1. A T HREE- D IMENSIONAL W ATER Q UALITY M ODEL OF S OUTHERN P UGET S OUND Greg Pelletier, P.E., Mindy Roberts, P.E., Skip Albertson, P.E., and Jan Newton, Ph.D. Washington State Department of Ecology, P.O. Box 47710, Olympia, WA, 98504-7710, e-mail: gpel461@ecy.wa.gov, www.ecy.wa.gov/programs/eap/spasm/ A BSTRACT A three-dimensional hydrodynamic and water quality model of southern Puget Sound was developed based on the EFDC model. The water quality model simulates the concentration of dissolved oxygen (DO) in response to primary production of phytoplankton, oxidation of organic material, and sediment flux (Figure 1 and Table 1). Of the 22 possible state variables in EFDC, the model of southern Puget Sound includes: two phytoplankton groups, three forms of organic carbon, three forms of organic phosphorus and nitrogen, dissolved reactive phosphorus, ammonia nitrogen, nitrate plus nitrite nitrogen, chemical oxygen demand, and dissolved oxygen. Fecal coliform was also included. A sediment process model is coupled with the water column model. The sediment model incorporates three processes: depositional flux of particulate organic matter (POM), diagenesis of POM, and the resulting sediment flux. The model was calibrated to limited data collected between October 1996 and September 1997. Because of the relative scarcity of data, the model is only crudely calibrated at the present time. COMPUTATIONAL GRID The spatial domain of southern Puget Sound was divided into 1906 discrete cells in a curvilinear orthogonal grid (Figure 2). The average grid cell is approximately 630 by 630 meters. Each grid cell has four vertical layers in the stretched-sigma coordinate system used by the EFDC model. The mass balance equations in the EFDC model are solved using the finite-difference method. The portion of the computational grid that represented Budd Inlet was extracted from the full grid to create a separate sub-model of Budd Inlet. This was done to facilitate model calibration using data from Budd Inlet. CALIBRATION PARAMETERS Calibration of the water quality model consists of selection of values to represent the kinetics and constants of the model equations. Two sets of parameters were evaluated for the model of southern Puget Sound. The first set is the selected calibration parameters for representation of Chesapeake Bay and the Peconic estuary (Cerco and Cole, 1994, Tetra Tech, 1998). The second set of parameters that were considered was developed for a modeling study of Budd Inlet (LOTT, 1999; Boatman and Edinger, 1999). The status of calibration of the EFDC model to southern Puget Sound consists of a comparison of the model results with observed concentrations using the two sets of model parameters, and a sensitivity analysis of model predictions for various ranges of key parameters. CALIBRATION RESULTS A comparison of predicted and observed chlorophyll, dissolved inorganic nitrogen (DIN), and dissolved oxygen (DO) in the surface and bottom layers of the central Budd Inlet station are shown in the figures below. The model runs presented below used the entire grid of southern Puget Sound. In general, the model describes many of the seasonal patterns of the observed data. In the surface layer (Figures A-C), chlorophyll increased during the April-September season. Concentrations of DIN were predicted to decline during the same period, which is consistent with observed patterns caused by uptake of nutrients by algae. The model did a good job of predicting DO in the bottom layer of the central part of Budd Inlet (Figure D). DO in the bottom layer of the inner inlet was over-predicted (Figures E-G). A sensitivity analysis was performed to determine which factors could improve calibration of DO in the inner inlet. SENSITIVITY ANALYSIS The following parameters were selected for a sensitivity analysis, and are listed in approximately decreasing order of sensitivity of the predicted bottom layer dissolved oxygen: Sediment oxygen demand (Figure E) Settling rates (Figure F) Algal basal metabolism rates (Figure G) Maximum algal growth rates Zooplankton algal predation rates Nitrogen half-saturation Calibration of the sediment model will be very important. Sediment flux and concentration data for southern Puget Sound are either very limited or nonexistent in most areas of southern Puget Sound, so this is an important data requirement for refinement of the calibration. Field studies for other key parameters, especially settling rates for phytoplankton and particulate organic matter, and algal basal metabolism rates would also be useful for further refinement of the model calibration. BOUNDARY CONDITIONS The concentrations of water quality state variables at the saltwater boundary of the model grid are referred to as boundary conditions. Salinity, temperature, algae, organic carbon, dissolved oxygen, nitrogen, phosphorus, and fecal coliform bacteria were specified based on monitoring data collected near the open boundaries for the simulation period of 10/1/96 through 9/30/97. The freshwater inflows and loads to the model domain were characterized for the same period as described by Roberts et al. in a separate paper for this conference. Central Budd Inlet (BUD005), surface: Central Budd Inlet (BUD005), bottom: Visit our project web site: www.ecy.wa.gov/programs/eap/sps/ A B C D E F G