Hydrodynamic Modeling of Tides and Hurricane Storm Surge for Pre- and Post-Dredging Conditions in the Lower St. Johns River, Florida Matthew V. Bilskie 1 1 PhD Student, University of Central Florida, Department of Civil, Environmental, and Construction Engineering, P.O. Box 162450, Orlando, FL, 32816, USA. Phone: 407-823-4843. E-mail: Matt.Bilskie@gmail.com Abstract The United States Army Corps of Engineers (USACE), Jacksonville District, is conducting a study for improving navigation near Jacksonville Harbor in the lower St. Johns River (SJR), Florida. A two-dimensional hydrodynamic model (ADCIRC) of the SJR is employed to study the effects that deepening the channel may have on circulation within the river. Model results from an inlet-based modeling domain, forced with astronomic tides, show low sensitivity to mean low water (MLW) and mean high water (MHW) between preand post-dredging conditions. Results from a large-scale hydrodynamic model (ADCIRC+SWAN), forced by astronomic tides and winds from Hurricane Dora, show minimal difference in peak surge, timing of peak surge, and inundation area. The model is then used to determine possible impacts a 30 cm rise in sea level may have on flooding. Model results demonstrate that peak surge elevations do not increase by the rise in sea level, depicting the dynamic nature of the estuary. Keywords: St. Johns River; Hydrodynamic; Shallow Water Equations; Sea Level Rise; ADCIRC+SWAN Introduction and Objective The United States Army Corps of Engineers (USACE), Jacksonville District, is conducting a study for improving navigation near Jacksonville Harbor in the lower St. Johns River (SJR), Florida. The objective is to assess the impacts that channel modifications have on the circulation in the river and the environmental impacts of any changes. Three experiments are performed to assess the impact of widening and dredging the lower St. Johns River. First, an inlet-based hydrodynamic modeling approach is used to compute mean low water (MLW) and mean high water (MHW) for the Timucuan marsh near the confluence of the SJR and the Atlantic Intracoastal Waterway (AICWW). Changes to MLW and MHW may affect salt marsh productivity (Hagen, et al., 2013). Second, a large-scale modeling approach is employed to study pre- and post-dredging impacts on hurricane storm surge flooding extent and peak surge. Third, the large-scale method is re-evaluated, taking into consideration a rise in sea level of 30 cm (1 ft). Study Area The study area is Jacksonville Harbor, located in the lower SJR, in northeastern Florida (Figure 1). The harbor extends from the mouth of the river, where it connects to the Atlantic Ocean, to river mile 20, near the end of the Jacksonville Port Authority Talleyrand Terminal. The AICWW intersects the SJR near river mile 5. Along this confluence and in the lower basin, Spartina alterniflora salt marsh thrives (Brody, 1994). Daily flooding of the salt marsh occurs as the marsh table elevations are near 0 NAVD88 (North American Vertical Datum of 1988) (Hagen, et al., 2013). Much of the salt marsh wetland is found along the AICWW, near the mouth of the river (Timucuan Ecological Preserve) (Figure 2). 1
Figure 1 - Jacksonville Harbor Federal Project ADCIRC+SWAN Model ADCIRC (Advanced Circulation) solves the depth-integrated shallow water equations for water levels and currents using unstructured finite elements in space (Luettich and Westerink, 2006). SWAN (Simulating Waves Nearshore), a third generation wave model, governed by the action-balance equation (wave-current interaction), solves for relative frequency and wave direction, σ and θ, respectively (Booij, et al., 1999, Holthuijsen, 2007): cσ cθ S N + x ( cg + cθ ) N + N + N = t θ σ σ (1) where N is the action density spectrum, x is the gradient in geographic space, c g is the propagation velocities in geographic space, and c θ is the propagation velocities in θ space. The ADCIRC and unstructured SWAN model (ADCIRC+SWAN) are coupled and operate on