Marjorie Friedrichs, Raleigh Hood and Aaron Bever Comparison of Shallow-water Models for Use in Supporting Chesapeake Bay Management Decision-making Marjorie Friedrichs, Raleigh Hood and Aaron Bever Model Comparison Team SCHISM team Joseph Zhang Harry Wang ROMS-GL team Dick Zimmerman John Klinck Chuck Gallegos Victoria Hill Mike Dinniman ROMS-RCA team Jeremy Testa Damian Brady Ming Li FVCOM & CH3D team Richard Tian Ping Wang Advisory Team: Lewis Linker Carl Cerco Larry Sanford Kevin Sellner

Shallow Water Multiple Model Effort Chesapeake Bay Shallow Water Multiple Model Effort Why focus on the shallow waters of the Chesapeake Bay? This is where we have seen degradation of water quality This is where we are likely to see early responses to management actions Why do we need improved shallow water models? We depend on models to assess the impacts of alternative management strategies Current models (e.g., CH3D) don’t resolve shallow waters very well. Why do we need multiple models of these waters? To increase scientific, management, and stakeholder confidence in the tools used to support and inform partnership collaborative decision making.

Challenges of Modeling Shallow Waters Challenging waters to model! Complex linkages between shallows & land, sediment and open Bay waters Processes vary on small time & space scales, requiring high resolution models Systems respond strongly to distant forcing: multiple spatial/temporal scales

Outline Study site: Chester River tributary Cruise and continuous monitoring data Four participating models Similarities (forcing) and differences (grids) Model performance: hydrodynamics (T, S) water quality (DO, chl, TSM) Sensitivity experiments Outer Boundary Conditions (OBC) Freshwater discharge Summary TSM = Total suspended matter = Total Suspended Solids = TSS

Study Site: Chester River Chesapeake Bay Study Site: Chester River Study site location Chester River

Observations: 2003-2006 A CHE0348 B XIH0077 C XHH4916 ET4.1 B XIH0077 = biweekly cruise data A, B, C = continuous moored data Data available. Today we will be showing results only from 2003, since this is the only year that we have results from all the models. C XHH4916 = CBP WQM data ET4.2 salinity, temperature, oxygen, chlorophyll, TSM

Horizontal resolution Participating models All models use consistent winds, boundary conditions, freshwater & nutrient discharge Model grids differ considerably Participating model Horizontal resolution Horizontal grid Vertical CH3D-ICM* low structured z-grid FVCOM-ICM medium triangular sigma ROMS-RCA high SCHISM-ICM hybrid *CH3D is the regulatory model currently used for management decisions in the Chesapeake Bay *CH3D-ICM is the regulatory model currently used by the CBP for management decisions in the Chesapeake Bay

CH3D Low res. 190 pts FVCOM Medium res. 8349 pts SCHISM Medium res. ROMS High res. 38432 pts Note that FVCOM has more resolution along coastline and SCHISM has more resolution in main channel

Channel depths also differ among models Bottom Depth [m] Note that we are going to show lots of plots that look like this – transects from the Chester River mouth to upstream. data Bay upstream SCHISM channel depths closely follow DEM; FVCOM & CH3D-ICM are shallower.

Simulation Experiments Three simulations for hydrodynamics (S, T) and water quality (chl, DO,TSM): Base Case (CH3D OBC) 10% reduction in watershed inflow SCHISM Outer Boundary Condition (OBC) in place of CH3D OBC

Simulation Experiments Three simulations for hydrodynamics (S, T) and water quality (chl, DO,TSM): Base Case (CH3D OBC) 10% reduction in watershed inflow SCHISM Outer Boundary Condition (OBC) in place of CH3D OBC Can our shallow water models reproduce observed hydrodynamics and water quality in the Chester River? How sensitive are the simulations to input fluxes from the rivers and at the Chester mouth?

Part 1: Hydrodynamic (T & S) multiple model comparisons Color scheme!

