Hector – a computationally-efficient, open-source, community-based global
and regional climate emulator
Corinne Hartin, Ben Bond-Lamberty, Robert Link, Ben Vega-Westhoff2, Ryan Sriver2,
Abigail Snyder, Ben Kravitz and Kate Calvin
Pacific Northwest National Laboratory/Joint Global Change Research Institute
2University of Illinois
U.S. Department of Energy, Office of Science, as part of research in Multi-Sector Dynamics, Earth and Environmental System Modeling Program
2
What is Climate Emulation?
reproduced from Provenzale, A. (2014)
An emulator is a fast approximation to a computationally expensive model that can be used as a surrogate for the model, to quantify uncertainty or to improve process understanding. (Holden et al., 2014)
Earth System Models (ESMs) are state of the art tools for climate predictions.
However, due to their computational demands they limit the range of studies that can be conducted with them.
What if we wanted to know average global mean temperature projections under other scenarios?
What if we wanted to investigate uncertainty across the climate and carbon systems?
Or the probability of exceeding a particular threshold?
3
What is Climate Emulation?
reproduced from Provenzale, A. (2014)
Idealized Comprehensive
FAIR
IPCC IR
Hector
MAGICC
ACC2
4
Simple Climate Carbon-Cycle Models
4
Meinshausen et al., 2011
5
• Free and open-source
• Community model
www.github.com/JGCRI/hector
Easy to use and well documented
Hartin et al., 2015 – GMD
Hartin et al., 2016 – BGS
pyhector
rhector
Hector GAMS
Issue tracking
Overview of Hector – Comprehensive Simple Climate Carbon-Cycle Model
• Complexity only where warranted
DOECLIM, BRICK
• Modular
Components can be enabled/disabled via inputs
E.g., test two different ocean submodelsagainst each other
• Modern, clean code structure
E.g., coupler enforces unit checking between submodels
Overview of Hector v2.0 - Climate and Carbon Cycle
CO2, CH4, N2O, halocarbons, BC, OC,
sulphate aerosols, strat H2O, trop O3
NPP,
RH,
carbon
fluxespH, CaCO3 saturations, carbon
fluxes
1D diffusive ocean – DOECLIM
for heat transport
Hartin et al., 2015, 2016
www.Github.com/JGCRI/hector
7
Overview of Hector v2.0 - Climate and Carbon Cycle
Hartin et al., 2015 – GMD; Hartin et al., 2016 – BGS
88
Complex Model Emulation – CMIP5
Global Mean Temperature Atmospheric CO2
99
Complex Model Emulation – CMIP5
Global Mean Temperature
10
Probabilistic Capabilities
Global Mean Temperature
HadCRUT
Global Sea-Level Rise
Church and White
Vega-Westhoff et al., submitted to Earth’s Future, 2018
low probability/high impact events
11
Regional Climate Emulation
Multiple realizations over College Park MD with realistic variability of the underlying ESM emulated - CESM1.
www.github.com/JGCRI/fldgen
Link et al., 2018 – GMDD
12
Current Hector Research
• Understanding the response of SCMs
• Probability of ocean acidification
• Updating Hector
Land biomes – permafrost
Schwarber et al., in review - ESD
13
Current Hector Research
• Understanding the response of SCMs
• Probability of ocean acidification
• Updating Hector
Land biomes – permafrost
• Coupled to GCAM
Investigate two-way feedbacks
14
Hector Capabilities
• All model code and tools are open source
Transparent and freely available
Issue tracking
Reproducible code and workflow
“How-to-guides”
Vignettes
• Hector ecosystem
Emulate any model or dataset of interest
Global
Regional
Uncertainty and sensitivity analysis
Solve for an emission pathway
• Easy to couple
GCAM, DOECLIM, BRICK
15
Future Community Involvement
• What would the IAM community would like to see in the future?
• www.github.com/JGCRI/hector
• www.github.com/JGCRI/fldgen
• www.github.com/openclimatedata/pyhector