An Adjoint Sensitivity An Adjoint Sensitivity Analysis of Analysis of the Southern California the Southern California Current Circulation and Current Circulation and Ecosystem Ecosystem Andy Moore, Andy Moore, Emanuele DiLorenzo, Hernan Emanuele DiLorenzo, Hernan Arango, Arango, Craig Lewis, Zack Powell, Craig Lewis, Zack Powell, Arthur Miller, Bruce Arthur Miller, Bruce Cornuelle Cornuelle
39
Embed
An Adjoint Sensitivity Analysis of the Southern California Current Circulation and Ecosystem
An Adjoint Sensitivity Analysis of the Southern California Current Circulation and Ecosystem. Andy Moore, Emanuele DiLorenzo, Hernan Arango, Craig Lewis, Zack Powell, Arthur Miller, Bruce Cornuelle. Outline. Motivation Model configuration and circulation Sensitivity and the adjoint - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
An Adjoint Sensitivity Analysis of An Adjoint Sensitivity Analysis of the Southern California Current the Southern California Current
Circulation and EcosystemCirculation and Ecosystem
Arthur Miller, Bruce CornuelleArthur Miller, Bruce Cornuelle
OutlineOutline
MotivationMotivation Model configuration and circulationModel configuration and circulation Sensitivity and the adjointSensitivity and the adjoint Indices of interestIndices of interest Examples of sensitivitiesExamples of sensitivities Seasonal variationsSeasonal variations SummarySummary
MotivationMotivation The California Current System is controlled The California Current System is controlled
by a number of different regimes (i.e. by a number of different regimes (i.e. upwelling, instability, topographic control).upwelling, instability, topographic control).
Sensitivity analysis can help to unravel this Sensitivity analysis can help to unravel this complex system.complex system.
Test hypotheses about other potentially Test hypotheses about other potentially important processes (i.e. stochastic forcing).important processes (i.e. stochastic forcing).
Sensitivity analysis is also an important Sensitivity analysis is also an important precursor for data assimilation and precursor for data assimilation and predictability analysis.predictability analysis.
The ROMS SCB DomainThe ROMS SCB Domain
7-20km resolution; forced by NCEP climatological winds and surface fluxes.ROMS has been used before in the CCS and validated by others (Marchesiello et al, 2003; Powell et al, 2005).
Outer domain:20km res, 20 levels.
Inner domain:7-20km res, 20 levels.Derives boundaryconditions from the outer domain.
A 4-Component Nitrogen-Based Trophic A 4-Component Nitrogen-Based Trophic ModelModel
Adjoint Approach to SensitivityAdjoint Approach to Sensitivity We must first define “sensitivity.”We must first define “sensitivity.” Consider the model state vector:Consider the model state vector: Consider a function or index, , defined in terms of space Consider a function or index, , defined in terms of space
and/or time integrals of .and/or time integrals of . Small changes in will lead to changes in Small changes in will lead to changes in
where:where:
We will define sensitivity as etcWe will define sensitivity as etc..
( , , , , , , , , )Tu v T S N P Z D
J
d J
J J J J JdJ du dv dT dS d
u v T S
J J J JdN dP dZ dD
N P Z D
, , ,J J J
u v T
dJ
Sensitivity AnalysisSensitivity Analysis
Consider a function Consider a function
ClearlyClearly
ButBut
So So
( )J G Φ
TJ G Φ Φ
( ) ( , ) ( )f i f it t t t Φ R Φ
( , )Tf iJ t t G Φ R Φ
Validity of the TL AssumptionValidity of the TL Assumption
TL assumption valid for ~30 days for perts TL assumption valid for ~30 days for perts that grow to an amplitude of:that grow to an amplitude of:
|SST|~0.5-1.0C|SST|~0.5-1.0C |v|~0.2 m/s|v|~0.2 m/s
These are lower bounds!These are lower bounds!
~ 0.08m
Seasonal Circulation Index RegionsSeasonal Circulation Index Regions
JSST
JKE
J90
IndicesIndices
21 CSST
A T
J SST SST dtdATA
0.15 For ,1
2 21 C CKE
A T Z
J u u v v dtdATAZ
For , 0.005
1BI f z N v
1BI BI
A T Z
J dtdATAZ
710
“Eady Index”
An index ofbaroclinic instability
1
IndicesIndices 2
90
1N C
X Z T
J N N dtdzdxTXZ
For ,
For ,
1 9.6
2
90
1P C
X Z T
J P P dtdzdxTXZ
0.3 1
What Physical Processes are likely to What Physical Processes are likely to Influence J?Influence J?
