Circulation in Narragansett Bay:
Water flow & mixing
Transport of chemical-biological material
Exchange between sub-regions of Bay
Motivation: Episodic hypoxia in NB &
Developing management tools for the
Bay ecosystem
CHRP Coupled Ecological Modeling
(GEMBox - ROMS Model)
Chemical Processes
Biological processes
Fundamental Column:
Detailed understanding of
physical processes
National Capitol Columns at the United States National Arboretum in Washington, D.C.
Role of physical processes in the health of
the Bay
• Physical Drivers:
– Tides
– Winds
– River discharge
– Density differences
http://www.geo.brown.edu/georesearch/insomniacs
Role of physical processes in the health of
the Bay
• Physical Drivers:
– Tides
– Winds
– River discharge
– Density differences
• Out with the bad
http://www.geo.brown.edu/georesearch/insomniacs
Role of physical processes in the health of
the Bay
• Physical Drivers:
– Tides
– Winds
– River discharge
– Density differences
• Out with the bad
• In with the good
http://www.geo.brown.edu/georesearch/insomniacs
Circulation in Narragansett Bay:
3 Basic Methods
Hydrographic data (currents, density)
Numerical Modeling (ROMS : ( Regional Ocean Modeling System)
Physical - Laboratory Modeling
Physics: Observations
Acoustic Doppler Current Profilers
Bottom mounted
Data coverage:
Excellent temporal
Poor Spatial
Ship mounted / underway
Data coverage:
Good spatial
Poor Temporal
Tilt Current Meters :
(tilt proportional to water velocity)
Water flow
Buoyant cylinder
Flexible membrane
Mooring weight
Numerical Model
Shallow Water Equations
Momentum balance x & y directions:
u + vu – fv = f + Fu + Du
t
x
v + vv + fu = f + Fv + Dv
t
y
Potential temperature and salinity :
T + vT = FT + DT
t
S + v S = FS + DS
t
The equation of state:
r= r (T, S, P)
Vertical momentum:
f = - r g
z
ro
Continuity equation:
u + v + w = 0
x y z
Initial Conditions
Forcing Conditions
ROMS Model
Regional Ocean
Modeling System
Output
Fluid Dynamics Laboratory Models
(a.k.a analog models)
Save the
Bay
Hurricane
Barrier
Fall, 2008 at GFD Lab, Australian
National University:
.
Developed flume tank to represent
the upper Providence River
Lab models provide excellent check
against numerical models
Shipping Channel
Save
the
Bay
Port
Edgewood
Edgewood Shoal
Circulation in Narragansett Bay:
3 Basic Methods
Lots of great students!!
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
Circulation in Narragansett Bay:
3 Basic Methods
Lots of great students
Heaps of excellent support
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
Quick summary:
PR
RI Sea Grant
GB
MHB
NB
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
1999-2001
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
RIS
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
Narragansett Bay Commission
2001-2005
PR
GB
MHB
NB
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
99-01
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
RIS
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
Narragansett Bay Commission
PR
& CHRP award, 2006
2005
GB
2006
MHB
NB
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
99-01
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
RIS
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
Narragansett Bay Commission
RI Sea Grant
& CHRP award, 2007
PR
2005
GB
2006
MHB
NB
2007
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
99-01
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
RIS
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
RI Sea Grant &
CHRP award, 2008
PR
2005
GB
2006
MHB
NB
2007
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
99-01
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
RIS 2008
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
RI Sea Grant &
CHRP award, 2009
PR
Most ambitious physical
sampling survey ever…..
2005
GB
2006
Focus on key CHRP Dye
Boxes
2009
MHB
NB
2007
Bridget Sullivan (1st developed SCRUM, then ROMS)
William Deleo
(data at Bay Mouth, detailed study of Mt. Hope Bay)
99-01
Kurt Rosenberger (data at Bay Mouth and in Rhode Island Sound)
Deanna Bergondo
(develop ROMS model/data for Providence River, initial CHRP proposal)
Justin Rogers
(ROMS, Mid-Bay Data )
Nicole LaSota
(ROMS Prov. River dye dispersion study)
Anna Pfeiffer-Herbert
Christelle Balt
RIS 2008
(Detailed data RIS, mouth, Greenwich Bay, Bay-Shelf Exchange)
(Greenwich Bay & Prov. River Data, ROMS flushing exeriments, Mixing schemes)
Data tell a physical story……..
Providence River Model
Full Bay Model
Domain
RIS-NB Model
Domain(Rogers, 2008)
Data Example
Circulation and Transport in Upper
Narragansett Bay
Justin Rogers
MS Thesis,08
West Passage (Tidal flows)
North - IN
Surf.
Bot.
