Exploring climate model large ensembles with explainable neural networks
1. EXPLORING
CLIMATE MODEL LARGE ENSEMBLES
WITH EXPLAINABLE NEURAL NETWORKS
@ZLabe
Zachary M. Labe
with Elizabeth A. Barnes
Colorado State University
Department of Atmospheric Science
22 September 2021
World Climate Research Programme (WCRP)
Workshop on “Attribution of multi-annual to decadal changes in the climate system”
4. We know some metadata…
+ What year is it?
+ Where did it come from?
TEMPERATURE
5. We know some metadata…
+ What year is it?
+ Where did it come from?
TEMPERATURE
Neural network learns nonlinear
combinations of forced climate
patterns to identify the year
6. ----ANN----
2 Hidden Layers
10 Nodes each
Ridge Regularization
Early Stopping
We know some metadata…
+ What year is it?
+ Where did it come from?
[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. in prep]
Surface Temperature Map Precipitation Map
+
TEMPERATURE
7. ----ANN----
2 Hidden Layers
10 Nodes each
Ridge Regularization
Early Stopping
We know some metadata…
+ What year is it?
+ Where did it come from?
[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
Surface Temperature Map Precipitation Map
+
TEMPERATURE
[e.g., Rader et al. in prep]
15. Layer-wise Relevance Propagation
BACK-PROPAGATE THROUGH NETWORK = EXPLAINABLE AI
WHY?
= LRP HEAT MAPS
Find regions of “relevance”
that contribute to the
neural network’s
decision-making process
[Labe and Barnes 2021, JAMES]
16. CLIMATE MODEL DATA PREDICT THE YEAR FROM MAPS OF TEMPERATURE
AEROSOLS
PREVAIL
GREENHOUSE GASES
PREVAIL
STANDARD
CESM1-LE
[Labe and Barnes 2021, JAMES]
17. OBSERVATIONS PREDICT THE YEAR FROM MAPS OF TEMPERATURE
AEROSOLS
PREVAIL
GREENHOUSE GASES
PREVAIL
STANDARD
CESM1-LE
[Labe and Barnes 2021, JAMES]
18. OBSERVATIONS
SLOPES
PREDICT THE YEAR FROM MAPS OF TEMPERATURE
AEROSOLS
PREVAIL
GREENHOUSE GASES
PREVAIL
STANDARD
CESM1-LE
[Labe and Barnes 2021, JAMES]
19. Higher LRP values indicate greater relevance
for the ANN’s prediction
Aerosol-driven
Greenhouse gas-driven
All forcings
Low High
[Labe and Barnes 2021, JAMES]
20. ----ANN----
2 Hidden Layers
10 Nodes each
Ridge Regularization
Early Stopping
TEMPERATURE
We know some metadata…
+ What year is it?
+ Where did it come from?
21. TEMPERATURE
We know some metadata…
+ What year is it?
+ Where did it come from?
Train on data from the
Multi-Model Large
Ensemble Archive
22. TEMPERATURE
We know some metadata…
+ What year is it?
+ Where did it come from?
NEURAL NETWORK
CLASSIFICATION TASK
HIDDEN LAYERS
INPUT LAYER
[Labe and Barnes, in prep]
24. KEY POINTS
Zachary Labe
zmlabe@rams.colostate.edu
@ZLabe
1. We can learn new climate science by using explainable AI methods in conjunction with
existing statistical tools
2. Explainable neural networks reveal patterns of climate change in large ensembles simulated
with different combinations of external forcing
3. Neural networks can be used to identify unique model differences and biases between large
ensemble simulations and observations
Labe, Z.M. and E.A. Barnes (2021), Detecting climate signals using
explainable AI with single-forcing large ensembles. Journal of
Advances in Modeling Earth Systems, DOI:10.1029/2021MS002464