  |
|
|
| To: | Files | |
| From: | Mindy Roberts | |
| Cc: | Andrew Kolosseus, Brandon Sackmann, Greg Pelletier (Ecology) Tarang Khangaonkar, Zhaoqing Yang, Taeyun Kim, and Rochelle Labiosa (PNNL) Ben Cope (EPA) Project Advisory Committee | |
| Date: | May 21, 2009 | |
| Subject: | Puget Sound Dissolved Oxygen Modeling Quarterly Progress Report #2 (January through March 2009) |
The Puget Sound Dissolved Oxygen Modeling project began in October 2008 to evaluate factors contributing to low levels of dissolved oxygen in Puget Sound. The tools will address two questions:
The project includes developing model boundary conditions, applying circulation and water quality models for the entire Puget Sound, and evaluating potential management scenarios. The project team includes staff from Ecology, Pacific Northwest National Laboratory, and the Environmental Protection Agency Region 10. The team works with a Project Advisory Committee representing other organizations. This memorandum summarizes progress from January through March 2009. A previous quarterly report summarizes earlier progress.
PNNL purchased a 184-processor cluster for $50,000, as stipulated under the contract, to be used for the Puget Sound Dissolved Oxygen Model. PNNL is covering the additional costs associated with installation of cooling and added air conditioning per hardware operational specifications. Later in the project, PNNL will develop a remote connection protocol where staff at Ecology and eventually other scientists and engineers in the region can log on and run the model. Currently, only PNNL staff use the computer cluster.
PNNL developed the model grid based on the bathymetry data generated from Digital Elevation Models of Puget Sound (Finlayson, 2005) and bathymetry data of Canadian waters from Institute of Ocean Sciences, Canada. The grid has roughly 9000 model nodes and 14,000 elements in the horizontal plane with 30 vertical layers (Figure 1).
The model grid has two open boundaries. The western boundary is specified at the mouth of Strait of Juan de Fuca. The northern open boundary is located between the Texada Island and Johnstone Strait, to avoid the complex deep channels of Johnstone Strait. Model grid sizes in the straits vary from 3000 meters near the open boundaries to 1000 meters near the entrance of Puget Sound. Average grid size is about 800 m in main basin of Puget Sound and 680 m in South Puget Sound. The current model grid does not consider the wetting and drying of tide flats near the mouths estuaries.
Additional cells were added to the Tacoma Narrows region to achieve numerical stability and to South Puget Sound to enable any cross-inlet circulation variations. Differences between this model and the more detailed South Puget Sound dissolved oxygen model include Hope Island and the West Bay of Budd Inlet, which may lead to differences in Budd Inlet predictions between the two modeling efforts. However, the project team does not believe this will extend into other areas of South Puget Sound and is appropriate for the scale being used for the Puget Sound model.
The complicated shapes around the mouths of the Stillaguamish and Snohomish Rivers could not be represented accurately without adding significantly more grid elements to describe, which would have increased model run times. For example, the original 5 sloughs of the Snohomish were simplified to 3 sloughs but all contributions are included. Overall circulation in Whidbey Basin should still be described well.
Areas such as the Strait of Juan de Fuca are represented by larger grid cells (nominally 3000 m).
A total of 19 major rivers are included explicitly in the model. Ungaged stream flows are currently not considered in the model set up and may be added at later stage. The ungaged flows are not expected to affect the quality of model prediction for the main circulation in Puget Sound because of the relative small flow contribution of the ungaged flow into Puget Sound. (The pollutant loads will be accounted for in the water quality model development.)
Open boundaries are forced by tides based on XTIDE predictions. Temperature and salinity profiles along the open boundaries are specified using monthly survey data from 2006. The data gaps were filled using interpolated values from adjacent months. For meteorological forcing, PNNL evaluated the pros and cons of using measured data from meteorological stations and using output from an atmospheric model. Although for calibration it would have been preferable to use measured data, not all radiation data are available at all met stations, and considerable data filling would have been required. The decision was then made to use the NOAA National Centers for Environmental Prediction (NCEP) North American Regional Reanalysis (NARR) data to force the model at the surface. The NARR data are 3-hour intervals in 32-km resolution. Parameters include wind speed and direction, shortwave- and longwave-downward and upward radiation flux, and sensible and latent net heat flux.
Ecology developed river inflows for the intermediate-scale model based on extrapolating USGS-gaged rivers to all areas tributary to Puget Sound. For the Canadian watershed area, Ecology compiled Fraser River flows and also extrapolated gaged river flows from Environment Canada to Canadian areas tributary to Georgia Strait. As a first step, the circulation model is set up with only the Fraser River flows (>80% of the Canadian watershed area tributary to model domain) in Canadian waters and the USGS gaged data for 19 major tributaries to Puget Sound.
The project team will revisit accounting for all ungaged areas in the next few months. While the amount of water may be relatively small and likely does not influence circulation, it is important to account for it as this flow likely affects the distribution of brackish water along the shorelines. The San Juan Islands and other surrounding areas near Rosario and Haro Straits have hundreds of small streams that radiate out to multiple waterbodies, we do not expect to account every inflow at its exact discharge location.
