Seasonal variations in water clarity in different regions of the Bay are influenced by the seasonal dynamics of the aquatic constituents. In the oligohaline region, large pulses of sediment are coincident with low water clarity events (i.e., high light attenuation) during the spring freshet. While light attenuation by ISS makes up 30% of the total kbgc in April at the oligohaline mainstem stations, the largest contributor is CDOM for the years 2001–2005 (Appendix, Fig. 12). During the summertime, the relative contribution by CDOM is larger, at an average contribution of about 70% from June to August, with phytoplankton contributing approximately 10%. Heavy rain during autumn and winter storms again increases delivery of sediments to the Bay, which increases the contribution by ISS during those months. In the meso- and polyhaline regions, the relative contribution by phytoplankton is larger while the contribution by ISS is smaller compared to the oligohaline stations. The kbgc calculated by the parameterized expression in Eq. 3 is sometimes less than 0.6 m− 1 in the meso- and polyhaline regions. This occurs at times when salinity, the modeled proxy for CDOM, in the polyhaline region is high and the phytoplankton concentration is low. During times when \(k_{bgc}~>~0.6~\mathrm {m}^{-1}\), modeled contributors to light attenuation are CDOM, phytoplankton, and detrital carbon, listed in order of decreasing importance. In situ measurements of light absorption in the Bay confirm the importance of CDOM on optical properties through all regions of the Bay. The light absorption coefficient for CDOM is highest in the oligohaline regions, decreasing linearly along a conservative mixing line from low to high salinity waters (Rochelle-Newall and Fisher 2002), with evidence of seasonal variability extending into the shelf waters of the Mid Atlantic Bight (Del Vecchio and Blough 2004).
These processes contribute to the differences in water clarity between the two model runs. The spring and fall sediment inputs contribute to peaks in kh during April and September in the oligohaline region for the “bgc-light” run. A seasonal cycle for the difference in \(k_{h}^{-1}\) is most apparent in the mesohaline region, with a peak in kh for the “bgc-light” run in April and May and minima in the winter months (Fig. 4). During the summertime, average river flow is lowest in 2001 and highest in 2003 (Fig. 1), which affects the spatial extent and magnitude of differences in kh between the two model runs. The most widespread differences in the penetration depth of solar heating occur during the high-flow year of 2003 (Appendix, Fig. 13). Interannual changes in river flow have previously been linked to phytoplankton dynamics in the Bay (Harding et al. 2016), with increased nutrient loadings during high-flow years being associated with increased chl-a and shift in flora to larger cells. In addition to increased light attenuation by phytoplankton, the modeled contribution by CDOM is also greater when there is more freshwater, further decreasing water clarity.
The increased light attenuation of the “bgc-light” simulation warms a thin surface layer while decreasing solar heating for deeper waters. Because the warmer surface layer is a small fraction of the volume of the Bay, most of the Bay is generally colder, and the heat content is smaller from April to September (Fig. 5). Since the two model runs were forced with the same atmospheric conditions, river fluxes, and oceanic boundary conditions, differences in heat content must originate from changes in surface heat fluxes at the air-water interface or horizontal heat fluxes (at the mouth of the Bay and from the rivers). Surface heat fluxes include the evaporative or latent heat flux and sensible heat flux, the conductive heat flux from the water to the air. Monthly climatology of the heat fluxes for our simulation years shows latent heat flux is the largest heat sink in the Bay throughout the year (Appendix, Fig. 14). Latent heat flux is more negative in the “bgc-light” run, due to the higher surface temperatures of this run. Latent heat flux acts as a larger heat sink from March to June, causing the relative decrease in heat content during these months. From July to September, the latent, sensible, and horizontal fluxes are less negative in the “bgc-light” run, narrowing the gap in heat content between the two simulations during these months. In the natural environment, changes in surface water heat fluxes affect the ambient air temperature. While our model simulations are forced with prescribed atmospheric conditions, Jolliff and Smith (2014) demonstrate how changes in water clarity affect air temperature in a coupled ocean-atmosphere system. For fully coupled modeling systems, water clarity variability is important for modeling air temperature variability as well.
