SlideShare a Scribd company logo
1 of 123
Download to read offline
Using explainable AI to identify
key regions of climate change in
GFDL SPEAR large ensembles
https://zacklabe.com/ @ZLabe
Zachary Labe
Postdoc in Seasonal-to-Decadal (S2D) Variability and Predictability Division
with Nathaniel Johnson and Thomas Delworth
15 March 2023
GFDL Lunchtime Seminar
1
1) Where do we
go from here?
2
1) Where do we
go from here?
2) How do we
disentangle
internal climate
variability?
Feb/Mar 2016
3
3) How do we
account for
regional
patterns of
change?
Explainable machine learning can
distinguish between regional patterns
of time-evolving climate change
in GFDL models.
4
SIGNIFICANCE
Machine Learning
is not new!
“A Bayesian Neural Network for
Severe-Hail Prediction (2000)”
“Classification of Convective Areas
Using Decision Trees (2009)”
“A Neural Network for Damaging
Wind Prediction (1998)”
“Generative Additive Models versus
Linear Regression in Generating
Probabilistic MOS Forecasts of
Aviation Weather Parameters (1995)”
”A Neural Network for
Tornado Prediction
Based on Doppler
Radar-Derived
Attributes (1996)”
”The Diagnosis of
Upper-Level Humidity
(1968)”
5
“An adaptive data processing system for weather forecasting”
It’s a neural network!
[Hu and Root (1964), APME]
6
Artificial Intelligence
Machine Learning
Deep Learning
Computer/Data Science
7
Computer/Data Science
Supervised
Learning
Unsupervised
Learning
Labeled data
Classification
Regression
Unlabeled data
Clustering
Dimension reduction
8
Artificial Intelligence
Machine Learning
Deep Learning
DATA-HUNGRY!
Do it better
e.g., parameterizations in climate models are not
perfect, use ML to make them more accurate
Do it faster
e.g., code in climate models is very slow (but we
know the right answer) - use ML methods to speed
things up
Do something new
• e.g., go looking for non-linear relationships you
didn’t know were there
WHY ELSE SHOULD WE CONSIDER
MACHINE LEARNING?
9
Do it better
e.g., parameterizations in climate models are not
perfect, use ML to make them more accurate
Do it faster
e.g., code in climate models is very slow (but we
know the right answer) - use ML methods to speed
things up
Do something new
• e.g., go looking for non-linear relationships you
didn’t know were there
Very relevant for
research: may be
slower and worse,
but can still learn
something
10
WHY ELSE SHOULD WE CONSIDER
MACHINE LEARNING?
Machine learning for meteorology
IDENTIFYING SEVERE THUNDERSTORMS
Molina et al. 2021
Martin et al. 2022
CLASSIFYING PHASE OF MADDEN-JULLIAN OSCILLATION
DETECTING CONVECTION FROM SATELLITES
Lee et al. 2021
LOCATING COLD FRONTS
Dagon et al. 2022
11
Machine learning for oceanography
CLASSIFYING ARCTIC OCEAN ACIDIFICATION
Krasting et al. 2022
LARGE-SCALE OCEAN CIRCULATION
Clare et al. 2022
ESTIMATING OCEAN SURFACE CURRENTS
Sinha and Abernathey, 2021
12
Machine learning for climate
PHYSICAL DRIVERS OF ENSO DYNAMICS
Shin et al. 2022
IDENTIFYING DECADAL STATE DEPENDENCE
Gordon and Barnes, 2022
INTERNAL/EXTERNAL CLIMATE FORCING
Po-Chedley et al. 2022
13
INPUT
[DATA]
PREDICTION
Machine
Learning
14
INPUT
[DATA]
PREDICTION
~Statistical
Algorithm~
15
INPUT
[DATA]
PREDICTION
Machine
Learning
16
Opening the black box
Artificial Intelligence
Machine Learning
Deep Learning
Artificial Neural Networks
17
Computer/Data Science
X1
X2
INPUTS
Artificial Neural Networks [ANN]
18
Linear regression!
Artificial Neural Networks [ANN]
X1
X2
W1
W2
∑ = X1W1+ X2W2 + b
INPUTS
NODE
19
Artificial Neural Networks [ANN]
X1
X2
W1
W2
∑
INPUTS
NODE
Linear regression with non-linear
mapping by an “activation function”
Training of the network is merely
determining the weights “w” and
bias/offset “b"
= factivation(X1W1+ X2W2 + b)
20
Artificial Neural Networks [ANN]
X1
X2
W1
W2
∑
INPUTS
NODE
= factivation(X1W1+ X2W2 + b)
ReLU Sigmoid Linear
21
X1
X2
∑
inputs
HIDDEN LAYERS
X3
∑
∑
∑
OUTPUT
= predictions
Artificial Neural Networks [ANN]
: : ::
INPUTS
22
Complexity and nonlinearities of the ANN allow it to learn
many different pathways of predictable behavior
Once trained, you have an array of weights and biases
which can be used for prediction on new data
INPUT
[DATA]
PREDICTION
Artificial Neural Networks [ANN]
23
What is the annual mean temperature of Earth?
24
THE REAL WORLD
(Observations)
What is the annual mean temperature of Earth?
Data from
Berkeley Earth Surface Temperature
1930 2022
25
What is the annual mean temperature of Earth?
THE REAL WORLD
(Observations)
Let’s run a
climate model
One ensemble member
2022
1930 2050
26
Data
from
SPEAR_M
ED
What is the annual mean temperature of Earth?
THE REAL WORLD
(Observations)
Let’s run a
climate model
again!
Two ensemble members
27
Data
from
SPEAR_M
ED
What is the annual mean temperature of Earth?
THE REAL WORLD
(Observations)
Let’s run a
climate model
again & again!
Three ensemble members
28
Data
from
SPEAR_M
ED
What is the annual mean temperature of Earth?
THE REAL WORLD
(Observations)
CLIMATE MODEL
LARGE ENSEMBLE
30 ensemble
members in
GFDL SPEAR
29
What is the annual mean temperature of Earth?
Mean of ensembles
= forced response (climate change)
THE REAL WORLD
(Observations)
CLIMATE MODEL
LARGE ENSEMBLE
30
ensemble members
In GFDL SPEAR
30
What is the annual mean temperature of Earth?
Mean of ensembles
= forced response (climate change)
THE REAL WORLD
(Observations)
CLIMATE MODEL
LARGE ENSEMBLE
30
ensemble members
In GFDL SPEAR
Range of ensembles
= internal variability (noise)
Mean of ensembles
= forced response (climate change)
31
But let’s remove
climate change…
Climate Change Signal
(ensemble mean)
Observations
Ensemble
Members
32
Ensemble
Members
Mean of
anomalies
After removing the
forced response…
= anomalies/noise!
33
Ensemble
members in
GFDL SPEAR
Maps of a given time period for each ensemble
Inputs for machine learning
34
Ensemble
members in
GFDL SPEAR
35
Training Data:
24 ensemble members
Maps of a given time period for each ensemble
36
Training Data:
24 ensemble members
Validation Data:
4 ensemble members
37
Training Data:
24 ensemble members
Validation Data:
4 ensemble members
Testing Data:
2 ensemble members
Historical Forcing – GFDL SPEAR Future Scenarios – GFDL SPEAR
38
Can a neural network
learn unique patterns of
climate change related
to each future emission
scenario?
1930 2010 2020 2100
39
Train a neural
network to predict
5 classes
(climate scenarios)
Yearly Maps of T2M
Yearly Maps of T2M
Neural
Network
Classify
Climate
Scenario
Artificial
Neural
Network
Output
=
5
Classes
Yearly Maps of T2M
Neural
Network
Binary Output Binary Output
40
Step #1
Read in gridded maps of a