the same unstructured mesh, therefore removing grid-to-grid interpolation. ADCIRC forces the SWAN model by providing water levels, currents, and wind speeds. SWAN computes the wave-related processes and forces, in part, ADCIRC with wave radiation stress gradients (Dietrich, et al., 2011). ADCIRC is forced with astronomic tides, winds, and pressures. Method Astronomic Tides An inlet-based ADCIRC model of the lower SJR is employed to simulate astronomic tides, harmonically decompose the tidal constituents, and compute spatially varying MLW and MHW over an 18.6 year period (tidal epoch). MLW is the average of all of the low water heights observed over a tidal epoch and MHW is the average of all of the high water heights observed over a tidal epoch (Center for Operational and Oceanographic Products and Services, 2011). 2
Herein, a previous model of the SJR (Hagen, et al., 2013) is modified by enhancing mesh resolution, topography, and bathymetry within the main river channel and in nine intertidal salt marsh regions (priority areas): 1) Timucuan salt marsh; 2) Ortega River; 3) Julington Creek; 4) Doctors Lake and Black Creek; 5) Dunn Creek; 6) Broward River; 7) Trout River; 8) Arlington River; and 9) Trout and Six Mile Creek (Figure 3). Mesh resolution ranges from over 500 m offshore, to 80 m in the main channel, and up to 20 m in tidal creek and salt marsh regions. Figure 2 2009 St. Johns River Water Management District Land Cover classification Model bathymetry and topography is presented in Bacopoulos et al. (2012). Source bathymetric data in the main river channel for existing and post-dredging (project depth) conditions was obtained through the USACE, Jacksonville District. Elevations for the nine updated salt marsh regions were linearly interpolated from the FEMA (Federal Emergency Management Agency) northeast Florida Georgia storm surge study ADCIRC+SWAN mesh. Tidal forcing is applied across the open ocean boundary in addition to the north and south AICWW boundaries. Amplitudes and phases for the seven dominating tidal constituents (M2, S2, N2, K1, O1, K2, and Q1) were obtained from the ADCIRC South Atlantic Bight tidal model (Bacopoulos, et al., 2011). Water levels and currents are simulated for 45 days and water levels are harmonically analyzed using 23 tidal constituents (Luettich, et al., 1992). The tidal resynthesis includes more constituents than being forced because additional constituents are formed due to the non-linear interaction of the tides (compound tides and overtides) as they approach shallow waters and complex geometries of the estuary (Hagen, et al., 2013). 3
Two experiments are performed using this approach. First, existing pre-dredged conditions are simulated, and second, the main river channel bathymetry is updated to represent post-dredging project depths. Results are then analyzed to determine the impact that dredging the main channel has on MLW and MHW. The method of computing MLW and MHW is similar to that of Hagen et al. (2013). The simulated tides are resynthesized for all areas that are continuously wetted, in which MLW and MHW are computed from the resynthesized tidal harmonics. D B A C Figure 3 (A) Inlet-based unstructured finite element of the SJR; (B) Pre-dredged model bathymetry and topography; (C) Difference in pre- and post-dredging (project depth) bathymetry of the main channel; (D) Location of the nine salt marsh regions updates in the inlet-based SJR model Hurricane Storm Surge A large scale ADCIRC+SWAN model with focus on the Georgia and northeastern Florida coast, including the St. Johns River, is employed to simulate hurricane storm surge with pre- and post-dredged conditions in the main river channel (Figure 4). The FEMA northeast Florida Georgia (neflga) ADCIRC+SWAN model was updated to include the aforementioned inlet-based SJR ADCIRC model, herein referred to as the neflga-lsjr model. This included cutting a hole in the neflga mesh, as well as a 300 m buffer area to provide proper mesh element size transition, and inserting, as fill, the SJR river mesh (Figure 5). Further, several other regions in the inlet-based mesh were modified to provided sufficient element size transition to the coarser neflga model. The result is a seamless large scale model of the St. Johns River and surrounding floodplain (Figure 1). 4
