Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
JGR-Oceans, Published 3 October 2013
Aaron Bever, Marjy Friedrichs, Carl Friedrichs, Malcolm Scully, Lyon Lanerolle
OUTLINE / SUMMARY
1.
Motivation regarding Chesapeake Bay hypoxia and hypoxic volume.
2.
Relation to US-IOOS Modeling Testbed program and general methods.
3.
Use 3D models to examine uncertainties in interpolating hypoxic volume.
•
•
Observed DO are not a “snapshot” (observed over ~ 2 weeks)
Observed DO have coarse spatial resolution
4.
Use 3D models to improve EPA-CBP interpolations of hypoxic volume.
5.
Use improved interpolations to assess different metrics for estimating
interannual variability in hypoxic volume.
Bever, A.J., M.A.M. Friedrichs, C.T. Friedrichs, M.E. Scully, and L.W. Lanerolle, 2013.
Combining observations and numerical model results to improve estimates of hypoxic
volume within the Chesapeake Bay, USA. Journal of Geophysical Research,
doi:10.1002/jgrc.20331
Definition of hypoxia and motivation for study:
(UMCES, Coastal Trends)
HYPOXIA
DO ≤ 2.0 mg/L
Hypoxic volume =
Total water volume with
dissolved oxygen (DO) ≤ 2.0 mg/L
Summer Hypoxic
Volume (km3)
Motivation (cont.):
Demersal Fish Trawl
(VIMS, ChesMMAP)
(UMCES, Coastal Trends)
Summer Hypoxic
Volume (km3)
Motivation (cont.):
Caution: Hypoxic volume estimates are
made by interpolating between widely
spaced profiles collected only every 2 to
4 weeks. What are the uncertainties?
Demersal Fish Trawl
(VIMS, ChesMMAP)
(UMCES, Coastal Trends)
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
JGR-Oceans, Published 3 October 2013
Aaron Bever, Marjy Friedrichs, Carl Friedrichs, Malcolm Scully, Lyon Lanerolle
OUTLINE
1.
Motivation regarding Chesapeake Bay hypoxia and hypoxic volume.
2.
Relation to US-IOOS Modeling Testbed program and general methods.
3.
Use 3D models to examine uncertainties in interpolating hypoxic volume.
•
•
Observed DO are not a “snapshot” (observed over ~ 2 weeks)
Observed DO have coarse spatial resolution
4.
Use 3D models to improve EPA-CBP interpolations of hypoxic volume.
5.
Use improved interpolations to assess different metrics for estimating
interannual variability in hypoxic volume.
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
Relationship to US-IOOS Modeling Testbed:
Part of Coastal & Ocean Modeling Testbed (COMT) Project headed by
Rick Luettich (IMS), funded by NOAA US-IOOS Office
COMT Mission: Accelerate the transition of scientific and technical
advances from the modeling research community to improve federal
agencies’ operational ocean products and services
Initial Phase: Estuarine Hypoxia, Shelf Hypoxia and Coastal
Inundation Modeling Testbeds; Cyber-infrastructure to advance
interoperability and archiving
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
Relationship to COMT Estuarine Hypoxia Modeling Testbed:
COMT Estuarine Hypoxia Team (Initial Phase)
Marjy Friedrichs (VIMS)
Carl Friedrichs (VIMS)
Aaron Bever (VIMS)
Jian Shen (VIMS)
Malcolm Scully (ODU)
Raleigh Hood/Wen Long (UMCES)
Ming Li (UMCES)
Kevin Sellner (CRC)
Federal partners
Carl Cerco (USACE)
David Green (NOAA-NWS)
Lyon Lanerolle (NOAA-CSDL)
Lewis Linker (EPA)
Doug Wilson (NOAA-NCBO)
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
General COMT Estuarine Hypoxia modeling methods:
• Compare relative skill and strengths/weaknesses of
various Chesapeake Bay models
• Assess how model differences affect water quality
simulations
• Recommend improvements to agency operational
products associated with managing hypoxia
Five hydrodynamic models configured for the Bay
Five hydrodynamic models configured for the Bay
TODAY’S TALK
Five dissolved oxygen (DO) models configured for the Bay
o ICM: EPA-CBP model; complex biology
o BGC: NPZD-type biogeochemical model
o 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
(not a function of x, y, temperature,
nutrients OR depth…)
Five dissolved oxygen (DO) models configured for the Bay
o ICM: EPA-CBP model; complex biology
o BGC: NPZD-type biogeochemical model
o 1eqn: Simple one equation respiration
(includes SOD)
TODAY’S TALK
o 1term-DD: depth-dependent respiration
(not a function of x, y, temperature,
nutrients…)
o 1term: Constant net respiration
(not a function of x, y, temperature,
nutrients OR depth…)
Coupled hydrodynamic-DO models
Today’s talk = Four combinations:
o
o
o
o
CH3D
CBOFS
ChesROMS
ChesROMS
+
+
+
+
ICM  EPA-CBP model
1term
1term
1term+DD
-- Physical models are similar, but grid resolution differs
-- Biological/DO models differ dramatically
-- All models run for 2004 and 2005 and compared to EPA Chesapeake
Bay Program DO observations
Relative model skill
How well do the models
represent the mean and
variability of
dissolved oxygen at
~40 CBP stations
in 2004 and 2005?