climate variable from
SPEAR simulations
Yearly Maps of T2M
Yearly Maps of T2M
Neural
Network
Classify
Climate
Scenario
Artificial
Neural
Network
Output
=
5
Classes
Yearly Maps of T2M
Neural
Network
Binary Output Binary Output
41
Step #2
Feed data into an
artificial neural network
with three hidden layers
Yearly Maps of T2M
Yearly Maps of T2M
Neural
Network
Classify
Climate
Scenario
Artificial
Neural
Network
Output
=
5
Classes
Yearly Maps of T2M
Neural
Network
Binary Output Binary Output
42
Step #3
Classify which climate
scenario (n=5) is
associated with each map
Yearly Maps of T2M
Yearly Maps of T2M
Neural
Network
Classify
Climate
Scenario
Artificial
Neural
Network
Output
=
5
Classes
Yearly Maps of T2M
Neural
Network
Binary Output Binary Output
43
Step #4
Why? à XAI
WHY
LAYER-WISE RELEVANCE PROPAGATION (LRP)
Volcano
Timber
Wolf
Image Classification LRP
LRP heatmaps show regions
of “relevance” that
contribute to the neural
network’s decision-making
process for a sample
belonging to a particular
output category
Neural Network
WHY
Backpropagation – LRP
https://heatmapping.org/ 44
WHY
LAYER-WISE RELEVANCE PROPAGATION (LRP)
Volcano
Timber
Wolf
Image Classification LRP
LRP heatmaps show regions
of “relevance” that
contribute to the neural
network’s decision-making
process for a sample
belonging to a particular
output category
Neural Network
WHY
Backpropagation – LRP
https://heatmapping.org/ 45
Image Classification LRP
https://heatmapping.org/
NOT PERFECT
Crock
Pot
Neural Network
Backpropagation – LRP
WHY
46
LAYER-WISE RELEVANCE PROPAGATION (LRP)
EXPLAINABLE AI (XAI)
THERE ARE MANY
METHODS
A bird!
XAI
47
[Adapted from Adebayo et al., 2020]
THERE ARE MANY
METHODS
EXPLAINABLE AI (XAI)
48
[Adapted from Adebayo et al., 2020]
Visualizing something we already know…
ENSO
49
Neural
Network
[0] La Niña [1] El Niño
Input a map of sea surface temperatures
50
[Toms et al. 2020, JAMES]
Visualizing something we already know…
Input maps of sea surface
temperatures (SST) to
identify El Niño or La Niña
Use ‘LRP’ to see how the
neural network is making
its decision
[Toms et al. 2020, JAMES]
Layer-wise Relevance Propagation
Composite SST Observations
LRP [Relevance]
SST Anomaly [°C]
0.00 0.75
0.0 1.5
-1.5
51
Warmer
Colder
High
Low
52
Returning to our application…
53
Predictions for
SPEAR_MED
Testing Data
Accuracy=92%
Nearer to predicted class
Further from predicted class
54
Predictions for
SPEAR_MED
Testing Data
Accuracy=92%
Nearer to predicted class
Further from predicted class
Global Mean
Surface Temperature
Can we identify changes in
future climate impacts after
rapid mitigation?
55
56
30 ensembles for GFDL
SPEAR_MED
9 ensembles for GFDL
SPEAR_MED
Input maps from
out-of-sample
ensembles into
classification
network
2020 2030 2100
57
2031 2040
Rapid Mitigation
Rapid Mitigation
30 ensembles for GFDL
SPEAR_MED
9 ensembles for GFDL
SPEAR_MED
58
Predictions for a single
ensemble member from
SSP5-3.4OS
SSP5-8.5
SSP2-4.5
2015 2055 2065 2095
59
Predictions for a single
ensemble member from
SSP5-3.4OS
SSP5-8.5
SSP2-4.5
2055-2060
rapid mitigation begins
2015 2095
60
Are these
predictions robust
across ensemble
members? (n=30)
SSP5-3.4OS
Transition from
SSP5-8.5 to SSP2-4.5
2015 2060 2100
61
What if we start
mitigation
10 years earlier?
Transition from
SSP5-8.5 to SSP2-4.5
Transition from
SSP5-8.5 to SSP2-4.5
Transition from
SSP2-4.5 to SSP1-1.9
2015 2060 2100
2015 2060 2100
62
What climate
patterns are
associated with
these transitions?
Transition from
SSP5-8.5 to SSP2-4.5
Transition from
SSP5-8.5 to SSP2-4.5
Transition from
SSP2-4.5 to SSP1-1.9
Rapid mitigation
Rapid mitigation
63
Difficult to distinguish
the patterns
associated with
scenario transitions
Composites of
relevance maps for
the mitigation
predictions
SSP5-3.4OS example for 2015 to 2100
Nearer to predicted scenario
Further from predicted scenario
Yearly Maps of T2M
Yearly Maps of T2M
Neural
Network
Classify
Climate
Scenario
Artificial
Neural
Network
Output
=
5
Classes
Yearly Maps of T2M
Neural
Network
Binary Output Binary Output
64
Steps #4-5
65
XAI composites of years associated with the transition from SSP5-8.5 to SSP2-4.5
(a) approx. 2055-2060 (b) approx. 2040-2045
66
North Atlantic is an
important indictor
region for climate
signals related to
identifying from
SSP5-8.5 to SSP2-4.5
Future Climate Change Rapid Mitigation
67
Framework can be
applied to different
geographic regions
and climate variables
Parallel approach for
detecting climate
intervention scenarios
Labe, Z.M., E.A. Barnes, and J.W. Hurrell (2023). Identifying the
regional emergence of climate patterns in the ARISE-SAI-1.5
simulations. EarthArXiv, DOI: 10.31223/X5394Z
1) Explainable AI methods identify
regions of climate change that
indicate the emission scenario
2) Framework for outlining benefits
from mitigation by testing out of
sample ensembles from SSP5-3.4OS
3) Patterns of fast and slow responses
to rapid climate mitigation are
revealed by explainable AI
KEY FINDINGS
• Indicator
patterns of
climate
change
• Quantify
benefits of
rapid climate
mitigation
• Regions of
rapid or slow
response to
mitigation
Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), A data-driven approach to
identifying key regions of change associated with rapid climate mitigation, in prep
1) Explainable AI methods identify
regions of climate change that
indicate the emission scenario
2) Framework for outlining benefits
from mitigation by testing out of
sample ensembles from SSP5-3.4OS
3) Patterns of fast and slow responses
to rapid climate mitigation are
revealed by explainable AI
KEY FINDINGS
• Indicator
patterns of
climate
change
• Quantify
benefits of
rapid climate
mitigation
• Regions of
rapid or slow
response to
mitigation
Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), A data-driven approach to
identifying key regions of change associated with rapid climate mitigation, in prep
1) Explainable AI methods identify
regions of climate change that
indicate the emission scenario
2) Framework for outlining benefits
from mitigation by testing out of
sample ensembles from SSP5-3.4OS
3) Patterns of fast and slow responses
to rapid climate mitigation are
revealed by explainable AI
KEY FINDINGS
• Indicator
patterns of
climate
change
• Quantify
benefits of
rapid climate
mitigation
• Regions of
rapid or slow
response to
mitigation
Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), A data-driven approach to
identifying key regions of change associated with rapid climate mitigation, in prep
71
NASA/GISS GISTEMPv4
72
73
A "warming hole”
[Eischeid et al. 2023, revised]
74