Figure 4 Bathymetry of the large-scale ADCIRC + SWAN model To study the affect of post-dredging conditions on hurricane storm surge, simulations of Hurricane Dora (1964) were performed using the neflga-lsjr SWAN+ADCIRC model. Hurricane Dora made landfall on September 10, about 10 miles north of St. Augustine Florida as a category 3 storm, creating, at the time, the largest water levels and winds in Jacksonville in over 80 years. Tides were ramped over a 15 day period, followed by a 9-day simulation of Hurricane Dora for a total simulation length of 24 days (August 19 to September 12, 1964). Amplitudes and phases for the eight dominating constituents (M2, S2, N2, K1, O1, K2, Q1, and P1) were forced at the open ocean boundary (60⁰ W meridian) along with winds and pressure from Hurricane Dora. Maximum water levels, inundation extent, and time-series of water levels (hydrographs) at specified locations were analyzed to show dredging the main channel may alter hurricane storm surge. Sea Level Rise To study the possible impact of future scenarios on existing and post-dredged conditions, Hurricane Dora was simulated using a sea level rise (SLR) scenario of 30 cm (1 ft). SLR was included in the model by offsetting the initial sea surface, by a given SLR, from the geoid. Maximum water levels, inundation extent, and hydrographs were examined. Plots of relative maximum water levels were generated to determine the spatially dynamic and non-linear interaction of the hydrodynamics to a rise in sea level: ζ R ζ ζ SLR SLR C = (2) where ζ R is the relative rise in sea level at each mesh node (unitless), ζ SLR is the maximum water level computed at each node under a given SLR, ζ C is the maximum water level computed under current conditions, and SLR is the rise in sea level (i.e. 30 cm). 5
Figure 5 FEMA neflga ADCIRC+SWAN model (green) with inlet-based SJR ADCIRC model (blue) boundary removed Jacksonville St. Johns River Results Figure 6 Topography and bathymetry of the neflga-lsjr unstructured mesh. Focus is on the lower St. Johns River Astronomic Tides MLW and MHW for the inlet-based SJR ADCIRC model were previously validated by Hagen et al. (2013) at two observation stations, Mayport and Clapboard Creek (Figure 7A). In both pre- and post-dredging conditions, MLW varies from about -80 cm NAVD88 in the Atlantic Ocean to about -65 cm NAVD88 in the main river channel and -30 to -50 cm NAVD88 up tidal creeks (Figure 7). MHW varies from about +80 cm NAVD88 in the Atlantic Ocean to about +78 cm NAVD88 in the main channel between the inlet and AICWW confluence, and down to +60 cm NAVD88 in tidal creeks. As the tidal influence decreases upstream, MLW and MHW approach 0 NAVD88. There is high spatial variability of both MLW and MHW within the tidal 6
A B Clapboard Creek Mayport Existing MLW Existing MHW C D Post-Dredging MLW Post-Dredging MHW E F Difference in MLW Difference in MHW Figure 7 Spatially computed (A) MLW and (B) MHW for existing conditions; (C) MLW and (D) MHW for post-dredging conditions; (E) difference in MLW and (F) difference in MHW between post-dredging and existing conditions 7
