Climate change and climate feedbacks Christopher S Bretherton

Climate change and climate feedbacks Christopher S. Bretherton Department of Atmospheric Sciences University of Washington Seattle, WA USA

1864 -1883 range 1988 -2007 range Weather = climate +

Global surface temperature - a good climate metric • Reliably derivable from observations for the last century. • Averages out regional variations • Unforced interannual variability ≈ ± 0. 1ºC • Well correlated with multidecadal regional variations. IPCC 2007

The simple core of global warming Absorbed sunlight heats Earth Warm surfaces radiate more energy. Earth warms until radiated infrared energy balances absorbed sunlight • CO 2, a greenhouse gas, intercepts Earth’s infrared radiation, reducing surface cooling efficiency. • With more CO 2, surface must warm to bring infrared loss into balance with absorbed sunlight. • Doubled CO 2 with no other surface or atmospheric composition changes would raise surface temperature 1. 2°C. • Already foreseen by Arrhenius (1895) …global climate models predict more warming due to ‘climate feedbacks’. CO 2(ppmv) • • • IPCC 2007

The climate sensitivity problem Models match 20 C climate, but vary on 21 C warming IPCC 2007 2. 5 -4. 3°C 2 x. CO 2 3 x. CO 2 …though every climate model suggests 21 C climate change will probably dwarf the significant changes we’ve seen already.

Climate sensitivity Manabe and Stouffer 1994 TCR 2. 3°C ECS 3. 5°C Equilibrium climate sensitivity ECS = global surface temperature change due to CO 2 doubling after climate equilibrates (takes over 100 years in practice). Transient climate response TCR = global surface temperature change at the time of CO 2 doubling when CO 2 is increased 1%/year. – TCR is relevant to climate change of the next 50 years. – TCR is smaller than ECS due to ocean thermal inertia.

Intermodel variability of climate sensitivity IPCC 2007 • ECS of global climate models varies from 2 -4. 5°C, with a median of 3°C; this range is little-reduced since 1990. • TCR of global climate models is 1. 2 -2. 5°C.

Global-mean precipitation also increases by 1. 5% / °C (but not everywhere!) Climate models agree less well on patterns of rainfall change than those of temperature change, but all climate models predict that the more the climate warms, the more rainfall patterns will change. IPCC 2007 Figure SPM. 7

Climate feedbacks CO 2 atm surf T f f 1: Water vapor + lapse rate: warmer more water vapor f 2: Snow/sea ice albedo: warmer less reflective snow/ice f 3: Clouds: warmer diverse changes. + long-term feedbacks: icesheets vegetation change carbon cycle (ocean + land)

Climate feedback strengths in climate models Clouds Albedo 0. 16± 0. 05 0. 06± 0. 02 Feedback factor f Water vapor/ lapse rate 0. 25± 0. 02 0. 5 Total 0. 48± 0. 05 0. 25 0 IPCC 2007

Water vapor/lapse rate feedback Outward Infrared radiation change 20 km warmer climate Height • Air 1°C warmer can hold 7% more water vapor. • Physical processes controlling atmospheric water vapor (evaporation, cloud formation/rainout, vertical motions) tend to maintain constant relative humidity on global scales. Thus atmospheric water vapor increases by ~7%/°C of global warming. • Water vapor is a greenhouse gas feedback f 1 ≈ 30%. • Warming is height-dependent, but overall ‘lapse rate’ + water vapor feedback almost unaltered. sfc cooler climate Temperature T WV

Observed water vapor feedbacks Satellite observations show mean tropical surface temperature is highly correlated with atmospheric column water vapor. Allan et al. 2003

Snow/ice albedo feedback • In a warmer climate, reflective snow and sea ice decrease, so Earth absorbs more sunlight. • A weak positive feedback in the current climate, but important for amplifying warming at high latitudes. • Much larger if ice sheet changes are considered responsible for much of ice age temperature fluctuations. • The seasonal cycle of albedo induced by winter snow cover is a good observational test of this feedback in climate models. IPCC 2007 from Hall and Qu 2006

Clouds - the most uncertain climate feedback Clouds are hard to represent in climate models, because: • Often thin, short-lived, • Often produced by small-scale turbulence (e. g. cumuli), • Complex interactions of water and ice particles - High, thin clouds warm (greenhouse effect dominates), - Low clouds cool (shading effect dominates).