Fast
in,
Faster
out
East Passage Channel (Tidal flows)
North - IN
Slow
in,
Slower
out
Decimal Day in 2006 (summer)
Examples what these data look like:
Data tell a physical story……..over lots of different scales (different forcing)
Providence River Model
Full Bay Model
Domain
RIS-NB Model
Domain(Rogers, 2008)
Residual (de-tided) Flow
West Passage
On-average…southward
Surface
Bottom
Residual (de-tided) Flow
West Passage
On-average…southward
Surface
Bottom
On-average…northward
East Passage Channel
Physical Story of the Bay, by Residual Q. Data
Water enters eastern EP
Dominant re-supply up EP
Outflow through WP*
------------------------------------------------Enhance:
Eastward winds
Southward winds
Northward, then Southward winds
Stall/reverse:
Northwestward winds stall / reverse
?
We have a very good idea about deep intrusions
where they come from, what drives them, and their thermal advection
Steady Re-supply up East Passage
Enhanced by:
Eastward winds
Southward winds
Northward, then Southward winds
Deep EP
?
Bottom Temperature @ ADCP
Data tell a physical story….
Data always limited in space and/or time…..
Models extend us towards full physical story of the Bay, all places, for all time
Providence River Model
Full Bay Model
Domain
RIS-NB Model
Domain(Rogers, 2008)
Data tell a physical story….
Data always limited in space and/or time…..
Models extend us towards a full physical story of the Bay, all places, for all time
-----------------------------------------------------------------------------------------------------Providence River Model
Models….do we trust them?
do the data and model wiggles match?
Full Bay Model
Domain
RIS-NB Model
Domain(Rogers, 2008)
ROMS Modeling for Narragansett Bay is built on years of GSO student efforts
Bergondo: Providence River model
Rogers: Bay - Rhode Island Sound Model
Rogers, Ullman, Balt: Full Narragansett Bay Model
Providence River Model
Full Bay Model
Domain
RIS-NB Model
Domain(Rogers, 2008)
ROMS Modeling for Narragansett Bay is built on years of GSO student efforts
Bergondo: Providence River model
Rogers: Bay - Rhode Island Sound Model
Rogers, Ullman, Balt: Full Narragansett Bay Model
Providence River Model
2006 wind, runoff, air conditions
Tides, mouth forced by large model
Full Bay Model
Domain
15 terrain-following vertical coordinates
<50 meter horizontal grids
RIS-NB Model
Domain(Rogers, 2008)
Models match instantaneous (tidal) flows/heights very well
Data ( R) vs Model (B)
Tidal response
Water Level
Skill = .97
Skill = .98
Skill = .98
Decimal Day, 2006
Validation using fixedpoint data
2006
VARIABLE
SKILL
Surface
Temperature
0.98
Surface Salinity
0.94
Bottom
Temperature
0.79
Bottom Salinity
0.76
From C. Balt,09
ROMS Assessment : Quantitative comparisons with fixed station data
Actual forcing data
ROMS Model
Time series model
output at buoy
locations
+
Time series buoy data
Statistical metrics:
Skill = 0 poor
Skill = 1 perfect
Tide heights: Skill >0.95 // Velocity fields: Skill 0.8 - 0.9 // T,S: Skill 0.75-0.98
Data tell a physical story….
Data always limited in space and/or time…..
Models extend us towards a full physical story of the Bay, all places, for all time
-----------------------------------------------------------------------------------------------------Providence River Model
Models….do we trust them?
do the data and model wiggles match?
does the model predict the gross character shown in data?
Full Bay Model
Domain
Modeled upper bay transport: Vertically integrated
along N. Prudence Line
--- Zero wind
Summer 2006 forcing, except winds
Modeled upper bay transport: Vertically integrated
along N. Prudence Line
--- Zero wind
--- Northward wind GYRE STALLS
Modeled upper bay transport: Vertically integrated
along N. Prudence Line
IN
OUT
--- Zero wind
--- Northward wind GYRE STALLS
--- Southwestward wind GYRE ENHANCED
Patterns match 06-09 ADCP data
Data tell a physical story….
Data always limited in space and/or time…..
Models extend us towards a full physical story of the Bay, all places, for all time
-----------------------------------------------------------------------------------------------------Providence River Model
Models….do we trust them?
do the data and model wiggles match?
does the model predict the gross character shown in data?
Full Bay Model
Domain
Quasi-trusted models….how can we RIS-NB
useModel
them?
Domain(Rogers, 2008)
Key Management Issue: The role of flushing dynamics as
the root of chronic water quality problems?
Out with the Bad
http://www.geo.brown.edu/georesearch/insomniacs
Flushing Time using Fraction of Water Method
(assume complete mixing)
Simple estimates: ~4 days
Simple estimates: ~10 days
Flushing processes may be more complex…….
Greenwich Bay:
Rogers thesis models using ROMS show longer times
Future: Test model predictions with dense current meter network
Providence River:
NBC data & Fluid Dynamics Lab Models suggest longer times
Future: Test predictions using dense current meter network
Flushing processes may be more complex…….