Summertime flow is generally a good predictor of the extent and intensity of the change in subsurface temperatures, except in 2005 which is the 2nd driest summer but exhibits similar subsurface temperature changes as the wettest summers, 2003 and 2004. The largest spring freshet occurred in 2005, which suggests springtime flow conditions and associated changes in water clarity are factoring in to the intensity and extent of summertime subsurface temperature differences between the two simulations. Previous studies have demonstrated links between winter to spring freshwater flow and summertime stratification, e.g., (Murphy et al. 2011), which supports this hypothesis. In our model runs, colder temperatures in the “bgc-light” run reach further seaward along the mainstem during years with higher flow (Appendix, Fig. 16). Due to the exponential dependence of both phytoplankton growth and zooplankton grazing on temperature, colder temperatures affect zooplankton concentrations both directly (by changing temperature) and indirectly (by decreasing zooplankton food availability).
In the “bgc-light” run, increased phytoplankton nutrient uptake in low salinity waters decreases the amount of nitrate that is transported downstream to higher salinity waters. This is supported by our Bay-integrated and surface results. Increases in phytoplankton in fresher waters are concurrent with declines in nitrate in saltier waters. Because fewer nutrients are transported downstream, there is less nitrate uptake and phytoplankton in the Bay. The colder temperatures and fewer phytoplankton of the “bgc-light” run contribute to the decline in zooplankton during the late summer to early fall months. Summertime flow conditions determine the magnitude and seaward extent of surface phytoplankton concentrations and the associated changes in nutrients. Increases in phytoplankton are confined to the northern Bay during years with the lowest June to August flow (e.g., 2001 and 2005), while the increases in phytoplankton extent are found further downstream in high flow years (e.g., 2003). These differences in the spatial distribution of the phytoplankton bloom also influence the location and extent of zooplankton concentrations. Relative and absolute differences in surface zooplankton between the “bgc-light” and the ”standard” runs along the Bay mainstem are larger and found further downstream during higher flow summers (e.g., 2003–2004).
During the spring freshet and fall storms, differences in temperature and phytoplankton averaged over several days are much less than the hourly differences. This is in part due to diurnal variations in surface processes. During the daytime hours, stratification is enhanced by surface warming while vertical mixing episodically homogenizes the upper water column during the evenings or wind events. This vertical mixing is more frequent during the fall, whereas during the spring, sustained density stratification allows for the development of greater temperature gradients. In the case of the spring freshet example, differences in hourly surface phytoplankton concentrations at the mainstem stations span − 33% to + 60%, with a difference of only + 2% averaged over those days and all the stations. Additionally, at station CB3.2 large relative increases in surface phytoplankton concentrations occurred after a period of high ISS concentrations. This suggests that as ISS settle out of the surface water, the warmer surface temperatures in the “bgc-light” run encourages growth, allowing phytoplankton to further utilize the abundant nutrients in the water.
Concluding Remarks
By comparing our two simulations, we showed how accounting for water clarity variability in the solar heating calculations can affect the estuarine system. In our ”standard” run, we had a “one-way” coupled configuration, in which the physical parameters influenced biogeochemistry, but the light attenuation coefficient for solar heating, kh, was constant at all locations and times. In the “bgc-light” run, a “two-way” coupled configuration, the variations in kh resulted in greater surface temperature variability and decreased bottom temperatures.