creeks. The AICWW confluence with the SJR has an influence on MHW and MLW, as MLW is lower and MHW is larger in the tidal creeks surrounding this region. The gradient in MLW and MHW is observed between the tidal creeks directly connected to the main river channel and AICWW and tidal creeks found farther upstream in the intertidal zone. Differences in MLW between existing and post-dredging conditions range from -5 cm NAVD88 and +3 cm NAVD88 (Figure 7). A +3 cm NAVD88 difference is observed upstream as far as Ortega River and Dunn Creek, about 48 river km (~30 river miles). A large difference in observed near the Broward River. The proposed dredged channel ends downstream of the Broward River, leaving a relatively higher gradient in river bathymetry (Figure 3). This is likely the cause of the high difference in MLW. Further, MLW is increased by +2 cm south of the river in the AICWW. MHW differences range from 0 (no difference) to +5 cm from upstream of the inlet to Palatka, FL, about 150 river km (~90 river miles). As expected, little to no differences in MLW and MHW is observed offshore. Although differences on the order of cm are found, these are minor and are within the uncertainty of measurements ( ± 7 cm) (Hagen, et al., 2013). Hurricane Storm Surge The neflga-lsjr ADCIRC+SWAN model was constructed using the most recent and best available bathymetry and topography and was validated for Hurricanes Dora and Frances (Bilskie and Hagen, 2013). Therefore, the model represents existing conditions, and impacts of dredging the main channel with respect to hurricane storm surge can be examined. Hurricane Dora produced surge elevations of 2.5 m NAVD88 near the mouth of the river and Timucuan salt marsh (Figure 8A). Higher storm surge (3.5 4 m NAVD88) were observed north of the inlet, along the AICWW. Maximum surge and inundation extent for post-dredging conditions were nearly identical to pre-dredging (Figure 8B). The dredged channel did produce slightly higher water levels near Bartram Island and Guana River (north of St. Augustine, FL) (Figure 8C). The 15 cm increase in water level near Bartram Island is due to the high bathymetric gradient at the end of the dredged channel, forcing momentum into a smaller cross-section Negligible differences were observed in maximum water levels for pre- and post-dredging conditions during Hurricane Dora. The timing of the surge levels were examined along the river channel and in tidal creeks, resulting in negligible differences at all locations (Figure 9). Sea Level Rise Plots of RSLR for a SLR scenario of 30 cm were generated to examine how SLR coupled with hurricane storm surge from Hurricane Dora may affect inundation extent and maximum water levels. Results were comparable between existing and post-dredging conditions. A rise in sea level increased maximum storm surge and introduced new regions to flooding (Figure 10). Dynamic, non-linear, interactions in water level and flooding extent under a 30 cm rise in sea level were observed (Figure 10C). White areas represent a static rise (i.e. maximum surge was increased by exactly 30 cm), areas that are blue have maximum water levels less than 30 cm, and areas with yellows and reds have maximum surge larger than 30 cm. Offshore, a static rise is observed; however, within the estuary the interactions between the SJR, AICWW, and surrounding floodplain result in dynamic changes in water level. Because SLR introduces new regions to flooding, peak surge in most of the SJR and AICWW is less than the rise in sea level; however, within areas that peak surge exceeded the rise in sea level, there were high gradients in topography, where the rise in sea level was not enough to initiate new flooding. Conclusions The USACE, Jacksonville District is conducting a study to determine the impacts on circulation in the SJR from channel deepening designed to improve Jacksonville Harbor navigation. An inlet-based and largescale two-dimensional hydrodynamic model was applied to assess possible changes in circulation within the SJR, tidal creeks, and salt marsh system. The models were forced with variations of astronomic tides, hurricane storm surge, wind-waves, and SLR. It was found that dredging the channel to a depth of 47 ft MLLW 8