Multiple model comparison: temperature Generally models simulate T well throughout transect, regardless of model resolution Surface Temperature All models get T July 15, 2003 – Sep 15, 2003 Sep 15, 2003 – Nov 15, 2003 Bay upstream Bay upstream

Multiple model comparison: Surface Salinity Summer 2003 Fall 2003 Main point is ICM overestimates S in upstream portions of river; only works for the first 40km into the Chester. Bay upstream Bay upstream ICM overestimates surface salinity in upper half of Chester FVCOM is always saltier than SCHISM Summer: ROMS performs best; Fall: FVCOM and SCHISM perform best 2003-2006 results (not shown) indicate that ICM does best in lower Chester, SCHISM does best in mid-Chester and FVCOM does best in upper Chester

Multiple model comparison: Salinity Stratification Summer 2003 Stratification = bottom S minus surface S Bay upstream ICM overestimates stratification in the channel and upstream ROMS & SCHISM often underestimate stratification FVCOM does best?

Multiple model comparison: S Stratification in Channel Summer 2003 Stratification = bottom S minus surface S We don’t have data all along the channel, but we can still compare the models and see that ICM has much higher stratification in the channel than the high resolution models. In reality, stratification is probably less than ICM and greater than the high res models. ICM generates much more stratification throughout Chester

Part 2: Water Quality (Chl, DO, TSM) multiple model comparisons (A work in progress!!) (New – under development!) Color scheme – note that these results are preliminary! Especially the new ICM results No results for ROMS yet.

Multiple model comparison: Surface Chlorophyll Summer 2003 TMDL WQSTM (phase 5.3.2) “new” WQSTM (work in progress) Surface Chl (ug/L) Bay upstream Lots of variability between models SCHISM produces low Chl near ~50km (improved coastline and bathymetry -> longer residence time)

Multiple model comparison: Surface Total Suspended Matter Summer 2003 TMDL WQSTM (phase 5.3.2) “new” WQSTM (work in progress) Surface Total Suspended Matter (mg/L) Bay upstream Models generally underestimate suspended matter “New” WQSTM represents improvement throughout most of Chester

Multiple model comparison: Bottom DO Summer 2003 TMDL WQSTM (phase 5.3.2) “new” WQSTM (work in progress) Bottom DO (mg/L) Bay upstream High resolution models show much improved bottom DO in upper Chester and at Chester mouth ICM models often show too much bottom DO depletion (too much stratification?)

Multiple model comparison: Surface DO Summer 2003 TMDL WQSTM (phase 5.3.2) “new” WQSTM (work in progress) Surface DO (mg/L) Bay upstream Models perform more similarly for surface DO Surface DO in upper Chester for “new” WQSTM needs investigation

Sensitivity Experiments  10% decrease in freshwater inflow  SCHISM vs. CH3D Outer Boundary Conditions

Multiple model comparison: Effect of changing OBC on surface salinity Summer 2003 Tidally-averaged surface salinity (PSU) Bay upstream Effect of change in OBC is felt 45 km up the tributary

Multiple model comparison: Effect of 10% change in freshwater input Summer 2003 Tidally-averaged surface salinity (PSU) Bay upstream Effect of 10% change in freshwater input on salinity is negligible 10% change in nutrient inputs has negligible effect on water quality too (not shown)

Summary Models simulate hydrodynamics relatively well All models reproduce temperature well Low resolution CBP model only works ~40km into the tributary; higher resolution models are needed to reproduce upstream salinity (~0psu) and weak stratification Models are very sensitive to bathymetry Water quality is more challenging to simulate Very preliminary results indicate that: High resolution models produce much better DO in upstream Chester (and at mouth) High resolution bathymetry and coastline affects residence time which may improve chlorophyll simulation All models underestimate TSM; “New” WQSTM simulation seems to produce higher (more realistic) values Sensitivity experiments: 10% reduction of freshwater/nutrient inflow has negligible effect on water quality & hydrodynamics (compared to changing models) Change in salinity OBC affects salinity ~45km up Chester River

Future Work Continue work on 2003-2006 base case comparisons (focusing on water quality) Examine sensitivity to alternate bathymetry Use new FEMA bathymetry Nutrient reduction scenarios (2002-2005) Historically high loads – “1985” Lowest loads ever seen – “all forest” = 90% reduction 2025 reductions – “TMDL” = 50% reduction We have seen that models produce very different water quality results in the shallow Chester River. Do the models produce equally different improvements resulting from nutrient reductions?