How to tackle long-standing uncertainties?

Transcription

1 How to tackle long-standing uncertainties? Sandrine Bony LMD/IPSL, CNRS, Paris Thanks to all members of the ClimaConf project Bjorn Stevens (MPI) & the WCRP Grand Challenge team Colloque ClimaConf; November 2014; Paris

2 Outline I. Long-standing challenges in climate science II. III. IV. How to gain confidence in model projections? What's missing? Illustration

3 Outline I. Long-standing challenges in climate science II. III. IV. How to gain confidence in model projections? What's missing? Illustration

4 From the Charney report to the IPCC AR5 Over 30 years, climate scientists have built a research framework : Very successful for detecting and attributing some global aspects of climate change, e. g.: - trends of global-mean temperature - trends of global sea level rise growing level of confidence Much less effective for quantifying other critical aspects of climate change, e. g.: - climate sensitivity - regional changes (e.g. precipitation) confidence remains low

5 Key Challenges of Climate Change Research Long-standing uncertainties (e.g. Climate Sensitivity) Regional climate change Decadal prediction and projections Carbon cycle-climate interactions Climate targets Aerosols, air quality and climate A seemingly diverse list of uncertainties, but all of them depend, to a large extent, on the answer to two questions : How much will the temperature rise for a given radiative forcing? (Climate Sensitivity) How does the atmospheric circulation respond? (Changing Climate Patterns)

6 Climate Sensitivity : conditions many aspects of the climate response at global and regional scales Equilibrium Climate Sensitivity : 1.5 K 4.5 K A concern for mitigation policies e.g, to maintain a warming target of 2 K, nearly twice as much CO2 could be emitted in a low (1.5K) climate sensitivity world as compared to a high (4.5K) sensitivity world. A concern for adaptation studies e.g. many regional changes scale with global T, and life strongly depends on temperature (crops production, physiological limit of human body's adaptability to heat stress, etc) Not a new story... but still a burning issue for climate science! Cattiaux et al. (2013)

7 For regional precipitation, model uncertainty dominates other forms of uncertainty Model uncertainty explains 60 to 90 % of the uncertainty beyond a few decades, with uncertainty in emission scenarios playing a much more minor role (Hawking and Sutton, 2011) IPSL model MPI model

8 The root cause of model uncertainty lies in the interaction between basic atmospheric processes and circulation +4K experiments in CMIP5 Aqua-Planet Models Stevens and Bony, Science, 2013

9 Outline I. Long-standing challenges in climate science II. III. IV. How to gain confidence in model projections? What's missing? Illustration

10 How to increase our confidence in climate change assessments? Improve our confidence in the models' predictive capabilities: Physical basis of climate models (cf J.-Y. Grandpeix's presentation) Ability of models to reproduce the observed climate state and its variations

11 Why has our confidence in projections progressed so slowly? Model development is a slow process: too few model developers much efforts invested in increasing the models' scope rather than the models' quality infinite list of processes to be improved; lack of focus IPCC AR5

12 Why has our confidence in projections progressed so slowly? Model development is a slow process: too few model developers much efforts invested in increasing the models' scope rather than the models' quality infinite list of processes to be improved; lack of focus IPCC AR5 But it is really the primary limiting factor of the rate of progress?

13 An early assessment of long-term climate change : The Charney report (1979) Jule Charney ( )

14 An early assessment of long-term climate change : The Charney report (1979) Available material : - simple climate models (EBM, 1D..) - a few early general circulation models - very few global observations Amazingly prescient in its assessment :... of the effects of increased CO 2 on climate : - timing of doubling of CO 2 concentration - 2 x CO 2 radiative forcing - pattern of surface warming - water vapor and sea-ice feedback estimates - climate sensitivity estimates : range : K ; likely value : 3 K... of key uncertainties : - cloud feedbacks - role of the ocean in carbon and heat uptake - regional precipitation changes - etc

15 The insights of the Charney Report were not an accident They reflect the power of the scientific approach underlying the assessment : In order to assess the climatic effects of increased CO2, we consider first the primary physical processes that influence the climate system as a whole. These processes are best studied in simple models whose physical characteristics may readily be comprehended. The understanding derived from these studies enables one better to assess the performance of the 3D circulation models. Our confidence in our conclusion that a doubling of CO2 will eventually result in significant temperature increases and other climate changes is based on the fact that the results of the radiative-convective and heat-balance model studies can be understood in purely physical terms and are verified by the more complex GCMs. Likely to remain a productive approach in the future

16 How to increase our confidence in climate change assessments? This situation may be expected to improve gradually as greater scientific understanding is acquired and faster computers are built Charney report (NRC, 1979) Might our lack of physical understanding of how the climate system works be the primary limiting factor?