minimally alters MLW and MHW, which is within the uncertainty of the computed MLW and MHW at Mayport and Clapboard Creek tidal stations; however, there is a large spatial variability of MLW and MWH throughout the estuary. Simulations of Hurricane Dora for pre- and post-dredging conditions did not yield differences in peak surge, timing of the peak surge, or inundation area; therefore, peak surge (of tracks and characteristics similar to Hurricane Dora) is not sensitive to changes in river depth. Further, simulations of Hurricane Dora under a 30 cm rise in sea level show non-linear interactions in the estuary between the SJR, AICWW, and salt marsh system, all of which indicate a need to perform dynamic assessments of sea level rise. Maximum surge elevations were not spatially increased by the rise in sea level. A B C Figure 8 - Simulated maximum inundation extent and peak surge from Hurricane Dora under (A) currents conditions, (B) post-dredging conditions, and (C) the difference A B Figure 9 Water levels during Hurricane Dora under pre- (black) and post-dredging (blue) conditions for (A) the mouth of SJR and (B) near river mile 20, south of Talleyrand Terminal (Figure 1) Acknowledgements This research was funded in part under Contract No. W912EP-10-D-0011 from Taylor Engineering, Inc. and the USACE. Computational resources and support that have contributed to results reported herein were provided from the University of Central Florida Stokes Advances Research Computing Center (URL: http://webstokes.ist.ucf.edu). The author also wishes to thank Scott C. Hagen, Peter Bacopoulos, Steve Bratos, and Chris Bender for sharing their modeling knowledge on the SJR. The statements and conclusions are those of the author and do not necessarily reflect the views of Taylor Engineering Inc., USACE, or their affiliates. 9
A B C Figure 10 Simulated maximum inundation extent and peak surge from Hurricane Dora under (A) currents conditions, (B) a sea level rise of 30 cm, (C) and the relative difference [Eq.(2)] References Bacopoulos, P., Hagen, S. C., Cox, A. T., Dally, W. R., and Bratos, S. (2012). "Observation and simulation of winds and hydrodynamics in St. Johns and Nassau Rivers." Journal of Hydrology, 420-421, 391-402. Bacopoulos, P., Parrish, D. M., and Hagen, S. C. (2011). "Unstructured mesh assessment for tidal model of the South Atlantic Bight and its estuaries." Journal of Hydraulic Engineering, Special Issue on Coastal Maritime Hydraulics, 49(4), 487-502. Bilskie, M. V., and Hagen, S. C. (2013). "Summary of Calibration and Validation Results." ADCIRC Storm Surge Modeling for Jacksonville Harbor Navigation Channel Design, University of Central Florida, Orlando, FL, 17. Booij, N., Ris, R. C., and Holthuijsen, L. H. (1999). "A third-generation wave model for coastal regions 1. Model description and validation." Journal of Geophysical Research, 104(C4), 7649-7666. Brody, R. W. (1994). "Volume 6 of the lower St. Johns River Basin reconnaissance: Biological resources." Technical Rep. SJ94-2, St. Johns River Water Management District, Palatka, FL. Center for Operational and Oceanographic Products and Services (2011). "Tides and Currents: Tidal Datums." <http://tidesandcurrents.noaa.gov/datum_options.html>. (2013). Dietrich, J. C., Zijlema, M., Westerink, J. J., Holthuijsen, L. H., Dawson, C. N., Luettich, R. A., Jensen, R. E., Smith, J. M., Stelling, G. S., and Stone, G. W. (2011). "Modeling hurricane waves and storm surge using integrally-coupled, scalable computations." Coastal Engineering, 58, 45-65. Hagen, S. C., Morris, J. T., Bacopoulos, P., and Weishampel, J. F. (2013). "Sea-Level Rise Impact on a Salt Marsh System of the Lower St. Johns River." J. Waterway, Port, Coastal, Ocean Eng., 139(2), 118-125. Holthuijsen, L. H. (2007). Waves in oceanic and coastal waters, Cambridge University Press, Cambridge. Luettich, R. A., and Westerink, J. J. (2006). "Formulation and Numerical Implementation of the 2D/3D ADCIRC Finite Element Model Version 44.XX." <http://adcirc.org/documentv46/adcirc_title_page.html>. Luettich, R. A., Westerink, J. J., and Scheffner, N. W. (1992). "ADCIRC: An Advanced Three-Dimensional Circulation Model For Shelves, Coasts, and Estuaries, I: Theory and Methodology of ADCIRC-2DDI and ADCIRC-3DL." U.S. Army Corps of Engineers. 10