Temperature anomalies [ °C ] relative to 1981-2010
Observations from NClimGrid
Climate model data from GFDL SPEAR_MED
United States – Summer
75
1920 2020
TEMPERATURE
76
TEMPERATURE
We know some metadata…
+ What year is it?
+ Where did it come from?
77
TEMPERATURE
We know some metadata…
+ What year is it?
+ Where did it come from?
78
We know some metadata…
+ What year is it?
+ Where did it come from?
[Labe and Barnes, 2022; ESS]
TEMPERATURE
79
TEMPERATURE
Neural network learns nonlinear
combinations of forced climate
patterns to identify the year
We know some metadata…
+ What year is it?
+ Where did it come from?
[Labe and Barnes, 2022; ESS]
80
----ANN----
2 Hidden Layers
10 Nodes each
Ridge Regularization
Early Stopping
[e.g., Barnes et al. 2019, 2020]
[e.g., Labe and Barnes, 2021]
TIMING OF EMERGENCE
(COMBINED VARIABLES)
RESPONSES TO
EXTERNAL CLIMATE
FORCINGS
PATTERNS OF
CLIMATE INDICATORS
[e.g., Rader et al. 2022]
Surface Temperature Map Precipitation Map
+
TEMPERATURE
We know some metadata…
+ What year is it?
+ Where did it come from?
[Labe and Barnes, 2022; ESS]
81
Seasonal Maps of T2M, TMAX, TMIN
Input
82
Seasonal Maps of T2M, TMAX, TMIN
Hidden Layers
Artificial Neural Network
Input
83
Seasonal Maps of T2M, TMAX, TMIN
Hidden Layers
Output
1921
2100
Artificial Neural Network
Input
84
Backpropagation
Seasonal Maps of T2M, TMAX, TMIN
Hidden Layers
Output
1921
2100
Artificial Neural Network
Input
85
Post hoc – XAI methods
Temperature
anomalies
[
°C
]
relative
to
1981-2010
TMIN
TMAX
1921 to 2022
1921 to 2022
86
Summer (June-August)
Neural Network
Predictions
for SPEAR/Obs
87
88
1921 2021 2100
ACTUAL YEARS
PREDICTED
YEARS
NOAA Monthly U.S.
NClimGrid v1.0
1921
2021
2100
Max year predicted in the 1921-1950
baseline for observations
Timing of
Emergence
89
1921-1950
Skill
1:1 Perfect Prediction
Maximum Temperature
90
Minimum Temperature
June – August – Timing of Emergence (ToE) for observations
91
June – August – skill for observations
How is the neural network able to detect the year prior to ~1990?
Temperature anomalies [ °C ] relative to 1981-2010
Machine learning predictions GFDL SPEAR_MED simulation
92
Machine Learning Explainability Methods – Attribution 93
Decrease
likelihood of year
Increase
likelihood of year
Western USA Central USA Eastern USA
94
Western USA Central USA Eastern USA
95
Western USA Central USA Eastern USA
96
Fully Coupled
ToE Statistical
Methods
97
First year that the 10-year running-mean
temperature exceeds and stays above the
mean 1921–1950 reference temperature by
more than two standard deviations
Most Areas = 2000s
98
Natural Forcings
99
Shuffling Map
Shuffling Time
100
1) Is it
aerosols?
101
Only available from
1921 to 2020
Proof of
Concept
102
50 km resolution 100 km resolution
103
2) Is it related to resolution?
SPEAR_MED SPEAR_LOW
MAE
(years)
3) Is it systematic in CMIP6?
104
Machine Learning Explainability Methods (Attribution) for CESM2-LE 105
Decrease
likelihood of year
Increase
likelihood of year
4) So, what is it? The land?
106
TRENDS FROM 1921 TO 1950
Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings
107
Warmer
Colder
TRENDS FROM 1921 TO 1950
Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings
108
Warmer
Colder
TRENDS FROM 1921 TO 1950
Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings
109
Warmer
Colder
TRENDS IN EVAPORATION
Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings
110
Increase
Decrease
111
ERA5
1979-2021
Input maps of P-E
for Western USA
Machine Learning
Explainability Methods
SUMMER
EVAPORATION
SUMMER
P-E
112
113
What about where we live?
Summer (June-August)
Eastern USA
X = mean
1. Increasing spatial resolution
improves the neural network skill
for identifying climate signals
2. Externally-forced temperature
signals have emerged in
observations in the United States
3. Trends in western United States
land surface fields are linked to
timing of emergence in SPEAR
KEY FINDINGS
• Machine
learning
prediction
skill
• ToE in
observations
• Physical
drivers of
climate
signals
Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), Forced signals in United States summer
temperatures revealed by explainable neural networks in climate models and observations, in prep
1. Increasing spatial resolution
improves the neural network skill
for identifying climate signals
2. Externally-forced temperature
signals have emerged in
observations in the United States
3. Trends in western United States
land surface fields are linked to
timing of emergence in SPEAR
KEY FINDINGS
• Machine
learning
prediction
skill
• ToE in
observations
• Physical
drivers of
climate
signals
Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), Forced signals in United States summer
temperatures revealed by explainable neural networks in climate models and observations, in prep
1. Increasing spatial resolution
improves the neural network skill
for identifying climate signals
2. Externally-forced temperature
signals have emerged in
observations in the United States
3. Trends in western United States
land surface fields are linked to
timing of emergence in SPEAR
KEY FINDINGS
• Machine
learning
prediction
skill
• ToE in
observations
• Physical
drivers of
climate
signals
Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), Forced signals in United States summer
temperatures revealed by explainable neural networks in climate models and observations, in prep
INPUT
[DATA]
PREDICTION
Machine
Learning
Explainable (or interpretable) AI
Learn new
climate science!
117
XAI can identify regional patterns of
climate change & variability
in GFDL large ensembles.
1)
118
Method can identify differences in
time-evolving climate signals between
other climate model large ensembles.
2)
119
Framework can be adapted for
monitoring and predicting patterns
of climate change in observations.
3)
120
DECADAL CLIMATE PREDICTION
Explainable machine learning for improving prediction skill
and identifying physical drivers
DETECTION AND ATTRIBUTION
Classification neural network for monitoring extreme events
in climate models and observations
FUTURE DIRECTIONS AT GFDL
DIAGNOSTIC TOOL FOR MODEL BIASES
Method for extracting forced climate signals across Earth system models
TAKEAWAYS
1. XAI can identify regional patterns of climate change & variability in GFDL large ensembles.
2. Method can identify differences in time-evolving forced climate signals between other
climate model large ensembles.
3. Framework can be adapted for monitoring and predicting patterns of climate change in
observations.
Zack Labe
zachary.labe@noaa.gov
122
15 March 2023
GFDL Lunchtime Seminar