17 Why is physical understanding so fundamental? Unlike in weather prediction, the reliability of model predictive capabilities for long-term climate change can not be established in a straightforward way... but only through indirect routes. Few observational tests are fully discriminating of long-term projections. Determine what questions to be asked of the models and observations. Confidence in climate projections thus remains disproportionately dependent on the development of understanding Bony et al., WCRP Monograph, 2013

18 Why is physical understanding so fundamental? A better identification and understanding of the processes which are critical for climate projections would help: to focus model development on key processes to target relevant observational tests More fundamentally: - models play an important role in climate change assessments - but robust conclusions require more than a consensus by comprehensive models - require the underpinning of physical arguments developed through the use of a hierarchy of models and (whenever possible) observations Might also help address the urgency issue! - advances often come from the rephrasing of old questions in a new way - one that makes long-standing problems finally tractable

19 Outline I. Long-standing challenges in climate science II. III. IV. How to gain confidence in model projections? What's missing? Illustration

20 Main gap in our physical understanding of the climate system? Climate change is not only a global energy budget problem It is also a circulation problem

21 Main gap in our physical understanding of the climate system? How do clouds interact with the atmospheric circulation? For 50 years, efforts to understand clouds and large-scale atmospheric circulations have developed separately... Improving our understanding of the interplay between clouds, circulation and climate will help tackle some of the fundamental puzzles of climate science; Has been identified as one (out of six) Grand Challenge of the World Climate Research Programme (WCRP). Four science questions have the potential to accelerate progress in understanding the Earth system and anticipating global and regional climate changes.

22 Q1: What controls the position, strength and variability of the storm tracks? - Most of the extra-tropical storms develop, organize and decay in spatially localised regions known as storm tracks - Organize clouds and precipitation, and control weather-related climate impacts. How may the storm tracks change as the troposphere becomes warmer and wetter, the stratosphere becomes cooler, and the cryosphere shrinks?

23 Q2: What controls the position, strength and variability of the tropical rain belts? Present-day distribution of tropical precipitation Observed rainfall trends from 1950 to 2000 Shifts of rain belts responsible for severe droughts (e.g. Sahelian drought) How will the rain belts (ITCZ, monsoons..) respond to anthropogenic forcings?

24 Q3: What is the role of convective aggregation in climate? The tendency of clouds to form clusters can modify the climate state. What role may it play as climate becomes warmer?

25 Q4: What is the role of convection in cloud feedbacks? Recent studies suggest that it is a critical question for low-cloud feedbacks Will answers to this question help refine estimates of climate sensitivity?

26 Outline I. Long-standing challenges in climate science II. III. IV. How to gain confidence in model projections? What's missing? Illustration

27 Cloud feedbacks: a long-standing puzzle of climate science... Recognized for more than 30 years to be at the heart of the climate sensitivity problem Many pieces: Cloud types (low, mid, high) Tropical vs extratropical Land vs ocean Cloud height vs water content vs microphysics Cloud-circulation coupling CO 2 vs temperature effects etc...

28 Cloud feedbacks: a long-standing puzzle of climate science... Recognized for more than 30 years to be at the heart of the climate sensitivity problem Many pieces: Cloud types (low, mid, high) Tropical vs extratropical Land vs ocean Cloud height vs water content vs microphysics Cloud-circulation coupling CO 2 vs temperature effects etc...