Advection
Q, P-E+R
Advection
InstabilityLong Rossby Waves
Short Rossby Waves
CoastallyTrappedWaves &
Tides
Turbulence/wave breaking
Note: All processes indicatedcan be significantly influencedby stochastic forcing.
A 4-Component Nitrogen-Based Trophic A 4-Component Nitrogen-Based Trophic ModelModel
The Signature of Advection in The Signature of Advection in
Day 5 Day 10 Day 15
Day 20 Day 25 Day 30
SSTJ Q
Seasonal Variations in Sensitivities ISeasonal Variations in Sensitivities IThe change in over the target arearequired to yield one change in
for .
SSTJ1
The change in Q over the target arearequired to yield one change in
for .
SSTJ
The change in v over the target arearequired to yield one change in
for .
SSTJ1
1
Low sensitivity
Low sensitivity
Low sensitivity
High sensitivity
High sensitivity
High sensitivity
0.035
0.01
32
20
2.3
0.3
Seasonal Variations in Sensitivities IISeasonal Variations in Sensitivities II
The change in over the target arearequired to yield one change in
for .
The change in over the target arearequired to yield one change in
for .
KEJ1
BIJ
1
Low sensitivity
Low sensitivity
High sensitivity
High sensitivity
0.0045
0.002
0.003
0.0003
Seasonal Variations in Sensitivities Seasonal Variations in Sensitivities IIIIII
The change in Q over the target arearequired to yield one change in
for .
The change in v over the target arearequired to yield one change in
for .
BI
J
BIJ
1
1
Low sensitivity
Low sensitivity
High sensitivity
High sensitivity
220
15
1
< 0.01
Interdependencies: Sensitivity of KE Interdependencies: Sensitivity of KE to Baroclinic Instabilityto Baroclinic Instability
Change in required to yield a one change in when varying only v for . Recall that
BI KEJ 1 710BI
Low sensitivity
High sensitivity
Log
scal
e
910
1310
Summary for Physical CirculationSummary for Physical Circulation
SST anomaly in coastal upwelling regions SST anomaly in coastal upwelling regions equally sensitive to variations in and Q, equally sensitive to variations in and Q, with v a close second.with v a close second.
Highest (Lowest) sensitivity in Fall (Spring)Highest (Lowest) sensitivity in Fall (Spring) KE anomaly most sensitive to variations in KE anomaly most sensitive to variations in
Adjoint Sensitivity for Ecosystem ModelAdjoint Sensitivity for Ecosystem Model
Oct: on day 13 3 day 30NO NO Oct: on day 13 day 30
NO P
3 day 30NO Z Oct: on day 1 Oct: on day 13 day 30
NO D
Mar: on day 13 day 30NO P
Jul: on day 13 day 30
NO P
Seasonal Variations in Sensitivities ISeasonal Variations in Sensitivities I
Change in required to yield a one change in for . Note the log-scale!
90PJ 1
Low sensitivity
High sensitivity
Log
scal
e
110
510
Seasonal Variations in Sensitivity IISeasonal Variations in Sensitivity IIThe change in N over the target area required to yield one change in
for .
The change in P over the target area required to yield one change in
for .
The change in Z over the target area required to yield one change in
for .
90PJ
90PJ
90PJ
1
1
1
Low sensitivity
Low sensitivity
Low sensitivity
High sensitivity
High sensitivity
High sensitivity
Summary of Biological CirculationSummary of Biological Circulation
For all NPZD-based indices, variations in are For all NPZD-based indices, variations in are found to be important.found to be important.
Variations in NPZD equally important (internal Variations in NPZD equally important (internal interactions important).interactions important).
NPZD concentrations strongly influenced by the NPZD concentrations strongly influenced by the physical environment.physical environment.
Highest (Lowest) sensitivities in Spring/Summer Highest (Lowest) sensitivities in Spring/Summer (Fall/Winter).(Fall/Winter).
Extraordinary sensitivities during some Spring Extraordinary sensitivities during some Spring periods suggestive of linear instability (i.e. we are periods suggestive of linear instability (i.e. we are perhaps the TL assumption a little too far!).perhaps the TL assumption a little too far!).
Other Ongoing ApplicationsOther Ongoing Applications
Intra-Americas SeaIntra-Americas Sea Monterey BayMonterey Bay