More Related Content

Similar to Using explainable AI to identify key regions of climate change in GFDL SPEAR large ensembles

Machine learning for evaluating climate model projections
Machine learning for evaluating climate model projectionsMachine learning for evaluating climate model projections
Machine learning for evaluating climate model projections
Zachary Labe
 
Learning new climate science by opening the machine learning black box
Learning new climate science by opening the machine learning black boxLearning new climate science by opening the machine learning black box
Learning new climate science by opening the machine learning black box
Zachary Labe
 
Reanalysis Datasets for Solar Resource Assessment - 2014 SPI
Reanalysis Datasets for Solar Resource Assessment - 2014 SPI Reanalysis Datasets for Solar Resource Assessment - 2014 SPI
Reanalysis Datasets for Solar Resource Assessment - 2014 SPI
Gwendalyn Bender
 
3178_IGARSS11.ppt
3178_IGARSS11.ppt3178_IGARSS11.ppt
3178_IGARSS11.ppt
grssieee
 
4 ROMAN_IGARSS'11.ppt
4 ROMAN_IGARSS'11.ppt4 ROMAN_IGARSS'11.ppt
4 ROMAN_IGARSS'11.ppt
grssieee
 
Program on Mathematical and Statistical Methods for Climate and the Earth Sys...
Program on Mathematical and Statistical Methods for Climate and the Earth Sys...Program on Mathematical and Statistical Methods for Climate and the Earth Sys...
Program on Mathematical and Statistical Methods for Climate and the Earth Sys...
The Statistical and Applied Mathematical Sciences Institute
 

Similar to Using explainable AI to identify key regions of climate change in GFDL SPEAR large ensembles (20)

Exploring climate change signals with explainable AI
Exploring climate change signals with explainable AIExploring climate change signals with explainable AI
Exploring climate change signals with explainable AI
 
Applications of machine learning for climate change and variability
Applications of machine learning for climate change and variabilityApplications of machine learning for climate change and variability
Applications of machine learning for climate change and variability
 
Using explainable neural networks for comparing climate model projections
Using explainable neural networks for comparing climate model projectionsUsing explainable neural networks for comparing climate model projections
Using explainable neural networks for comparing climate model projections
 
An intro to explainable AI for polar climate science
An intro to  explainable AI for  polar climate scienceAn intro to  explainable AI for  polar climate science
An intro to explainable AI for polar climate science
 
Big Data, Big Computing, AI, and Environmental Science
Big Data, Big Computing, AI, and Environmental ScienceBig Data, Big Computing, AI, and Environmental Science
Big Data, Big Computing, AI, and Environmental Science
 
Machine learning for evaluating climate model projections
Machine learning for evaluating climate model projectionsMachine learning for evaluating climate model projections
Machine learning for evaluating climate model projections
 
Learning new climate science by opening the machine learning black box
Learning new climate science by opening the machine learning black boxLearning new climate science by opening the machine learning black box
Learning new climate science by opening the machine learning black box
 
Making effective science figures
Making effective science figuresMaking effective science figures
Making effective science figures
 
Using explainable machine learning for evaluating patterns of climate change
Using explainable machine learning for evaluating patterns of climate changeUsing explainable machine learning for evaluating patterns of climate change
Using explainable machine learning for evaluating patterns of climate change
 
The Emerging Cyberinfrastructure for Earth and Ocean Sciences
The Emerging Cyberinfrastructure for Earth and Ocean SciencesThe Emerging Cyberinfrastructure for Earth and Ocean Sciences
The Emerging Cyberinfrastructure for Earth and Ocean Sciences
 
Reanalysis Datasets for Solar Resource Assessment - 2014 SPI
Reanalysis Datasets for Solar Resource Assessment - 2014 SPI Reanalysis Datasets for Solar Resource Assessment - 2014 SPI
Reanalysis Datasets for Solar Resource Assessment - 2014 SPI
 
Evaluating global climate models using simple, explainable neural networks
Evaluating global climate models using simple, explainable neural networksEvaluating global climate models using simple, explainable neural networks
Evaluating global climate models using simple, explainable neural networks
 
EcoTas13 BradEvans e-MAST
EcoTas13 BradEvans e-MASTEcoTas13 BradEvans e-MAST
EcoTas13 BradEvans e-MAST
 
The Large Interferometer For Exoplanets (LIFE): the science of characterising...
The Large Interferometer For Exoplanets (LIFE): the science of characterising...The Large Interferometer For Exoplanets (LIFE): the science of characterising...
The Large Interferometer For Exoplanets (LIFE): the science of characterising...
 
A PHYSICAL METHOD TO COMPUTE SURFACE RADIATION FROM GEOSTATIONARY SATELLITES
A PHYSICAL METHOD TO COMPUTE SURFACE RADIATION FROM GEOSTATIONARY SATELLITES  A PHYSICAL METHOD TO COMPUTE SURFACE RADIATION FROM GEOSTATIONARY SATELLITES
A PHYSICAL METHOD TO COMPUTE SURFACE RADIATION FROM GEOSTATIONARY SATELLITES
 
3178_IGARSS11.ppt
3178_IGARSS11.ppt3178_IGARSS11.ppt
3178_IGARSS11.ppt
 
Understanding climate model evaluation and validation
Understanding climate model evaluation and validationUnderstanding climate model evaluation and validation
Understanding climate model evaluation and validation
 
4 ROMAN_IGARSS'11.ppt
4 ROMAN_IGARSS'11.ppt4 ROMAN_IGARSS'11.ppt
4 ROMAN_IGARSS'11.ppt
 
Program on Mathematical and Statistical Methods for Climate and the Earth Sys...
Program on Mathematical and Statistical Methods for Climate and the Earth Sys...Program on Mathematical and Statistical Methods for Climate and the Earth Sys...
Program on Mathematical and Statistical Methods for Climate and the Earth Sys...
 
2 1 xie_solar_2016_pv_systems
2 1 xie_solar_2016_pv_systems2 1 xie_solar_2016_pv_systems
2 1 xie_solar_2016_pv_systems
 

More from Zachary Labe

More from Zachary Labe (20)

Welcome to GFDL for Take Your Child To Work Day
Welcome to GFDL for Take Your Child To Work DayWelcome to GFDL for Take Your Child To Work Day
Welcome to GFDL for Take Your Child To Work Day
 
Reexamining future projections of Arctic climate linkages
Reexamining future projections of Arctic climate linkagesReexamining future projections of Arctic climate linkages
Reexamining future projections of Arctic climate linkages
 
Techniques and Considerations for Improving Accessibility in Online Media
Techniques and Considerations for Improving Accessibility in Online MediaTechniques and Considerations for Improving Accessibility in Online Media
Techniques and Considerations for Improving Accessibility in Online Media
 
Using accessible data to communicate global climate change
Using accessible data to communicate global climate changeUsing accessible data to communicate global climate change
Using accessible data to communicate global climate change
 
Water in a Frozen Arctic: Cross-Disciplinary Perspectives
Water in a Frozen Arctic: Cross-Disciplinary PerspectivesWater in a Frozen Arctic: Cross-Disciplinary Perspectives
Water in a Frozen Arctic: Cross-Disciplinary Perspectives
 
Distinguishing the regional emergence of United States summer temperatures be...
Distinguishing the regional emergence of United States summer temperatures be...Distinguishing the regional emergence of United States summer temperatures be...
Distinguishing the regional emergence of United States summer temperatures be...
 
