Office of Coast Survey / CSDL A Hierarchy of Physical Models for Ecological Applications Lyon Lanerolle 1,2, Richard Patchen 1, Richard Stumpf 3, Frank.

O ffi c e o f C o a s t S u r v e y / C S D L

A Hierarchy of Physical Models for Ecological Applications

Lyon Lanerolle1,2, Richard Patchen1, Richard Stumpf3, Frank Aikman III1, Timothy Wynne3, Michelle Tomlinson3 and Jiangtao Xu1,4

1NOAA/NOS/OCS/Coast Survey Development Laboratory,1315 East-West Highway, Silver Spring, MD 20910; 2Earth Resources Technology (ERT) Inc.,10810 Guilford Road, Suite 105, Annapolis Junction, MD 20701; 3NOAA/NOS/NCCOS/Coastal and Oceanographic Assessment Status & Trends Branch,

1305 East-West Highway, Silver Spring, MD 20910; 4University Corporation for Atmospheric Research/Visiting Scientist Program, P. O. Box 3000, Boulder, CO 80307.

O ffi c e o f C o a s t S u r v e y / C S D L

Introduction and Motivation

• Origin of modeling efforts due to research collaborations attempting to improve HAB and Hypoxia predictions:

1D, 2D HAB set-ups due to NCCOS-CSDL collaboration3D HAB set-up due to NOAA/IOOS partnership projectHypoxia set-up due to SURA Testbed (Estuarine Hypoxia) project

• Hierarchy of modeling approaches due to :Different levels of complexity of processes (e.g. 1D, 2D, 3D, etc.)Nature of dominant physical processes (e.g. vertical mixing,

upwelling, etc.)• All applications are physically forced usually with no behavior;

all based on Rutgers University’s ROMS model• Modeling set-ups have dual purposes : (a) research tool and

(b) real-time, operational forecast system (if needed)

O ffi c e o f C o a s t S u r v e y / C S D L

1D Vertical Mixing ModelMotivation and Model Set-up

• Motivation : to study and predict cyanobacterial bloom/scum formation in water bodies

• Model Design Specifications : (i) highly portable (speedily applied to any water body) and (ii) minimum number of model inputs – a simple grid, representative bathymetric value, a Coriolis parameter, a T and S profile for initialization and a fixed-point time-series of met. variables

• Model set-up : ROMS with Bulk Fluxes, GLS k-ω closure, quadratic bottom drag (Cd=0.003), wall BCs, ∆t=300s (baroclinic)

• Calibration : using idealized fields (thereafter applied to Western Lake Erie)

• Computational Efficiency : 30-day simulation/per minute on LINUX box in serial mode (highly efficient)

Acknowledgment : This work was funded by National Center for Environmental Health at the Centers for Disease Control and Prevention (CDC)

O ffi c e o f C o a s t S u r v e y / C S D L

1D Vertical Mixing ModelWestern Lake Erie Application and Results

1. Periods of high/low viscosity due to met. conditions2. Simulated tracer and particles respond in the expected way

• 9 x 9 horizontal grid• 20 vertical σ-levels• 7.7m flat bathymetry (from Obs.)• IC from linear interp. of a surface and bottom T, S value• Met. forcing from NOAA/NDBC Marblehead, OH station

Marblehead, OH station

O ffi c e o f C o a s t S u r v e y / C S D L

2D Upwelling Transect ModelMotivation and Model Set-up

• Motivation : Try to enhance WFS HAB event predictions of NOS/CO-OPS Bulletins by accounting for upwelling

• Hypothesis : Upwelling contributes to HAB events on WFS

• Model Design Specifications : Set-up model to study upwelling driven flow with cross-shore transport component

• Model Set-up (ROMS): Grid : transect with 400 x 9 points, 80

vertical σ-levels

Bathymetry : NOS soundings

ICs : MODAS/Basin model (NGOM) for T and S and geostrophic velocities

BCs : periodic and far-field radiationForcings : met. forcing only (VENF1-

CMAN and NAM)

Vertical Mixing : GLS k-ω model

Time Step : 150s (baroclinic)

O ffi c e o f C o a s t S u r v e y / C S D L

2D Upwelling Transect ModelModel Output and Results

• Algal cells simulated using Lagrangian particles (blue square – begin, black circle – end)• Model capable of running as Nowcast/Forecast system and generating above graphic• Computational Efficiency : 7-day simulation/hour or better on LINUX box in serial mode