Cloud feedbacks in two climate models With doubled CO 2, lowlatitude low-lying clouds systematically: • decreased in GFDL AM 2 (positive albedo feedback) and ECS = 4. 5 K • increased in NCAR CAM 2 (negative albedo feedback) and ECS = 1. 5 K - No good observational test of cloud feedbacks on mean climate!

Some questions raised • Does the significant global warming of the last century observationally constrain climate sensitivity? • Is a CO 2 -based view of global warming accurate? • Are climate sensitivity uncertainties as large as the climate differences between emissions scenarios? • Should climate sensitivity uncertainties affect current energy policy decisions?

Present climate forcing - it’s not just CO 2 Human forcings - greenhouse gases (+) - aerosols (-) - land use, waste heat Natural forcings - volcanoes - solar variability …& on long timescales… - Milankovitch cycles - drifting continents IPCC 2007 Aerosol effects (esp. from cloud brightening) are the biggest uncertainty. Radiative forcing of CO 2 doubling = 4 W/m 2 = 2% outgoing radiation reduction

Forcing evolution • Until ~1970, aerosols cancelled 1/3 -2/3 of greenhouse gases. • Since 1990, aerosols have leveled as US/Europe pollution controls balance Asian industrialization. • In 21 st century, trends should be CO 2 -dominated. IPCC 2007 MIROC models - others vary somewhat

Climate models reproduce 20 th century climate well IPCC 2007 break in warming: More aerosols? More eruptions? …but do so through different cancellations of forcings.

Observational estimates of climate sensitivity We’ve had about 1°C of warming, and a 40% CO 2 increase since 1850. Can’t we use this to estimate climate sensitivity? - Issue: warming counteracted by uncertain aerosol forcing so possible range is large (ECS > 1. 5°C). IPCC 2007 Glacial-interglacial climate change suggests ECS < 5°C - but climate sensitivity may depend on climate! All available obs+models are consistent with ECS of 2 -4°C

Climate projection IPCC 2007 Emissions uncertainty Carbon cycle feedback uncertainty Climate feedback uncertainty

21 st century emissions scenarios Uncertainties reflect carbon cycle feedbacks 1900 2000 2100

By 2070, emission scenario dominates climate change uncertainty B 1 A 1 B A 2 …so climate models are ‘skillful’ for energy policy choices.

Fossil CO 2 - a cumulative problem Geological burial of CO 2 takes 1000 s of years! Climate sensitivity affects model response to this.

Melting of Greenland Ice Cap? 4 x. CO 2 Hadley Centre simulations (Ridley et al. 2005) • Complete melting = 7 m sea level rise. • ~50% melting happened 125, 000 years ago in the last interglacial with Greenland only 1 -3°C warmer than now.

Conclusions • Positive feedbacks in the atmosphere (water vapor, clouds, snow/ice albedo) double the surface warming due to greenhouse gases like CO 2. • Climate models agree well on the water vapor feedbacks but differ somewhat on the cloud feedback (and carbon cycle feedbacks). • The ± 25% uncertainty this introduces into global temperature projections is less than the differences between reasonable emissions scenarios after 2070. • Fossil CO 2 is a cumulative problem. Failure to reduce global CO 2 emission within the next 20 -40 years will lead to large climate change for hundreds of years. • The countries of the world (including the USA and China) need to change rapidly to energy technologies that do not emit CO 2, such as carbon sequestration from coal, wind and solar energy, and nuclear energy.