Researching and Communicating Our Changing Climate
Researching and Communicating Our Changing ClimateResearching and Communicating Our Changing Climate
Researching and Communicating Our Changing Climate
 
Revisiting projections of Arctic climate change linkages
Revisiting projections of Arctic climate change linkagesRevisiting projections of Arctic climate change linkages
Revisiting projections of Arctic climate change linkages
 
Visualizing climate change through data
Visualizing climate change through dataVisualizing climate change through data
Visualizing climate change through data
 
Contrasting polar climate change in the past, present, and future
Contrasting polar climate change in the past, present, and futureContrasting polar climate change in the past, present, and future
Contrasting polar climate change in the past, present, and future
 
Climate change extremes by season in the United States
Climate change extremes by season in the United StatesClimate change extremes by season in the United States
Climate change extremes by season in the United States
 
Guest Lecture: Our changing Arctic in the past and future
Guest Lecture: Our changing Arctic in the past and futureGuest Lecture: Our changing Arctic in the past and future
Guest Lecture: Our changing Arctic in the past and future
 
Climate Projections - What Really is Business as Usual?
Climate Projections - What Really is Business as Usual?Climate Projections - What Really is Business as Usual?
Climate Projections - What Really is Business as Usual?
 
Monitoring indicators of climate change through data-driven visualization
Monitoring indicators of climate change through data-driven visualizationMonitoring indicators of climate change through data-driven visualization
Monitoring indicators of climate change through data-driven visualization
 
Sea Ice Anomalies
Sea Ice AnomaliesSea Ice Anomalies
Sea Ice Anomalies
 
Techniques and Considerations for Improving Accessibility in Online Media
Techniques and Considerations for Improving Accessibility in Online MediaTechniques and Considerations for Improving Accessibility in Online Media
Techniques and Considerations for Improving Accessibility in Online Media
 
Career pathways and research opportunities in the Earth sciences
Career pathways and research opportunities in the Earth sciencesCareer pathways and research opportunities in the Earth sciences
Career pathways and research opportunities in the Earth sciences
 
Telling data-driven climate stories
Telling data-driven climate storiesTelling data-driven climate stories
Telling data-driven climate stories
 
Evaluating and communicating Arctic climate change projection
Evaluating and communicating Arctic climate change projectionEvaluating and communicating Arctic climate change projection
Evaluating and communicating Arctic climate change projection
 
Arctic climate through the lens of data visualization
Arctic climate through the lens of data visualizationArctic climate through the lens of data visualization
Arctic climate through the lens of data visualization
 

Recently uploaded

POGONATUM : morphology, anatomy, reproduction etc.
POGONATUM : morphology, anatomy, reproduction etc.POGONATUM : morphology, anatomy, reproduction etc.
POGONATUM : morphology, anatomy, reproduction etc.
Silpa
 
The Mariana Trench remarkable geological features on Earth.pptx
The Mariana Trench remarkable geological features on Earth.pptxThe Mariana Trench remarkable geological features on Earth.pptx
The Mariana Trench remarkable geological features on Earth.pptx
seri bangash
 
LUNULARIA -features, morphology, anatomy ,reproduction etc.
LUNULARIA -features, morphology, anatomy ,reproduction etc.LUNULARIA -features, morphology, anatomy ,reproduction etc.
LUNULARIA -features, morphology, anatomy ,reproduction etc.
Silpa
 
Cyathodium bryophyte: morphology, anatomy, reproduction etc.
Cyathodium bryophyte: morphology, anatomy, reproduction etc.Cyathodium bryophyte: morphology, anatomy, reproduction etc.
Cyathodium bryophyte: morphology, anatomy, reproduction etc.
Silpa
 
Digital Dentistry.Digital Dentistryvv.pptx
Digital Dentistry.Digital Dentistryvv.pptxDigital Dentistry.Digital Dentistryvv.pptx
Digital Dentistry.Digital Dentistryvv.pptx
MohamedFarag457087
 
Reboulia: features, anatomy, morphology etc.
Reboulia: features, anatomy, morphology etc.Reboulia: features, anatomy, morphology etc.
Reboulia: features, anatomy, morphology etc.
Silpa
 
CYTOGENETIC MAP................ ppt.pptx
CYTOGENETIC MAP................ ppt.pptxCYTOGENETIC MAP................ ppt.pptx
CYTOGENETIC MAP................ ppt.pptx
Silpa
 

Recently uploaded (20)

POGONATUM : morphology, anatomy, reproduction etc.
POGONATUM : morphology, anatomy, reproduction etc.POGONATUM : morphology, anatomy, reproduction etc.
POGONATUM : morphology, anatomy, reproduction etc.
 
Gwalior ❤CALL GIRL 84099*07087 ❤CALL GIRLS IN Gwalior ESCORT SERVICE❤CALL GIRL
Gwalior ❤CALL GIRL 84099*07087 ❤CALL GIRLS IN Gwalior ESCORT SERVICE❤CALL GIRLGwalior ❤CALL GIRL 84099*07087 ❤CALL GIRLS IN Gwalior ESCORT SERVICE❤CALL GIRL
Gwalior ❤CALL GIRL 84099*07087 ❤CALL GIRLS IN Gwalior ESCORT SERVICE❤CALL GIRL
 
PSYCHOSOCIAL NEEDS. in nursing II sem pptx
PSYCHOSOCIAL NEEDS. in nursing II sem pptxPSYCHOSOCIAL NEEDS. in nursing II sem pptx
PSYCHOSOCIAL NEEDS. in nursing II sem pptx
 
The Mariana Trench remarkable geological features on Earth.pptx
The Mariana Trench remarkable geological features on Earth.pptxThe Mariana Trench remarkable geological features on Earth.pptx
The Mariana Trench remarkable geological features on Earth.pptx
 
Dr. E. Muralinath_ Blood indices_clinical aspects
Dr. E. Muralinath_ Blood indices_clinical  aspectsDr. E. Muralinath_ Blood indices_clinical  aspects
Dr. E. Muralinath_ Blood indices_clinical aspects
 
LUNULARIA -features, morphology, anatomy ,reproduction etc.
LUNULARIA -features, morphology, anatomy ,reproduction etc.LUNULARIA -features, morphology, anatomy ,reproduction etc.
LUNULARIA -features, morphology, anatomy ,reproduction etc.
 
Cyanide resistant respiration pathway.pptx
Cyanide resistant respiration pathway.pptxCyanide resistant respiration pathway.pptx
Cyanide resistant respiration pathway.pptx
 
Clean In Place(CIP).pptx .
Clean In Place(CIP).pptx                 .Clean In Place(CIP).pptx                 .
Clean In Place(CIP).pptx .
 
Factory Acceptance Test( FAT).pptx .
Factory Acceptance Test( FAT).pptx       .Factory Acceptance Test( FAT).pptx       .
Factory Acceptance Test( FAT).pptx .
 
300003-World Science Day For Peace And Development.pptx
300003-World Science Day For Peace And Development.pptx300003-World Science Day For Peace And Development.pptx
300003-World Science Day For Peace And Development.pptx
 
Cyathodium bryophyte: morphology, anatomy, reproduction etc.
Cyathodium bryophyte: morphology, anatomy, reproduction etc.Cyathodium bryophyte: morphology, anatomy, reproduction etc.
Cyathodium bryophyte: morphology, anatomy, reproduction etc.
 
Genome sequencing,shotgun sequencing.pptx
Genome sequencing,shotgun sequencing.pptxGenome sequencing,shotgun sequencing.pptx
Genome sequencing,shotgun sequencing.pptx
 
Digital Dentistry.Digital Dentistryvv.pptx
Digital Dentistry.Digital Dentistryvv.pptxDigital Dentistry.Digital Dentistryvv.pptx
Digital Dentistry.Digital Dentistryvv.pptx
 
Reboulia: features, anatomy, morphology etc.
Reboulia: features, anatomy, morphology etc.Reboulia: features, anatomy, morphology etc.
Reboulia: features, anatomy, morphology etc.
 
