WORK PLANS - Laserfiche WebLink

Home Browse Search

Geosyntec'% consultants 3.1.2 EVS® Gridding Technique EVS® was used to interpolate chemical concentrations at RRI by generating a 3D grid from spatial X, Y, and Z data and concentration (C) input data. X, Y, and Z define the location of each point and C is the concentration value of the chemical at that point. The top of the model was constrained by water level data from site wells and the San Joaquin River in fourth quarter 2015 (Table 4). The bottom of the model was constrained at 85 feet below msl based on the approximate depth of a clay aquitard observed in borings throughout RRI (Tetra Tech, 2002). Water quality monitoring data are input at the exact geographic coordinates (X, Y) of the respective sample and mid-screen elevation of the respective well screened interval (Z) or at the exact elevation of a discrete grab- groundwater sample. Cell sizes are not uniform across the model domain, but rather are adapted in shape and size such that the position of each chemical data point is within a unique cell. In this manner, the interpolation will completely honor the input data such that the interpolated concentration value in a cell containing input data is equal to the input concentration value.Once the grid is defined,EVS®utilizes a 3D kriging algorithm to interpolate the log-transformed input concentration data and assign an interpolated concentration value to each cell in the grid. 3D kriging algorithm was used as the geostatistical basis for the 3D contaminant concentration interpolation. The kriging algorithm used in EVS®attempts to develop an estimate of the spatial correlation between pairs of data points as the basis for interpolation and extrapolation of COC concentrations at locations without data. Data weighting factors and variance are calculated using a semi-variogram model that encompasses the expected spatial variability of COC concentrations based on the spatial variability in the COC data. The estimated variance between modeled COC concentrations at each grid location and the spatially-weighted average of COC data are then minimized to produce a"best-fit" model. This "best-fit" model was manually fit to the data for each map by adjusting the range and sill in the semivariogram model until the kriged results matched the data. The horizontal to vertical anisotropy value for each model was also manually adjusted to match the data.The horizontal to vertical anisotropy ratios was set at 6:1 for the models. The secondary anisotropy ratio for site-specific inset models was set in the approximate direction of groundwater flow at 1.5:1. A secondary anisotropy was not established for the island-wide models, as the apparent direction of groundwater gradient varied throughout the site. RRI Groundwater Monitoring Work Plan 17 1 June 2018