The primary implication of this work for future water quality and climate change modeling studies is that running the models in a “one-way” coupled configuration underestimates temperature variability, both temporally and spatially. As we showed in our spring freshet example, the largest episodic differences in temperature may occur during a seasonal transition. Additionally, temperature shifts may be sustained over the summer months, as demonstrated by the colder summertime bottom temperatures in the “bgc-light” run. Changes in vertical temperature gradients will have greater impacts on circulation and biogeochemistry in water bodies where temperature plays a large role in density stratification, as in Cahill et al. (2008) and Jolliff and Smith (2014), in contrast to the salinity-stratified Chesapeake Bay.
For models that assume a constant light attenuation coefficient for solar heating that is much different from the observed light attenuation in the water body, temperature differences and impacts on biogeochemistry will be larger in magnitude than reported in this study. It is difficult to assess the prevalence of this inconsistency because many modeling studies do not report the irradiance calculations used for solar heating. We recommend other modelers report their hydrodynamic light attenuation assumption, Jerlov type or otherwise, in their model documentation. For our model runs, the differences in temperature between the two simulations are much smaller than the difference between the model and observations, and neither systematically agrees better with the observations. Therefore, there was no change in the model agreement with in situ measurements of temperature from discrete bi-weekly to monthly measurements by the Chesapeake Bay Program along mainstem stations (see Appendix B).
In this study, we used a parameterization of light attenuation that was highly dependent on salinity, a proxy for CDOM. This additive model of light-attenuating materials allowed us to determine the contributions by different modeled biogeochemical constituents in the oligo- and mesohaline regions of the Bay. While it likely adequately represents the spatial pattern of light absorption by CDOM (Rochelle-Newall and Fisher 2002), this parameterization predicted a light attenuation coefficient that was sometimes smaller than typical values of kd from in situ measurements in the polyhaline region. This is in part due to the lack of explicit resuspension of inorganic sediment in our implementation of ChesROMS-ECB. Here, resuspension is represented implicitly through the imposition of a lower bound on kbgc (0.6m− 1). Future studies should investigate and incorporate the processes that affect water clarity in the lower reaches of the Bay, such as wave-induced sediment resuspension and coastal erosion.
Projected increases in precipitation and streamflow during the winter and spring due to climate change (Najjar et al. 2010; Irby et al. 2018) may bring more intense spring freshet events, depending on the suddenness of snow and ice melt. As shown in our study, decreased water clarity events may be followed by warmer surface temperatures during the spring-to-summer transition from cold to warm temperatures (Fig. 6a). These changes are in addition to the long-term temperature rise that has already been observed and is expected to continue. While warmer temperatures generally increase phytoplankton growth, it can also shift the distributions of specific phytoplankton taxa which could have implications for trophic interactions. As demonstrated in our study, increasing surface temperatures in the upper Bay can affect the distribution of nutrients by decreasing nutrient transport into the meso- and polyhaline regions of the Bay. This resulted in a decline in phytoplankton in these regions of the Bay, decreasing zooplankton populations. These indirect effects on higher trophic levels are poorly understood and warrant further study.
We demonstrated the impacts of variable water clarity on solar heating by showing the difference between the “bgc-light” minus “standard” simulations. The temperature changes associated with increasing water clarity would be the opposite of what we reported in this manuscript; i.e., more solar heating at deeper depths. While the recent recovery of submerged aquatic vegetation (SAV) in the Chesapeake Bay demonstrates the success of management efforts (Lefcheck et al. 2018), warmer subsurface and bottom temperatures in shallow areas may be an unintended consequence of water clarity improvements. Along with rising temperatures from climate change, these changes together may threaten recent gains in SAV restoration. While spectrally resolved light calculations have been incorporated into modeling studies for SAV habitat prediction (del Barrio et al. 2014), the impact of improved calculations for solar heating are unknown. On the other hand, we report how decreases in water clarity may be linked to colder bottom temperatures. Unusually cold winter temperatures can impact other organisms of importance to the Bay as well, such as mortality for blue crabs during severe winters (Bauer and Miller 2010). The interaction between water clarity and solar heating should be incorporated in future modeling studies involving species’ habitability to better simulate environmental temperature variability.