CYTOGENETIC MAP................ ppt.pptx
CYTOGENETIC MAP................ ppt.pptxCYTOGENETIC MAP................ ppt.pptx
CYTOGENETIC MAP................ ppt.pptx
 
module for grade 9 for distance learning
module for grade 9 for distance learningmodule for grade 9 for distance learning
module for grade 9 for distance learning
 
PATNA CALL GIRLS 8617370543 LOW PRICE ESCORT SERVICE
PATNA CALL GIRLS 8617370543 LOW PRICE ESCORT SERVICEPATNA CALL GIRLS 8617370543 LOW PRICE ESCORT SERVICE
PATNA CALL GIRLS 8617370543 LOW PRICE ESCORT SERVICE
 
Atp synthase , Atp synthase complex 1 to 4.
Atp synthase , Atp synthase complex 1 to 4.Atp synthase , Atp synthase complex 1 to 4.
Atp synthase , Atp synthase complex 1 to 4.
 
GBSN - Biochemistry (Unit 2) Basic concept of organic chemistry
GBSN - Biochemistry (Unit 2) Basic concept of organic chemistry GBSN - Biochemistry (Unit 2) Basic concept of organic chemistry
GBSN - Biochemistry (Unit 2) Basic concept of organic chemistry
 
Human & Veterinary Respiratory Physilogy_DR.E.Muralinath_Associate Professor....
Human & Veterinary Respiratory Physilogy_DR.E.Muralinath_Associate Professor....Human & Veterinary Respiratory Physilogy_DR.E.Muralinath_Associate Professor....
Human & Veterinary Respiratory Physilogy_DR.E.Muralinath_Associate Professor....
 

Using explainable AI to identify key regions of climate change in GFDL SPEAR large ensembles

