East and 200-West Areas to establish grid size requirements for accurate computations. Given this information, the interpretation of borehole data was applied to assign soil types for each of the approximately 400 study areas on a grid block–specific basis. An example of the interpreted borehole data is presented as Figure N–3, where the lithology of the cross-section is vertically exaggerated. Figure N–3 represents a geologist’s interpretation of the subsurface geology at B- and BX-Tank Farms in the 200-East Area. Single or multiple cross sections of interpreted borehole data were used to specify a three-dimensional spatial distribution of soil types that is encoded into STOMP input files for each of the study areas. An example of this translation into STOMP input data is presented in Figure N–4 for the TX Tank Farm in the 200-West Area. The STOMP data of Figure N–4 for the 200-West Area show layers of Plio-Pleistocene soils present at the TX Tank Farm that are not present in the borehole data of Figure N–3 for the 200-East Area at the B- and BX-Tank Farms. Not all of the 16 soil types are present at all locations; within the specific cross section presented as Figure N–4, for example, only 7 of those soil types are found. This contrast is representative of the level of detail of the spatial distribution of soil types that is captured in the interpretation and translation process. N–3 Draft Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington Key: STOMP=Subsurface Transport Over Multiple Phases. Figure N–1. Flow Diagram for Selection of Values of Vadose Zone Hydraulic Parameters N–4 Appendix N ▪ Vadose Zone Flow and Transport Borehole 299-E33-338 0 0 0 8 HGr 0 0 HSa 0 6 m.) c e ( bl a er t 0 at 0 w 0 e 4 v o b a n o ati v e el 00 HGr 0 2 PPlSi PPlGr 0 Key: HGr=Hanford Gravel; HSa=Hanford Sand; PPlGr=Plio-Pleistocene Gravel; 0.0 0.1 0.2 0.3 0.4 PPlSi=Plio-Pleistocene Silt. vol. % moisture Figure N–2. Predicted and Measured Moisture Content Profiles N–5 Draft Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington Figure N–3. Borehole Stratigraphy Data Figure N–4. Vertical Cross Section of a Grid for a STOMP Vadose Zone Model Volume for the TX Tank Farm (200-West Area) The final element of the model development process, the establishment of final values of the van Genuchten parameters, was accomplished by selecting parameter values that match conditions observed in the unconfined aquifer. Two data sets were employed: (1) observed conditions in the unconfined aquifer that could reasonably be associated with a single source in the vadose zone; and (2) observed conditions in the unconfined aquifer, primarily concentrations of hydrogen-3 (tritium), that are associated with a group of sources. The first data set was used to select the parameter values; the second, to verify the final set of parameter values. The three sets of concentration data for the unconfined aquifer for single-source sites were (1) the concentration of beta activity below the BY Cribs, (2) the beta activity below the BC Cribs, and (3) the activity of iodine-129 in the vicinity of the 216-T-26 Crib. The review of area-specific geology established that three soil types, Hanford Gravel, Hanford Sand, and Ringold Gravel, jointly represent more than 90 percent of the sediments present in the vadose zone at Hanford. Also, a travel time sensitivity analysis conducted for simple layered geometry established that the movement of water and solute through the vadose zone is largely controlled by these three soil types, with a secondary contribution from Plio-Pleistocene Silt in the 200-West Area. In particular, this finding is applicable for the three areas for which single-source data are available. Accordingly, the refinement of hydraulic parameter values focused on Hanford Gravel, Hanford Sand, and Ringold Gravel. For these three soil types, a systematic search of the parameter space was conducted. To ensure that the entire space of admissible parameter values was investigated, a statistical search and screening were performed. The search involved specification of the range of values for each parameter and random selection of values from uniform distributions defined over the range. The screening involved calculation of the moisture content at a specified constant rate of recharge and comparison thereof with the range of moisture content observed at the site. The step of the procedure identified 18 million combinations of sets of hydraulic parameter values that met the initial screening requirement. The simulation of movement through the vadose at the three single-source sites was implemented using the STOMP computer code. Predicted fluxes of solute in the water were then used to estimate concentrations in the unconfined aquifer; in the near-field, a mixing-box model was used, and at distances removed from the source, a particle tracking model. At this stage, hydraulic properties of the Plio-Pleistocene Silt were adjusted as needed to match conditions at the 216-T-26 Crib. Sets of values that passed each of these tests were judged acceptable for use in vadose zone analysis. This step of the analysis is described in the following paragraphs using the BY Cribs as an example. N–6 Appendix N ▪ Vadose Zone Flow and Transport A time series of measurements of gross-beta activity and technetium-99 concentrations at a single location in the unconfined aquifer below the BY Cribs is presented as Figure N–5. The gross-beta data include contributions from beta-emitters other than technetium-99, while more recently, concentrations of technetium-99 have been measured separately and reported in addition to the concentrations of gross-beta activity. Using TC & WM EIS data for inventory of technetium-99, historical dates of aqueous discharge and current values of vadose zone hydraulic parameters, the time series of concentration of technetium-99 below the BY Cribs was estimated using the STOMP model and is presented in Figure N–6. The predicted concentration profile reflected in that figure shows an early peak due to rapid movement of the large initial aqueous discharge and a long-term plateau due to a more gradual release of technetium-99 retained in the vadose zone. The early peak of the predicted technetium-99 profile occurs at the same time as the early peak of the measured total beta profile (see Figure N–5) but is lower because of the presence of radionuclides other than technetium-99 among beta emitters. The concentration level measured and predicted for technetium-99 for the current time period are in general agreement. Thus, the predicted concentration profile for technetium-99 shows qualitative agreement with the reported concentration of gross-beta activity, supporting continued investigation of this set of values for the vadose zone hydraulic parameters. 100000000 10000000 1000000 L) 100000 Ci/ p n ( o Gross Beta ntrati 10000 Tc-99 e c n Co 1000 100 10 1 4/12/49 2/15/56 12/20/62 10/24/69 8/28/76 7/3/83 5/7/90 3/11/97 1/14/04 Time Figure N–5. Time Series of Measured Gross Beta Activity Below the BY Cribs (observed at well 299-E33-7) N–7 Draft Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington Figure N–6. Time Series of Predicted Concentration of Technetium-99 Below the BY Cribs In addition to reports of time series of concentrations at single locations, the site monitoring program reports estimates of the spatial distribution of contaminants at specific points in time. Estimates of isopleths of concentration of technetium-99 near the BY Cribs based on measurements reported for 2007 are presented in Figure N–7. These data were used to provide additional testing of the proposed set of values of vadose zone hydraulic parameters. The approach used TC & WM EIS source data for the BY Cribs, the STOMP vadose zone model, the MODFLOW-predicted transient flow field, and a particle tracking transport model to predict spatial distribution of technetium-99 in the unconfined aquifer for calendar year 2005. The results of this analysis are presented in Figure N–8. The predicted concentrations show both qualitative and quantitative agreement with measured concentrations, with high levels near the sources and decreasing levels in the northwest direction. The predicted concentrations also show movement to the southeast due to transient flow in that direction under the influence of high aqueous discharges from past Hanford operations. N–8 Appendix N ▪ Vadose Zone Flow and Transport Key: DWS=Drinking Water Standard; LLWMA=low-level waste management area; WMA=waste management area. Figure N–7. Isopleths of Concentration of Technetium-99 near the BY Cribs N–9 Draft Tank Closure and Waste Management Environmental Impact Statement for the Hanford Site, Richland, Washington Figure N–8. Particle Tracking Model Output of Technetium-99 Concentrations for Calendar Year 2005 due to Sources at the BY Cribs N–10