by Judith Curry
Later this week, the Royal Society is hosting a Workshop on Handling Uncertainty in Weather and Climate Prediction, With Application to Health, Agronomy, Hydrology, Energy and Economics.
The web page for the workshop is [here.] From the website:
This meeting follows on from the 2010 Anniversary Discussion Meeting on “Handling Uncertainty in Science” but with a focus on weather and climate prediction and downstream applications. How is uncertainty represented in weather and climate prediction? How reliable are representations of uncertainty? How can decision makers in weather and climate sensitive sectors make useful decisions in the light of uncertain input? Are current ensemble weather and climate prediction systems useful for decision making across a variety of application sectors? How should probability forecasts be presented to the public?
The agenda and list of speakers can be found [here]. I am unfamiliar with about 70% of the speakers, which makes this an exciting conference for me with the prospect of learning new things. The collection of abstracts can be found here [RS abstracts]
I am giving a talk entitled ‘Climate models: fit for what purpose?’ A link to my draft presentation is found here [RS uncertainty 12]. My presentation doesn’t cover a lot of new ground relative to material that I’ve presented previously, but it makes a slightly different argument than I’ve previously made. In particular, it responds to the recent report from the NRC National Strategy for Advancing Climate Models.
Here are the main points from my presentation:
Model credibility and fitness for purpose
Credibility: linked to the model and simulations themselves (expertise and trust)
Fitness for Purpose: suitability and confidence with regards to the (potentially) many uses of the model and simulations.
The user will treat simulation results as ‘fit’ when there is enough evidence that it fulfills its intended purpose, so that useful decisions can be made about the system of interest.
Climate models: fit for decision support?
CO2 mitigation policies:
- GCMs have a role to play, but large ensembles from lower order models with interactive carbon cycle may be the better solution for determining sensitivity
Regional climate change:
- Little to no skill here; increased resolution not helping
- Dynamical & statistical downscaling adds little value
- Many extreme weather events not explicitly simulated
- Depends on poorly simulated modes of natural internal variability
Decision making under uncertainty
The decision-analytic framework influences how climate models are used and developed.
The current focus on the precautionary principle and optimal decision making is driving climate model development & applications in directions for which they are not fit.
Summary
The objective of improving climate models for societal needs is based upon three dubious premises:
- climate models are fit for this purpose
- climate models are useful for this purpose
- climate models are the best choice for this purpose
GCMs are currently incapable of simulating:
- Regional climate variability and change
- Network of teleconnection climate regimes on DEC-CEN timescales
- Predictions of emergent phenomena, e.g. abrupt climate change
It is unlikely that the current path of development will improve this
Fundamental issues for climate models: with the focus on societal needs and the large investment in IPCC production runs, climate models are becoming less fit for the purpose of increasing our understanding of the climate system
Recommendations
Re-orient the science-policy interface to decision making under deep uncertainty, robust decision making and scenario thinking
Re-orient climate modeling development and applications to support the creation of a broad range of future scenarios
- refocus climate model development to emergent phenomena, extreme events, and natural internal variability
- emphasis on ensemble size
More complete exploration of climate model uncertainty, including unknowns and model structural uncertainty
JC comment: Assuming that the presentations are available online, I will post them, and will have at least one additional thread on this Workshop in the coming week. I would appreciate any comments that you have on my presentation.