U.S. IOOS Testbed Comparisons:Hydrodynamics and Hypoxia
Marjy FriedrichsVirginia Institute of Marine Science
Including contributions from the entire
Estuarine Hypoxia Testbed team
With special thanks to Aaron Bever
Four Teams:
Cyberinfrastructure TeamCoastal Inundation Team
Shelf Hypoxia Team (Gulf of Mexico)Estuarine Hypoxia Team (Chesapeake Bay)
U.S. IOOS Modeling Testbed
Estuarine Hypoxia Team:Carl Friedrichs (VIMS)
Marjorie Friedrichs (VIMS)Aaron Bever (VIMS)
Jian Shen (VIMS)Malcolm Scully (ODU)
Raleigh Hood/Wen Long (UMCES, U. Md.)Ming Li (UMCES, U. Md.)
John Wilkin/Julia Levin (Rutgers U.)Kevin Sellner (CRC)
Federal partnersCarl Cerco (USACE)
David Green (NOAA-NWS)Lyon Lanerolle (NOAA-CSDL)
Lew Linker (EPA)Doug Wilson (NOAA-NCBO)
U.S. IOOS Modeling Testbed
To help improve operational and scenario-based modeling of hypoxia in Chesapeake Bay
Methods:1. Compare hindcast skill of multiple CB models on
seasonal time scalesHydrodynamics & dissolved oxygen (2004 and 2005)
2. Generate metrics by which future models can be tested
Overarching Goal:
U.S. IOOS Modeling Testbed
Outline
• What CB models and metrics are we using? – 5 Hydrodynamic models and 5 Biological (DO) models– RMSD; target diagrams
• What is the relative hydrodynamic skill of these CB models?
– Is this a function of resolution? Forcing?
• What is the relative DO skill of these CB models?
– Is this a function of model complexity?
• Summary and outlook for forecasting
Outline
• What CB models and metrics are we using? – 5 Hydrodynamic models and 5 Biological (DO) models– RMSD; target diagrams
• What is the relative hydrodynamic skill of these CB models?
– Is this a function of resolution? Forcing?
• What is the relative DO skill of these CB models?
– Is this a function of model complexity?
• Summary and outlook for forecasting
Methods (i) Models: 5 Hydrodynamic Models (so far)
(& J. Wiggert/J. Xu, USM/NOAA-CSDL)
Biological models: o ICM: CBP model; complex biologyo bgc: NPZD-type biogeochemical modelo 1eqn: Simple one equation respiration
(includes SOD)o 1term-DD: depth-dependent respiration
(not a function of x, y, temperature, nutrients…)o 1term: Constant net respiration
Multiple combinations: o CH3D + ICMo EFDC + 1eqn, 1termo CBOFS2 + 1term, (1term+DD soon!)o ChesROMS + 1term, 1term+DD, bgc
Biological-Hydrodynamic models
Outline
• What CB models and metrics are we using? – 5 Hydrodynamic models and 5 Biological (DO) models– RMSD; target diagrams
• What is the relative hydrodynamic skill of these CB models?
– Is this a function of resolution? Forcing?
• What is the relative DO skill of these CB models?
– Is this a function of model complexity?
• Summary and outlook for forecasting
Data from 40 CBP stations
mostly 2004some 2005 results
bottom T, bottom S, stratification = max dS/dz,
depth of max dS/dz
bottom DO, hypoxic volume
= ~40 CBP stations used inthis model-data comparison
o bottom temperatureo bottom salinityo maximum stratification (dS/dz)o depth of maximum stratification
Hydrodynamic Model Comparisons
Use consistent forcing for each model to examine model skill in hindcasting spatial & temporal variability of:
Bottom Temperature (2004)
Models all successfully reproduce seasonal/spatial variability of bottom temperature (ROMS models do best)
unbiasedRMSD [°C]
variability
bias [°C]
mean
outer circle: mean of data
unbiasedRMSD
[°C]
bias [°C]
unbiasedRMSD[psu]
bias [psu]
unbiasedRMSD[psu/m]
bias [psu/m]
unbiasedRMSD
[m]
bias [m]
(a) Bottom Temperature (b) Bottom
Salinity
(c) Stratificationat pycnocline (d) Depth of
pycnocline
Models do better at hindcasting bottom T & S than stratification
Stratification is a challenge for all the models:
• All underestimate strength and variability of stratification
• All underestimate variability of pycnocline depth.
Hydrodynamic Model Skill
Used 4 models to test sensitivity of hydrodynamic skill to:
o Vertical grid resolution (CBOFS2)o Freshwater inflow (CBOFS2; EFDC)o Vertical advection scheme (CBOFS2)o Horizontal grid resolution (UMCES-ROMS)o Coastal boundary condition (ChesROMS)o Mixing/turbulence closure (ChesROMS)o 2004 vs. 2005 (all models; in progress)
Sensitivities not yet tested: BathymetryVertical grid type: sigma vs. z-grid
Sensitivity Experiments
Outline
• What CB models and metrics are we using? – 5 Hydrodynamic models and 5 Biological (DO) models– RMSD; target diagrams
• What is the relative hydrodynamic skill of these CB models?