  • 1. Using explainable AI to identify key regions of climate change in GFDL SPEAR large ensembles https://zacklabe.com/ @ZLabe Zachary Labe Postdoc in Seasonal-to-Decadal (S2D) Variability and Predictability Division with Nathaniel Johnson and Thomas Delworth 15 March 2023 GFDL Lunchtime Seminar
  • 2. 1 1) Where do we go from here?
  • 3. 2 1) Where do we go from here? 2) How do we disentangle internal climate variability? Feb/Mar 2016
  • 4. 3 3) How do we account for regional patterns of change?
  • 5. Explainable machine learning can distinguish between regional patterns of time-evolving climate change in GFDL models. 4 SIGNIFICANCE
  • 6. Machine Learning is not new! “A Bayesian Neural Network for Severe-Hail Prediction (2000)” “Classification of Convective Areas Using Decision Trees (2009)” “A Neural Network for Damaging Wind Prediction (1998)” “Generative Additive Models versus Linear Regression in Generating Probabilistic MOS Forecasts of Aviation Weather Parameters (1995)” ”A Neural Network for Tornado Prediction Based on Doppler Radar-Derived Attributes (1996)” ”The Diagnosis of Upper-Level Humidity (1968)” 5
  • 7. “An adaptive data processing system for weather forecasting” It’s a neural network! [Hu and Root (1964), APME] 6
  • 8. Artificial Intelligence Machine Learning Deep Learning Computer/Data Science 7
  • 9. Computer/Data Science Supervised Learning Unsupervised Learning Labeled data Classification Regression Unlabeled data Clustering Dimension reduction 8 Artificial Intelligence Machine Learning Deep Learning DATA-HUNGRY!
  • 10. Do it better e.g., parameterizations in climate models are not perfect, use ML to make them more accurate Do it faster e.g., code in climate models is very slow (but we know the right answer) - use ML methods to speed things up Do something new • e.g., go looking for non-linear relationships you didn’t know were there WHY ELSE SHOULD WE CONSIDER MACHINE LEARNING? 9
  • 11. Do it better e.g., parameterizations in climate models are not perfect, use ML to make them more accurate Do it faster e.g., code in climate models is very slow (but we know the right answer) - use ML methods to speed things up Do something new • e.g., go looking for non-linear relationships you didn’t know were there Very relevant for research: may be slower and worse, but can still learn something 10 WHY ELSE SHOULD WE CONSIDER MACHINE LEARNING?
  • 12. Machine learning for meteorology IDENTIFYING SEVERE THUNDERSTORMS Molina et al. 2021 Martin et al. 2022 CLASSIFYING PHASE OF MADDEN-JULLIAN OSCILLATION DETECTING CONVECTION FROM SATELLITES Lee et al. 2021 LOCATING COLD FRONTS Dagon et al. 2022 11
  • 13. Machine learning for oceanography CLASSIFYING ARCTIC OCEAN ACIDIFICATION Krasting et al. 2022 LARGE-SCALE OCEAN CIRCULATION Clare et al. 2022 ESTIMATING OCEAN SURFACE CURRENTS Sinha and Abernathey, 2021 12
  • 14. Machine learning for climate PHYSICAL DRIVERS OF ENSO DYNAMICS Shin et al. 2022 IDENTIFYING DECADAL STATE DEPENDENCE Gordon and Barnes, 2022 INTERNAL/EXTERNAL CLIMATE FORCING Po-Chedley et al. 2022 13
  • 18. Artificial Intelligence Machine Learning Deep Learning Artificial Neural Networks 17 Computer/Data Science
  • 20. Linear regression! Artificial Neural Networks [ANN] X1 X2 W1 W2 ∑ = X1W1+ X2W2 + b INPUTS NODE 19
  • 21. Artificial Neural Networks [ANN] X1 X2 W1 W2 ∑ INPUTS NODE Linear regression with non-linear mapping by an “activation function” Training of the network is merely determining the weights “w” and bias/offset “b" = factivation(X1W1+ X2W2 + b) 20
  • 22. Artificial Neural Networks [ANN] X1 X2 W1 W2 ∑ INPUTS NODE = factivation(X1W1+ X2W2 + b) ReLU Sigmoid Linear 21
  • 24. Complexity and nonlinearities of the ANN allow it to learn many different pathways of predictable behavior Once trained, you have an array of weights and biases which can be used for prediction on new data INPUT [DATA] PREDICTION Artificial Neural Networks [ANN] 23
  • 25. What is the annual mean temperature of Earth? 24
  • 26. THE REAL WORLD (Observations) What is the annual mean temperature of Earth? Data from Berkeley Earth Surface Temperature 1930 2022 25
  • 27. What is the annual mean temperature of Earth? THE REAL WORLD (Observations) Let’s run a climate model One ensemble member 2022 1930 2050 26 Data from SPEAR_M ED
  • 28. What is the annual mean temperature of Earth? THE REAL WORLD (Observations) Let’s run a climate model again! Two ensemble members 27 Data from SPEAR_M ED
  • 29. What is the annual mean temperature of Earth? THE REAL WORLD (Observations) Let’s run a climate model again & again! Three ensemble members 28 Data from SPEAR_M ED
  • 30. What is the annual mean temperature of Earth? THE REAL WORLD (Observations) CLIMATE MODEL LARGE ENSEMBLE 30 ensemble members in GFDL SPEAR 29
  • 31. What is the annual mean temperature of Earth? Mean of ensembles = forced response (climate change) THE REAL WORLD (Observations) CLIMATE MODEL LARGE ENSEMBLE 30 ensemble members In GFDL SPEAR 30
  • 32. What is the annual mean temperature of Earth? Mean of ensembles = forced response (climate change) THE REAL WORLD (Observations) CLIMATE MODEL LARGE ENSEMBLE 30 ensemble members In GFDL SPEAR Range of ensembles = internal variability (noise) Mean of ensembles = forced response (climate change) 31
  • 33. But let’s remove climate change… Climate Change Signal (ensemble mean) Observations Ensemble Members 32
  • 34. Ensemble Members Mean of anomalies After removing the forced response… = anomalies/noise! 33
  • 35. Ensemble members in GFDL SPEAR Maps of a given time period for each ensemble Inputs for machine learning 34
  • 36. Ensemble members in GFDL SPEAR 35 Training Data: 24 ensemble members Maps of a given time period for each ensemble
  • 37. 36 Training Data: 24 ensemble members Validation Data: 4 ensemble members
  • 38. 37 Training Data: 24 ensemble members Validation Data: 4 ensemble members Testing Data: 2 ensemble members
  • 39. Historical Forcing – GFDL SPEAR Future Scenarios – GFDL SPEAR 38 Can a neural network learn unique patterns of climate change related to each future emission scenario? 1930 2010 2020 2100
  • 40. 39 Train a neural network to predict 5 classes (climate scenarios)
  • 41. Yearly Maps of T2M Yearly Maps of T2M Neural Network Classify Climate Scenario Artificial Neural Network Output = 5 Classes Yearly Maps of T2M Neural Network Binary Output Binary Output 40 Step #1 Read in gridded maps of a climate variable from SPEAR simulations
  • 42. Yearly Maps of T2M Yearly Maps of T2M Neural Network Classify Climate Scenario Artificial Neural Network Output = 5 Classes Yearly Maps of T2M Neural Network Binary Output Binary Output 41 Step #2 Feed data into an artificial neural network with three hidden layers
  • 43. Yearly Maps of T2M Yearly Maps of T2M Neural Network Classify Climate Scenario Artificial Neural Network Output = 5 Classes Yearly Maps of T2M Neural Network Binary Output Binary Output 42 Step #3 Classify which climate scenario (n=5) is associated with each map
  • 44. Yearly Maps of T2M Yearly Maps of T2M Neural Network Classify Climate Scenario Artificial Neural Network Output = 5 Classes Yearly Maps of T2M Neural Network Binary Output Binary Output 43 Step #4 Why? à XAI
  • 45. WHY LAYER-WISE RELEVANCE PROPAGATION (LRP) Volcano Timber Wolf Image Classification LRP LRP heatmaps show regions of “relevance” that contribute to the neural network’s decision-making process for a sample belonging to a particular output category Neural Network WHY Backpropagation – LRP https://heatmapping.org/ 44
  • 46. WHY LAYER-WISE RELEVANCE PROPAGATION (LRP) Volcano Timber Wolf Image Classification LRP LRP heatmaps show regions of “relevance” that contribute to the neural network’s decision-making process for a sample belonging to a particular output category Neural Network WHY Backpropagation – LRP https://heatmapping.org/ 45
  • 47. Image Classification LRP https://heatmapping.org/ NOT PERFECT Crock Pot Neural Network Backpropagation – LRP WHY 46 LAYER-WISE RELEVANCE PROPAGATION (LRP)
  • 48. EXPLAINABLE AI (XAI) THERE ARE MANY METHODS A bird! XAI 47 [Adapted from Adebayo et al., 2020]
  • 49. THERE ARE MANY METHODS EXPLAINABLE AI (XAI) 48 [Adapted from Adebayo et al., 2020]
  • 50. Visualizing something we already know… ENSO 49
  • 51. Neural Network [0] La Niña [1] El Niño Input a map of sea surface temperatures 50 [Toms et al. 2020, JAMES]
  • 52. Visualizing something we already know… Input maps of sea surface temperatures (SST) to identify El Niño or La Niña Use ‘LRP’ to see how the neural network is making its decision [Toms et al. 2020, JAMES] Layer-wise Relevance Propagation Composite SST Observations LRP [Relevance] SST Anomaly [°C] 0.00 0.75 0.0 1.5 -1.5 51 Warmer Colder High Low
  • 53. 52 Returning to our application…
  • 54. 53 Predictions for SPEAR_MED Testing Data Accuracy=92% Nearer to predicted class Further from predicted class
  • 55. 54 Predictions for SPEAR_MED Testing Data Accuracy=92% Nearer to predicted class Further from predicted class
  • 56. Global Mean Surface Temperature Can we identify changes in future climate impacts after rapid mitigation? 55
  • 57. 56 30 ensembles for GFDL SPEAR_MED 9 ensembles for GFDL SPEAR_MED Input maps from out-of-sample ensembles into classification network 2020 2030 2100
  • 58. 57 2031 2040 Rapid Mitigation Rapid Mitigation 30 ensembles for GFDL SPEAR_MED 9 ensembles for GFDL SPEAR_MED
  • 59. 58 Predictions for a single ensemble member from SSP5-3.4OS SSP5-8.5 SSP2-4.5 2015 2055 2065 2095
  • 60. 59 Predictions for a single ensemble member from SSP5-3.4OS SSP5-8.5 SSP2-4.5 2055-2060 rapid mitigation begins 2015 2095
  • 61. 60 Are these predictions robust across ensemble members? (n=30) SSP5-3.4OS Transition from SSP5-8.5 to SSP2-4.5 2015 2060 2100
