Analysis of Hypoxia and Sensitivity to Nutrient Pollution in Salish Sea

Abstract

Increasing levels of nutrients, persistent hypoxia, harmful algal blooms, and increased frequency of fish kills are degrading the ecological health of the Salish Sea. An improved version of a diagnostic hydrodynamic and biogeochemical model (nutrients, phytoplankton, carbon, dissolved oxygen, and pH) of the Salish Sea has been developed with the ability to simulate characteristic circulation and water quality features. Sensitivity tests were conducted to assess the responsiveness of the system to land-based (rivers and wastewater sources) nutrient loading. The influence of Fraser River on the magnitude of estuarine exchange with the Pacific Ocean and nearshore habitat was examined given that it contributes nearly half of the total freshwater discharged to the Salish Sea. A large region of hypoxia in Hood Canal that extends over 30-40 km during its peak was reproduced and attributed primarily to the existence of a two-layer classic fjord-type circulation and a nearly stagnant deep bottom layer that occupies nearly 60% of the water column. Nitrate mass in the euphotic zone from land-based and oceanic sources is depleted to near-zero limiting levels during summer. Under such conditions, the Salish Sea is responsive to changes in nutrient loads entering the euphotic zone directly. A hypothetical scenario involving the elimination of land-based nutrient sources results in notable water-quality improvement, featuring a reduction in algal biomass (≈5.4%), reduction in sediment oxygen demand (≈17.1%), and significant reduction in hypoxic area (≈39%) and exposure in area-days to bottom layer hypoxia (≈62%) within the Salish Sea. KHANGAONKAR ET AL.4735

Salish Sea Response to Global Climate Change, Sea Level Rise, and Future Nutrient Loads

“…Figure upper left panel shows the model domain, which encompasses Vancouver Island completely, allowing tides to propagate into the Salish Sea around Vancouver Island through Johnstone Strait and the Strait of Juan de Fuca. The hydrodynamic component of the Salish Sea Model uses the finite volume community ocean model (FVCOM; Chen et al, ) with an unstructured grid framework and has been discussed in detail previously (Khangaonkar et al, , , ).…”

“…With the aid of improvements described above, observed biogeochemical cycles (i.e., salinity, temperature, DO, nitrate, algal biomass, and pH) in the Salish Sea, including near‐bed hypoxia in locations such as the Lynch Cove region of Hood Canal, have been reproduced. The biogeochemical component of the Salish Sea Model has been described in detail previously (Khangaonkar et al, , ). The model has been tested over multiple years and subjected to sensitivity level testing with variation in forcing/loading either due to anthropogenic causes or due to stressors associated with climate change projections (Bianucci et al, ; Roberts et al, ).…”

“…Evidence of low dissolved oxygen (DO) levels and hypoxia (DO concentrations <2 mg/L) has been documented in several basins (e.g., Albertson et al, , ; Barnes & Collias, ; Embrey & Inkpen, ; Harrison et al, ; Mohamedali et al, ; Newton et al, ; Paulson et al, ; Puget Sound Action Team, ; Roberts et al, ; Stockner et al, ). Khangaonkar et al () showed that eliminating all land‐based nutrient sources would result in notable water‐quality improvement, including a reduction in algal biomass (≈5.4%), reduction in sediment oxygen demand (≈17.1%), significant reduction in hypoxic area (≈39%), and reduction in exposure in area days to bottom‐layer hypoxia (≈62%) within the Salish Sea.…”

“…These data products are typically on a grid (e.g., 30 × 30 km) that cannot adequately resolve coastal nearshore estuarine systems. Site‐specific models such as the Salish Sea Model in the Pacific Northwest region of the United States (Khangaonkar et al, , , ) or the Chesapeake Bay Model in the Mid‐Atlantic region (Cerco & Noel, ; Irby et al, ; Jiang & Xia, ; Ye et al, ) are typical examples of coastal estuarine models that have the required nearshore resolution. They are typically operated in the “hindcast” mode using observed river flow, ocean boundary, and meteorological data.…”

A Hybrid Lagrangian-Eulerian Particle Model for Ecosystem Simulation

“…The unstructured 395 grid can be designed to provide a customized variable resolution to both coastline and 396 bathymetry. With the merits of ideal geometric fitting and local refinement of mesh 397 resolution, FVCOM has been used in numerous applications to estuaries, coastal oceans, 398 and the Great Lakes [Yang et al, 2013;Xue et al, 2015;Anderson et al, 2016;Xue et al 399 2017; Khangaonkar et al, 2018, Ye et al, 2018. These characteristics make the model 400 well suited for the study of Sandusky Bay.…”