For the water quality model, the mass loads from the entire watershed, including the ungaged areas, will be accounted for. Ecology developed GIS datalayers of the terrestrial areas tributary to the model domain in the US and Canada (Figures 2 and 3). The project team proposes this discretization for water quality constituents and possibly for river inflows. Watershed loads will be developed based on multiple linear regression techniques applied to monthly monitoring data from over 50 locations and applied to the entire model domain.
No year-round continuous temperature monitoring data are available for the major rivers tributary to Puget Sound. Ecology's ambient monitoring program captures monthly grabs and summer-only continuous temperatures. Because density is far more sensitive to salinity than to temperature (5°C differences contribute to 1 psu variation), the river temperatures will be based on monthly mean values recorded in the Cedar River by the USGS. Mindy Roberts (Ecology) compared these values to the monthly instantaneous measurements at the mouths of seven rivers (Figure 4) and found the values within roughly 2°C of each other.
Fraser River temperatures were based on Environment Canada monitoring. Brandon noted that Canadian river temperatures are often warmer than the US river temperatures. Because of the relatively low influence on density, the project team believes this would not influence the results significantly.
Wastewater treatment plant contributions will be added to the water quality model, but the separate flows are not included in the circulation model because they are insufficient to affect stratification or circulation.
The project team developed a general approach to calibration, based on the information presented in the Quality Assurance Project Plan. PNNL will calibrate water surface elevations against NOAA real-time measured tide gages (six in the model domain, including Port Townsend, Seattle, Tacoma, Cherry Point, Friday Harbor, Port Angeles) and additional water surface elevations predicted by XTide.
The project team will use year 2006 for model calibration. The 2006-2007 time period is the most data rich in recent years. While more data are available in 2007, the summer was uncharacteristically cool and wet, which produced less-intense low dissolved oxygen values.
Predicted velocities will be calibrated against Ecology's ADCP data in South Puget Sound and any available velocity data within the model domain.
Finally, vertical profiles of salinity and temperature will be compared between the model and Ecology's ambient monitoring data.
A 12-month model run for the circulation model takes 37 hours at a 10-second time step in parallel mode with 56 cores. The project team had targeted a run time of a day to facilitate water quality scenario development. Because the water quality model will use a coarser time step than 10 seconds, which was required for hydrodynamic model stability, these model run times are expected to meet the project objectives.
The water quality model will use an offline version of CE-QUAL-ICM coupled to the scalar transport scheme of FVCOM. PNNL developed a direct link between FVCOM and CE-QUAL-ICM with outside funding. Testing will be complete in the next quarter.
PNNL began developing the water quality model boundary conditions. For the western boundary, PNNL has compiled monthly marine nutrient and other water quality data from Canadian sources. As fewer water quality data are available for Georgia Strait north of the Fraser River, quarterly data will be used at the northern boundary. All data will be interpolated to the model time step. It should be noted that ranges for water quality state variable concentrations in the Georgia Strait will serve as one of the sensitivity tests to be conducted with the water quality model. Ecology will provide boundary conditions for the rivers and wastewater treatment plants.
The Box Model is being developed to describe large-scale or long-term processes relevant to dissolved oxygen. Brandon Sackmann (Ecology) developed marine boundary conditions for the Puget Sound box model using data from Ecology's long-term ambient monitoring program. The circulation model has been completed, and Brandon is developing the water quality time series for boundary conditions from the watershed (with assistance from Mindy on the wastewater treatment plants).
The inflow time series are being developed in conjunction with those for both the intermediate-scale Puget Sound Dissolved Oxygen Model and for Ecology's South Puget Sound Dissolved Oxygen Model. See below for additional information on the time series of inflows. The box model will merge river inflows into the larger watersheds tributary to the 10 basins, but the data on which they are based are common to all three efforts. In addition, the box model will simulate a 10-year period (1999-2008), whereas both the Intermediate-scale Puget Sound DO Model and the South Sound DO model use 2006 as the calibration year.
The intermediate-scale Puget Sound Dissolved Oxygen Model has been set up with flows from rivers included as boundary source terms. Streamflow data were obtained from USGS and Environment Canada for 30 rivers that discharge into the model domain. These data were used to estimate the daily river flow associated with each of the 66 watershed areas defined in the model. Flows from ungaged rivers were estimated by first choosing a representative gaged reference river, based on its similarity to the ungaged area, and then multiplying the reference flow by the ratios of the ungaged-to-gaged area and average annual precipitation. In cases where rivers are gaged well upstream from the river mouth, the drainage area downstream of the station is used as the ungaged area in the calculation.
Ecology is currently exploring ways to incorporate interactive data visualizations into project web pages and reports to make large datasets more readily accessible. The example that follows gives users the opportunity to perform more in-depth and self-guided explorations of flow data for rivers that discharge into Puget Sound and Canadian waters by allowing them to vary the range of the x-axis and by providing real-time readouts of data values at each point in time.
Figure 1. Circulation and dissolved oxygen model domain.
Figure 2. U.S. watersheds tributary to Puget Sound and the Strait of Juan de Fuca.
Figure 3. Canadian watersheds tributary to Georgia Strait and the Strait of Juan de Fuca.
Figure 4. Cedar River average monthly temperatures developed from USGS data compared with instantaneous monthly temperatures recorded by Ecology's ambient monitoring program at major rivers discharging to Puget Sound.
Copyright © Washington State Department of Ecology. See http://www.ecy.wa.gov/copyright.html.