Particles respond to wind-driven upwelling (~31 August)

O ffi c e o f C o a s t S u r v e y / C S D L

3D Nested/Coupled ModelMotivation and Modeling Strategy

• Motivation : Try to enhance predictions of NOS/CO-OPS Bulletin HAB patch extent and movement on WFS - which is inherently 3D in nature

• Model Design Specifications:

Need a fully 3D model of WFS

Need to include enough of shelf

Need Tampa Bay and Charlotte Harbor

Need to be computationally efficient• Modeling Strategy : Nest/Couple high-resolution ROMS

model to already available basin-scale, POM-based NOAA/NOS Gulf of Mexico Model (NGOM)

O ffi c e o f C o a s t S u r v e y / C S D L

3D Nested/Coupled ModelModel Coupling and Set-up

• Grid : Covers WFS and refined along coast, Tampa Bay and Charlotte Harbor; 298 x 254 horizontal points and 30 vertical σ-levels• Bathymetry : NOS soundings (need to match at lateral boundaries) • Initial Conditions : NGOM interpolated water levels, T and S (spin-up from rest)• Boundary Conditions : NGOM interpolated surface forcings (wind stress, air P, heat flux); SST correction; water levels, currents, T and S at lateral boundaries; tides from ADCIRC added; 19 additional rivers (Tampa Bay and Charlotte Harbor) • ROMS details : Coupling BCs, Quadratic bottom drag, GLS k-ω model, ∆t=90 s (baroclinic)• Computational Efficiency : 24-day simulation/hour [MPI, IBM Power 6 cluster , 96 proc.]

NGOM ROMS

O ffi c e o f C o a s t S u r v e y / C S D L

3D Nested/Coupled ModelModel Results

Tracer patch method : Passive/Inert tracer evolution within ROMSParticle tracking method : CSDL’s Chesapeake Bay Oyster Larvae Tracker (CBOLT)

Observed Initial Patch Digitized Initial Patch

Initialization HAB patch

7-day Hindcast

Need 3D velocities as 2D depth-averaged velocities miss near-

shore upwelling behavior

O ffi c e o f C o a s t S u r v e y / C S D L

Water Quality ModelMotivation and Model Set-up

• Motivation : to study & predict spatio-temporal evolution of hypoxia in Chesapeake Bay

• Strategy : begin with simplest WQ model and then build up to complex models

• Model : examine hypoxia via DO using a 1-equation model with constant respiration (Malcolm Scully/ODU)

• Model Set-up : embed DO model within NOAA/NOS Chesapeake Bay Operational Forecast System (CBOFS)

DO in ROMS is a passive/inert tracer• Simulation : synoptic hindcast from June 01,

2003 - August 31, 2005• ICs and BCs : DO saturation from T and S [Weiss

(1970)]; no river DO sources• Computational Efficiency : 6-day sim./hour

[MPI, IBM Power 6 cluster , 96 proc.]

Const. resp. rate of 0.55 gO2/m3/day

Fixed at saturation (surface also)

O ffi c e o f C o a s t S u r v e y / C S D L

Water Quality ModelModel Results (Preliminary)

• Total DO content (kg) diminishes during the summer months as expected• Hypoxic volumes show agreement with those derived from CBP

observations*• Hypoxic zones present in deep, narrow channels during summer months*Courtesy of Malcolm Scully/ODU and Rebecca Murphy/JHU

DO ≤ 2 mg/L at 1m above botm.

DO ≤ 2 mg/L

DO ≤ 1 mg/L

DO ≤ 0.2 mg/L

O ffi c e o f C o a s t S u r v e y / C S D L

Summary and Conclusions• Through collaborative efforts to study and predict (with improved

accuracy) HABs and Hypoxia a suite of physical models have been developed at NOAA/NOS/OCS/Coast Survey Development Lab.

• The models range in complexity depending on the nature of the dominant processes driving HABs and Hypoxia; they span multiple dimensions (1D - 3D)

• Predictive capabilities of the models have been demonstrated and they reveal insights in to the driving physical mechanisms

• These models, although developed as research tools, also have the potential to be cast in to Operational Forecast Systems (OFS) to routinely generate HAB and Hypoxia forecasts :

http://tidesandcurrents.noaa.gov/hab/