  • 62. 61 What if we start mitigation 10 years earlier? Transition from SSP5-8.5 to SSP2-4.5 Transition from SSP5-8.5 to SSP2-4.5 Transition from SSP2-4.5 to SSP1-1.9 2015 2060 2100 2015 2060 2100
  • 63. 62 What climate patterns are associated with these transitions? Transition from SSP5-8.5 to SSP2-4.5 Transition from SSP5-8.5 to SSP2-4.5 Transition from SSP2-4.5 to SSP1-1.9 Rapid mitigation Rapid mitigation
  • 64. 63 Difficult to distinguish the patterns associated with scenario transitions Composites of relevance maps for the mitigation predictions SSP5-3.4OS example for 2015 to 2100 Nearer to predicted scenario Further from predicted scenario
  • 65. Yearly Maps of T2M Yearly Maps of T2M Neural Network Classify Climate Scenario Artificial Neural Network Output = 5 Classes Yearly Maps of T2M Neural Network Binary Output Binary Output 64 Steps #4-5
  • 66. 65 XAI composites of years associated with the transition from SSP5-8.5 to SSP2-4.5 (a) approx. 2055-2060 (b) approx. 2040-2045
  • 67. 66 North Atlantic is an important indictor region for climate signals related to identifying from SSP5-8.5 to SSP2-4.5 Future Climate Change Rapid Mitigation
  • 68. 67 Framework can be applied to different geographic regions and climate variables Parallel approach for detecting climate intervention scenarios Labe, Z.M., E.A. Barnes, and J.W. Hurrell (2023). Identifying the regional emergence of climate patterns in the ARISE-SAI-1.5 simulations. EarthArXiv, DOI: 10.31223/X5394Z
  • 69. 1) Explainable AI methods identify regions of climate change that indicate the emission scenario 2) Framework for outlining benefits from mitigation by testing out of sample ensembles from SSP5-3.4OS 3) Patterns of fast and slow responses to rapid climate mitigation are revealed by explainable AI KEY FINDINGS • Indicator patterns of climate change • Quantify benefits of rapid climate mitigation • Regions of rapid or slow response to mitigation Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), A data-driven approach to identifying key regions of change associated with rapid climate mitigation, in prep
  • 70. 1) Explainable AI methods identify regions of climate change that indicate the emission scenario 2) Framework for outlining benefits from mitigation by testing out of sample ensembles from SSP5-3.4OS 3) Patterns of fast and slow responses to rapid climate mitigation are revealed by explainable AI KEY FINDINGS • Indicator patterns of climate change • Quantify benefits of rapid climate mitigation • Regions of rapid or slow response to mitigation Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), A data-driven approach to identifying key regions of change associated with rapid climate mitigation, in prep
  • 71. 1) Explainable AI methods identify regions of climate change that indicate the emission scenario 2) Framework for outlining benefits from mitigation by testing out of sample ensembles from SSP5-3.4OS 3) Patterns of fast and slow responses to rapid climate mitigation are revealed by explainable AI KEY FINDINGS • Indicator patterns of climate change • Quantify benefits of rapid climate mitigation • Regions of rapid or slow response to mitigation Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), A data-driven approach to identifying key regions of change associated with rapid climate mitigation, in prep
  • 73. 72
  • 75. [Eischeid et al. 2023, revised] 74
  • 76. Temperature anomalies [ °C ] relative to 1981-2010 Observations from NClimGrid Climate model data from GFDL SPEAR_MED United States – Summer 75 1920 2020
  • 78. TEMPERATURE We know some metadata… + What year is it? + Where did it come from? 77
  • 79. TEMPERATURE We know some metadata… + What year is it? + Where did it come from? 78
  • 80. We know some metadata… + What year is it? + Where did it come from? [Labe and Barnes, 2022; ESS] TEMPERATURE 79
  • 81. TEMPERATURE Neural network learns nonlinear combinations of forced climate patterns to identify the year We know some metadata… + What year is it? + Where did it come from? [Labe and Barnes, 2022; ESS] 80
  • 82. ----ANN---- 2 Hidden Layers 10 Nodes each Ridge Regularization Early Stopping [e.g., Barnes et al. 2019, 2020] [e.g., Labe and Barnes, 2021] TIMING OF EMERGENCE (COMBINED VARIABLES) RESPONSES TO EXTERNAL CLIMATE FORCINGS PATTERNS OF CLIMATE INDICATORS [e.g., Rader et al. 2022] Surface Temperature Map Precipitation Map + TEMPERATURE We know some metadata… + What year is it? + Where did it come from? [Labe and Barnes, 2022; ESS] 81
  • 83. Seasonal Maps of T2M, TMAX, TMIN Input 82
  • 84. Seasonal Maps of T2M, TMAX, TMIN Hidden Layers Artificial Neural Network Input 83
  • 85. Seasonal Maps of T2M, TMAX, TMIN Hidden Layers Output 1921 2100 Artificial Neural Network Input 84
  • 86. Backpropagation Seasonal Maps of T2M, TMAX, TMIN Hidden Layers Output 1921 2100 Artificial Neural Network Input 85 Post hoc – XAI methods
  • 89. 88 1921 2021 2100 ACTUAL YEARS PREDICTED YEARS NOAA Monthly U.S. NClimGrid v1.0 1921 2021 2100
  • 90. Max year predicted in the 1921-1950 baseline for observations Timing of Emergence 89 1921-1950 Skill 1:1 Perfect Prediction
  • 92. June – August – Timing of Emergence (ToE) for observations 91 June – August – skill for observations
  • 93. How is the neural network able to detect the year prior to ~1990? Temperature anomalies [ °C ] relative to 1981-2010 Machine learning predictions GFDL SPEAR_MED simulation 92
  • 94. Machine Learning Explainability Methods – Attribution 93 Decrease likelihood of year Increase likelihood of year
  • 95. Western USA Central USA Eastern USA 94
  • 96. Western USA Central USA Eastern USA 95
  • 97. Western USA Central USA Eastern USA 96
  • 99. First year that the 10-year running-mean temperature exceeds and stays above the mean 1921–1950 reference temperature by more than two standard deviations Most Areas = 2000s 98
  • 102. 1) Is it aerosols? 101 Only available from 1921 to 2020
  • 104. 50 km resolution 100 km resolution 103 2) Is it related to resolution? SPEAR_MED SPEAR_LOW MAE (years)
  • 105. 3) Is it systematic in CMIP6? 104
  • 106. Machine Learning Explainability Methods (Attribution) for CESM2-LE 105 Decrease likelihood of year Increase likelihood of year
  • 107. 4) So, what is it? The land? 106
  • 108. TRENDS FROM 1921 TO 1950 Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings 107 Warmer Colder
  • 109. TRENDS FROM 1921 TO 1950 Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings 108 Warmer Colder
  • 110. TRENDS FROM 1921 TO 1950 Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings 109 Warmer Colder
  • 111. TRENDS IN EVAPORATION Fully-Coupled [Historical + SSP5-8.5] SPEAR_MED, but NO anthropogenic aerosols SPEAR_MED, but NO anthropogenic forcings 110 Increase Decrease
  • 112. 111 ERA5 1979-2021 Input maps of P-E for Western USA
  • 114. 113 What about where we live? Summer (June-August) Eastern USA X = mean
  • 115. 1. Increasing spatial resolution improves the neural network skill for identifying climate signals 2. Externally-forced temperature signals have emerged in observations in the United States 3. Trends in western United States land surface fields are linked to timing of emergence in SPEAR KEY FINDINGS • Machine learning prediction skill • ToE in observations • Physical drivers of climate signals Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), Forced signals in United States summer temperatures revealed by explainable neural networks in climate models and observations, in prep
  • 116. 1. Increasing spatial resolution improves the neural network skill for identifying climate signals 2. Externally-forced temperature signals have emerged in observations in the United States 3. Trends in western United States land surface fields are linked to timing of emergence in SPEAR KEY FINDINGS • Machine learning prediction skill • ToE in observations • Physical drivers of climate signals Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), Forced signals in United States summer temperatures revealed by explainable neural networks in climate models and observations, in prep
  • 117. 1. Increasing spatial resolution improves the neural network skill for identifying climate signals 2. Externally-forced temperature signals have emerged in observations in the United States 3. Trends in western United States land surface fields are linked to timing of emergence in SPEAR KEY FINDINGS • Machine learning prediction skill • ToE in observations • Physical drivers of climate signals Labe, Z.M., N.C. Johnson, and T.L. Delworth (2023), Forced signals in United States summer temperatures revealed by explainable neural networks in climate models and observations, in prep
  • 119. XAI can identify regional patterns of climate change & variability in GFDL large ensembles. 1) 118
  • 120. Method can identify differences in time-evolving climate signals between other climate model large ensembles. 2) 119
  • 121. Framework can be adapted for monitoring and predicting patterns of climate change in observations. 3) 120
  • 122. DECADAL CLIMATE PREDICTION Explainable machine learning for improving prediction skill and identifying physical drivers DETECTION AND ATTRIBUTION Classification neural network for monitoring extreme events in climate models and observations FUTURE DIRECTIONS AT GFDL DIAGNOSTIC TOOL FOR MODEL BIASES Method for extracting forced climate signals across Earth system models
  • 123. TAKEAWAYS 1. XAI can identify regional patterns of climate change & variability in GFDL large ensembles. 2. Method can identify differences in time-evolving forced climate signals between other climate model large ensembles. 3. Framework can be adapted for monitoring and predicting patterns of climate change in observations. Zack Labe zachary.labe@noaa.gov 122 15 March 2023 GFDL Lunchtime Seminar