= ~40 EPA-CBP stations used in
this model-data comparison
Relative model skill: animation
Bottom DO [mg/L]
Constant Respiration
Observed
Modeled
Complex Biology
Relative model skill: Target diagrams
Dimensionless version of
plot normalizes by standard
deviation of observations
Model skill: Bottom DO
-- The models all have significant skill (normalized RSMD < 1) in reproducing
observed bottom dissolved oxygen (DO).
-- The four models all reproduce observations of bottom DO about equally well.
-- Unlike observations, model output is continuous in space and time.
-- So use the continuous model output to estimate uncertainties caused by CBP
interpolations of discontinous observed data.
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
JGR-Oceans, Published 3 October 2013
Aaron Bever, Marjy Friedrichs, Carl Friedrichs, Malcolm Scully, Lyon Lanerolle
OUTLINE
1.
Motivation regarding Chesapeake Bay hypoxia and hypoxic volume.
2.
Relation to US-IOOS Modeling Testbed program and general methods.
3.
Use 3D models to examine uncertainties in interpolating hypoxic volume.
•
•
Observed DO are not a “snapshot” (observed over ~ 2 weeks)
Observed DO have coarse spatial resolution
4.
Use 3D models to improve EPA-CBP interpolations of hypoxic volume.
5.
Use improved interpolations to assess different metrics for estimating
interannual variability in hypoxic volume.
Observation-derived Hypoxic Volume estimates
Observations:
 Of 99 CBP stations (red dots),
30-65 are sampled each
“cruise”
Note: Cruises use 2 boats from 2
institutions to collect vertical profiles;
last for up to 2 weeks
Interpolation Method:
 CBP Interpolator Tool
 HV = DO < 2 mg/L
 Full Bay
Uncertainties arise from:
 Temporal errors: observations
are not a snapshot
 Spatial errors: discrete data
cannot resolve entire Bay
(“CBP” = EPA Chesapeake Bay Program)
Model-derived Hypoxic Volume estimates
Integrated 3D:
 Hypoxic volume is computed
from integrating over all grid
cells
Interpolated Absolute Match:
 Same 30-65 stations are
“sampled” at same time/place
as observations are available
Interpolated Spatial Match:
 Same stations are “sampled”,
but samples are taken
synoptically (i.e., all at once)
Interpolation Method:
 CBP Interpolator Tool
 HV = DO < 2 mg/L
 Full Bay
(“CBP” = EPA Chesapeake Bay Program)
Model Skill Assessment for Hypoxic Volume (HV)
Modeled
Absolute Match
vs.
Observation-derived
Interpolation
unbiased
RMSD
Modeled
Integrated 3D
vs.
Observation-derived
Interpolation
unbiased
RMSD
• Skill of Absolute Match vs. Observation-derived HV (both interpolated) is higher.
• Absolute Match skill compared to Integrated 3D  uncertainties in data-derived HV.