Greenwich Bay:
Rogers thesis models using ROMS show longer times
Future: Test model predictions with dense current meter network
Providence River:
NBC data & Fluid Dynamics Lab Models suggest longer times
Future: Test predictions using dense current meter network
CHRP models must represent flushing accurately
Providence River flushing may not follow simple estimates
Models & Data Suggest a Bimodel Residence Time
NBC
Save the
Bay
Edgewood
shoals
EYC
channel
Fluid Dynamics Laboratory Models
(a.k.a analog models)
Save the
Bay
Hurricane
Barrier
Fall, 2008 at GFD Lab, Australian
National University:
.
Developed flume tank to represent
the Providence River
Model includes shipping channel &
the Edgewood Shoals
Parameters: river flow , NBC
discharge, tides : (no winds)
Shipping Channel
Save
the
Bay
Port
Edgewood
Edgewood Shoal
We have modified a 3 meter long flume tank at the Fluid Dynamics Lab of the
Australian National University to represent the Providence River-Fields PointEdgewood region of the estuary.
NBC
NBC
Port
Port
Edgewood
66 cm (1000m)
Z= 2cm
33 cm
(400 m)
Shoal
Channel
Shoal
Channel
Broad
Edgewood
Broad
Scaling:
Re#= 105
Scale on runoff prism and tidal prism :
30 CMS = 7 liters/min
1 tide cycle = 20 seconds
Shoal
Broad
NBC
Channel
Relevance to Water Quality Models
Lab Models: small physics that numerical models can’t represent
Do small scale processes influence exchange & mixing between
shoal and channel & overall water quality?
Shoal
Broad
NBC
Channel
Movies: no wind/density forcing
1.
30 CMS (7liters/min), no tide (2 CMS NBC Fields Pt).
2.
2 CMS runoff, 1.5 meter tide
3.
30 CMS & 1 meter tide (late addition of 2 CMS NBC outfall)
Shoal
Broad
NBC
Channel
Do small scale processes
influence exchange &
mixing? YES
Eddies 10-50 meters
Lab Models: Discharge, no tide
Small scale eddies: tilted/sheared
Vertical & horizontal structure to
shoal-channel exchange:
>15 days: shoal bottom
water retention
Surface floaters rapid exchange,
but recycle
Do small scale processes influence exchange & mixing? YES
Spring tide only, no runoff:
Mid-shoal surface off in 4 cycles (2 days)
Mid-shoal deep >> 10 cycles (5 days)
Inside shoal >> 10 cycles (5 days)
Combined 30 CMS runoff & 1m tide: flushing ~5 cycles
Two exchange modes
1. tilted-stretched eddies
Combined 30 CMS runoff & 1m tide: flushing ~5 cycles
Two exchange modes
1. tilted-stretched eddies
2. late flood, eastward plumes
Conclusions:
Lab models show:
two modes of shoal-channel exchange
deformed eddies at shoal-channel interface
tides & runoff drive off-shoal plumes
strong vertical flow structures, deep shoal water isolation
flushing times of 5-20 tide cycles
Mixed basin model for Narragansett Bay is not appropriate
NBC outfall on shoals increases flushing by factor of 3
Tilt Current Meters: 1) Gyre vorticity / shape vs. environmental forcing
2) Small scale eddy field at channel-shoal intersection
x
x
Testing Model Predictions
x
Flushing processes may be more complex…….
Greenwich Bay:
Rogers thesis models using ROMS show longer times
Future: Test model predictions with dense current meter network
Providence River:
NBC data & Fluid Dynamics Lab Models suggest longer times
Future: Test predictions using dense current meter network
Retention in Greenwich Bay: Wind matters
Position of floats after 10 days of simulation after float cloud introduced,
No wind
No sea
breeze
NNE-ward wind
Applied sea
breeze
summer 2006
J.M. Rogers
Movies of Greenwich Bay Flushing:
Narraganset Bay-RIS ROMS model.
Summer 2006 tides & density fields
1) 2006 Runoff, no wind
2) 2006 Runoff, northwestward wind
Retention in Greenwich Bay: Wind matters
Residual flows are predicted to be distinctly in the two cases.
No wind
No sea
breeze
NNE-ward wind
Applied sea
breeze
summer 2006
J.M. Rogers
Retention in Greenwich Bay: Wind matters
Residual flows are predicted to be distinctly in the two cases.
No wind
Test with data
No sea
breeze
NNE-ward wind
Applied sea
breeze
summer 2006
J.M. Rogers
Flushing processes may be more complex…….
Greenwich Bay:
Rogers thesis models using ROMS show longer times
Future: Test model predictions with dense current meter network
Providence River:
NBC data & Fluid Dynamics Lab Models suggest longer times
Future: Test predictions using dense current meter network
CHRP models must represent flushing accurately
Flushing processes may be more complex…….
CHRP Coupled Eco-model
Greenwich Bay:
Rogers thesis models using ROMS show longer times
Future: Test model predictions with dense current meter network
Fundamental Column:
Providence River:
Detailed understanding of
physical processes
NBC data & Fluid Dynamics Lab Models suggest longer times
Future: Test predictions using dense current meter network
CHRP models must represent flushing accurately