29 Example of a piece placed in the puzzle: rise of the cloud-top height in cloud feedback LW cloud feedback in CMIP5 models All models predict that in regions of deep convection, cloud-tops rise with global warming, in a way that amplifies the warming (larger temperature contrast between the cloud-top and the surface) How confident can we be in this result? Zelinka et al., J. Climate, 2012, 2013 Vial, Dufresne & Bony, Clim Dyn, 2013

30 Example of a piece placed in the puzzle: rise of the cloud-top height in cloud feedback - Explained through a conceptual representation of the Tropics and physical reasoning (basic thermodynamics and dependence of radiative cooling on water vapor): Fixed anvil temperature hypothesis - Tested with cloud-resolving models - Tested with GCMs used in various configurations CRM simulation - Verified with observations High confidence in this robust component of the model cloud feedbacks Hartmann and Larson, GRL, 2002 Eitzen et al., J. Climate, 2009 Kuang and Hartmann, J. Climate, 2010 Zelinka and Hartmann, JGR, 2010

31 Cloud feedbacks: a long-standing puzzle of climate science... Recognized for more than 30 years to be at the heart of the climate sensitivity problem Many pieces: Cloud types (low, mid, high) Tropical vs extratropical Land vs ocean Cloud height vs water content vs microphysics Cloud-circulation coupling CO 2 vs temperature effects etc...

32 Spread of climate sensitivity estimates among GCMs Primarily arises from the uncertain response of marine low-level clouds to warming. Shallow cumulus clouds in the trades Photo B. Stevens Bony et al., Clim. Dyn., 2004 Bony & Dufresne, GRL, 2005 Webb et al., Clim. Dyn., 2006 IPCC AR4 & AR5

33 Understanding the low-level cloud feedback Much progress recently in understanding the processes that primarily control this feedback These ideas have motivated analyses of GCMs, e.g. Sherwood et al. (Nature, 2014) which identifies the strength of water vapor mixing within the first few km of the atmosphere as a key constraint for climate sensitivity. Motivates many modeling and observational studies currently... Hope for an improved assessment in the future? Rieck, Nuijens & Stevens, JAS, 2012 Brient and Bony, Clim Dyn 2013 Zhang et al., JAMES, 2013 Webb and Lock, Clim Dyn 2013 Qu et al., Clim. Dyn Bretherton and Blossey, JAMES, 2014 Tomassini, Voigt & Stevens, QJRMS, 2014 Sherwood, Bony & Dufresne, Nature, 2014

34 Conclusion Our ability to provide more reliable insights about climate changes primarily depends on our ability to physically understand how the climate system works. Accelerating progress in climate change assessments requires an approach that: - tackles fundamental puzzles of climate science (including old ones) - emphasizes physical concepts and testable ideas around which scientific activity can organize (e.g. test different story lines for each of the four questions) - improves our understanding of the interplay between clouds, circulation and climate through the development and testing of hypotheses that link changes in regional patterns, extremes, climate sensitivity etc in a self-consistent way. - integrates theory, hierarchy of models, observations, process understanding (different from the IPCC approach)

35 Related papers Bony S., B. Stevens, D. M. W. Frierson, C. Jakob, M. Kageyama, R. Pincus, T. G. Shepherd, S. C. Sherwood, A. P. Siebesma, A. H. Sobel, M. Watanabe, and M. J. Webb: Clouds, Circulation and Climate Sensitivity. Nature Geoscience, submitted Bony S., B. Stevens, I. Held, J. Mitchell, J.-L. Dufresne, K. Emanuel, P. Friedlingstein, S. Griffies and C. Senior, 2013: Carbon Dioxide and Climate : Perspectives on a Scientific Assessment. Monograph on Climate Science for Serving Society: Research, Modeling and Prediction Priorities, Springer, G.R. Asrar and J.W. Hurrel (eds.), pp , DOI / _14, Stevens, B. and S. Bony, 2013: What are climate models missing? Science, 340 (6136), , DOI: /science Stevens, B. and S. Bony, 2013: Water in the atmosphere. Physics Today, June 2013,

36 Thank You

37 Our physical understanding of the climate system relies on a spectrum of models and theories Model simplicity (relative to system) System complexity Our confidence in modelling results depends on our ability : to connect robust behavior across a spectrum of relevant models to connect this behavior both to basic physical principles and to observations Bony et al., WCRP Monograph, 2013