Hypoxic Volume Estimates
• When observations
and model are
interpolated in
same way, good
match
20
CH3D-ICM
= Absolute Match
Hypoxic Volume, km
3
10
0
20
ChesROMS+1term
10
0
20
Observations-derived
10
0
05/01
06/01
07/01
08/01
09/01
Date in 2004, Month/Day
10/01
11/01
Integrated 3D HV
• But interpolated
HV underestimates
actual HV for every
cruise
06/01
07/01
08/01
09/01
Date in 2004, Month/Day
10/01
11/01
Integrated 3D HV
Absolute Match
20 CH3D-ICM
CH3D-ICM
= Absolute Match
10
3
0
• When observations
05/01
and model are
interpolated in
same way, good
match
Hypoxic Volume Estimates
Hypoxic Volume, km
10
Cruise Date Range
0
20
ChesROMS+1term
ChesROMS+1term
10
0
20
Data-derived
Data-derived
10
0
05/01
06/01
09/01
08/01
07/01
Date in 2004, Month/Day
10/01
11/01
Integrated 3D HV
Absolute Match
06/01
• When observations
and model are
interpolated in
same way, good
match
• But interpolated
HV underestimates
actual HV for every
cruise
• Much of this
disparity could be
due to temporal
errors (red bars)
07/01
08/01
09/01
Date in 2004, Month/Day
20
10/01
11/01
Integrated 3D HV
Absolute Match
Spatial Match
Spatial Match Range
Cruise Date Range
CH3D-ICM
10
3
05/01
Hypoxic Volume Estimates
Hypoxic Volume, km
0
0
20
ChesROMS+1term
10
0
20
=
Data-derived
10
0
05/01
06/01
07/01
08/01
09/01
Date in 2004, Month/Day
10/01
11/01
Integrated 3D HV
Absolute Match
Spatial Match
Spatial Match Range
Date in 2004, Month/Day
• When data and
model are
interpolated in
same way, good
match
• But interpolated
HV underestimates
actual HV for every
cruise
• Much of this
disparity could be
due to temporal
errors (red bars)
• Same pattern
across all 4 models
for both 2004 &
2005
Integrated 3D HV
Absolute Match
Spatial Match
Spatial Match Range
Cruise Date Range
Uncertainties in data-derived hypoxic volumes
Spatial errors show
interpolated HV is almost
always too low (up to 5 km3)
Spatial error = Integrated 3D HV
minus Spatial Match
Uncertainties in data-derived hypoxic volumes
Spatial errors show
interpolated HV is almost
always too low (up to 5 km3)
The temporal errors from
non-synoptic sampling can
be as large as spatial errors
(~5 km3)
Spatial error = Integrated 3D HV
minus Spatial Match
Range of Spatial match over a cruise
gives size of potential temporal error
Spatial errors show
interpolated HV is almost
always too low (up to 5 km3)
The temporal errors from
non-synoptic sampling can
be as large as spatial errors
(~5 km3)
Similar patterns across all
4 models for both 2004 &
2005
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
JGR-Oceans, Published 3 October 2013
Aaron Bever, Marjy Friedrichs, Carl Friedrichs, Malcolm Scully, Lyon Lanerolle
OUTLINE
1.
Motivation regarding Chesapeake Bay hypoxia and hypoxic volume.
2.
Relation to US-IOOS Modeling Testbed program and general methods.
3.
Use 3D models to examine uncertainties in interpolating hypoxic volume.
•
•
Observed DO are not a “snapshot” (observed over ~ 2 weeks)
Observed DO have coarse spatial resolution
4.
Use 3D models to improve EPA-CBP interpolations of hypoxic volume.
5.
Use improved interpolations to assess different metrics for estimating
interannual variability in hypoxic volume.
Improving observation-derived hypoxic volumes
Blue triangles = 13 selected CBP stations
 Reduce Temporal errors:
1. Choose subset of 13 CBP
stations
2. Routinely sampled within
2.3 days of each other
3. Characterized by high DO
variability
Improving observation-derived hypoxic volumes
Blue triangles = 13 selected CBP stations
 Reduce Temporal errors:
1. Choose subset of 13 CBP
stations
2. Routinely sampled within
2.3 days of each other
3. Characterized by high DO
variability
But why 13 stations?