– Is this a function of resolution? Forcing?
• What is the relative DO skill of these CB models?
– Is this a function of model complexity?
• Summary and outlook for forecasting
- Simple models reproduce dissolved oxygen (DO) and hypoxic volume about as well as more complex models.
- All models reproduce DO better than they reproduce stratification.- A five-model average does better than any one model alone.
Dissolved Oxygen Model Comparison
Bottom DO
HypoxicVolume
Hypoxic Volume Time Series
2004
Models generally overestimate hypoxic volumes computed from station data – but what about uncertainties in these interpolated estimates?
Hypoxic Volume Time Series
2004 2004
Absolute model-data match3D modeled hypoxic volume
Interpolated hypoxic volumes contain large uncertainties (factor of two)
How can the simple 1-term models resolve seasonal cycle of HV without nutrient or temperature dependent net respiration?
Wind & Solubility!
Summary
• There currently exist multiple hydrodynamic and DO models for Chesapeake Bay
• Hydrodynamic skill is similar in all models• Simple constant net respiration rate models reproduce
seasonal DO cycle as well as complex models- But can they reproduce interannual variability? - The simpler models cannot be used to test impacts of decreasing
nutrient inputs to the Bay• Models reproduce DO better than stratification• Averaging output from multiple models provides better
hypoxia hindcasts than relying on any individual model alone
Outlook for DO forecasting
• Strong dependence on solubility (temperature) and winds is a good thing for forecasting, since these are variables we know relatively well! (At least better than respiration rates)
• Particularly easy to implement 1-term DO model into the CBOFS hydrodynamic model presently being run operationally at NCEP
• Caveat: To date we have tested these models only on seasonal time scales (i.e. not daily and interannual scales)
EXTRA SLIDES
Bottom DO – temporal variability
1-term DO model ICM (complex CBP model)
1-term DO model underestimates high DO and overestimates low DO: high not high enough, low not low enough
Total RMSD = 0.9 ± 0.1 Total RMSD = 0.9 ± 0.1
1-term DO model
Bottom DO – spatial variability
Overall model-data fit to CBP station bottom DO data is similar
1-term DO modelICM
(CBP model)
Total RMSD = 1.0 ± 0.1 Total RMSD = 1.1 ± 0.1
Depth of maximum stratification
-0.2-0.4-0.6-0.8
0.2
-0.2
-0.2
CBOFS2
CBOFS2 stratification is insensitive to: vertical grid resolution, vertical advection scheme and freshwater river input
Spatial variability of stratification for each month
ChesROMS
EFDC
UMCES-ROMS
CH3D
CBOFS2
month
CH3D/EFDC slightly better in terms of spatial variability
Temporal variability of depth of max strat. at 40 stations
ChesROMS
EFDC
UMCES-ROMS
CH3D
CBOFS2
Salinity [psu]
Model skill is similar in terms of temporal variability
Spatial variability of depth of max strat. for each month
ChesROMS
EFDC
UMCES-ROMS
CH3D
CBOFS2
month
CH3D slightly better in terms of spatial variability
Atm forcing; Horiz grid resolution
Max. stratification is not sensitive to horizontal grid resolution or changes in atmospheric forcing
CH3D, EFDC
ROMS
Stratification
Atm forcing; Horiz grid resolution
Models do better in 2005 than 2004!
2005
2004
Stratification
Atm forcing; Horiz grid resolution
Bottom salinity IS sensitive to horizontal grid resolution
High horiz res
Low horiz res
Bottom Salinity
(by M. Scully)
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Date in 2004
Hyp
oxic
Vol
ume
in k
m3
20
10
0
Base Case
(by M. Scully)
Effect of physical forcing on hypoxia
ChesROMS+1-term model
Seasonal changes in hypoxia are not a function of seasonal changes in freshwater.
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Date in 2004
Hyp
oxic
Vol
ume
in k
m3
20
10
0
Base Case
Freshwater river input constant
(by M. Scully)(by M. Scully)
Effect of physical forcing on hypoxia
ChesROMS+1-term model
Seasonal changes in hypoxia may be largely due to seasonal changes in wind.
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Date in 2004
Hyp
oxic
Vol
ume
in k
m3
20
10
0
Base CaseJuly wind year-round
(by M. Scully)(by M. Scully)
Effect of physical forcing on hypoxia
ChesROMS+1-term model
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Date in 2004
Hyp
oxic
Vol
ume
in k
m3
20
10
0
Base Case
January wind year-round
(by M. Scully)(by M. Scully)
Seasonal changes in hypoxia may be largely due to seasonal changes in wind.
Effect of physical forcing on hypoxia
ChesROMS+1-term model
EXTRA SLIDES
2004 simulation vs. 2004 data 2004 simulation vs. 2005 data
STRATIFICATION