Improving observation-derived hypoxic volumes
Modeled
Integrated 3D
vs.
Spatial Match for
Different Station Sets
Improving observation-derived hypoxic volumes
 Reduce Spatial errors:
1. For each model and
each cruise, derive a
correction factor as a
function of interpolated
HV that “corrects” this
13-station Spatial Match
HV to equal the
Integrated 3D HV.
Improving observation-derived hypoxic volumes
 Reduce Spatial errors:
Before
Scaling
1. For each model and
each cruise, derive a
correction factor as a
function of interpolated
HV that “corrects” this
13-station Spatial Match
HV to equal the
Integrated 3D HV.
2. Apply correction factor
to HV time-series
3. Scaling-corrected
“interpolated” HV more
accurately represents
true HV
After
Scaling
Interannual (1984-2012) corrected (i.e., scaled) time
series of observed Hypoxic Volume
Combining Observations & Numerical Model Results to Improve
Estimates of Hypoxic Volume within the Chesapeake Bay
JGR-Oceans, Published 3 October 2013
Aaron Bever, Marjy Friedrichs, Carl Friedrichs, Malcolm Scully, Lyon Lanerolle
OUTLINE
1.
Motivation regarding Chesapeake Bay hypoxia and hypoxic volume.
2.
Relation to US-IOOS Modeling Testbed program and general methods.
3.
Use 3D models to examine uncertainties in interpolating hypoxic volume.
•
•
Observed DO are not a “snapshot” (observed over ~ 2 weeks)
Observed DO have coarse spatial resolution
4.
Use 3D models to improve EPA-CBP interpolations of hypoxic volume.
5.
Use improved interpolations to assess different metrics for estimating
interannual variability in hypoxic volume.
Interannual DO Assessment
 How do we determine which years are good/bad?
Or whether we’re seeing a recent reduction in hypoxia?
• Length of time waters are hypoxic
• Percent of Bay (volume) that is hypoxic
 Choose metrics dependent on ecological function of interest:
•
Prolonged low HV could be worse for some species than an
extensive short duration hypoxic event, and vice versa.
Different HV metrics can give different results
for which years are “worst”
Interannual DO Assessment
1995 - 1997
Of these three years, 1996 appears to have the least hypoxia
Interannual DO Assessment
1995 - 1997
Average Summer HV
Annual HV Time-Series
cruises = late June, both July
both Aug, early Sept
In 1996 Maximum HV is relatively low BUT Average Summer HV is relatively high;
Maximum Annual HV is probably not the best DO metric
Red dashed lines
denote period of
“summer averaging”
2011 looks “good”,
because much
hypoxia occurs
outside of
“summer” time
period
Cumulative HV
Average Summer HV
vs.
Cumulative HV
• Performance of relative years
changes
Average Summer HV
vs.
Cumulative HV
• Performance of relative years
changes
• Average Summer HV doesn’t
taken into account long HV
duration
• If climate change affects time
of onset, this will not be seen
when using Avg Summer HV
Summary/Conclusions
 Information from multiple models (2004-2005) has been used
to assess uncertainties in present CBP interpolated hypoxic
volume estimates
• Temporal uncertainties: up to ~5 km3
• Spatial uncertainties: up to ~5 km3
 These are significant, given maximum HV is ~10-15 km3
 A method for correcting interpolate HV time series for
temporal and spatial errors has been presented, based on
the 3D structure of multiple model DO results
• 13 observed stations do as well for HV as 40-60 or more
 Different HV metrics can give different results in terms of
assessing DO improvement
• Cumulative HV is a good way to take into account shifts in
onset of hypoxia that could occur with climate change
Extra Slides
Average Summer HV
vs.
Cumulative HV
• Performance of relative years
changes
• Average Summer HV doesn’t
take into account long HV
duration
• If climate change affects time
of onset, this will not be seen
when using Avg Summer HV
As in previous slide, without HV correction
This demonstrates that the correction of HVs does not significantly
affect the Average Summer HV vs. Cumulative HV conclusions
Cumulative HV
CBP13 scaled is
now much more
inline with the
model estimates
of 3D HV.