The scientific basis for climate change projections: History, Status, Unsolved problems Isaac Held, Princeton, Feb 2008 Katrina-like storm spontaneously generated in atmospheric model Regions projected to have more (green) and less (brown) precipitation In 21st century The Keeling curve poorly understood well understood
(C. Keeling)
Ice cores + direct measurements provide a beautiful record of the history of atmospheric carbon dioxide www globalwarmingart.com. Jouzel, Lorius et al-- Vostok late 70 s-80 s Etheridge, et al -- Law Dome 1990 s
Carbon dioxide methane Nitrous oxide CFCs Methane and Nitrous Oxide: (1976) Wang, W.-C., et al CFCs: (1975) V. Ramanathan (1974) Molina-Rowland: catalytic destruction of ozone by chlorine; (1985) Farman et al: ozone hole
http://www.esrl.noaa.gov/gmd/aggi/ Carbon dioxide methane Nitrous oxide CFCs other
www.globalwarmingart.com
Was the 20th century warming 1) primarily forced by increasing greenhouse gases? or, 2) primarily forced by something else? or, 3) primarily an internal fluctuation of the climate? Claim: Our climate theories STRONGLY support 1) A central problem for the IPCC has been to evaluate this claim and communicate our level of confidence appropriately
Global ocean heat content Energy is going into ocean => More energy is entering the atmosphere from space than is going out www.globalwarmingart.com 1955 1980 2005 Almost all parts of the Earth s surface have warmed over the past 100 years IPCC 4th Assessment Report.
Theories/models required to discuss future changes in climate An atmospheric model under development at GFDL one month of simulated near surface flow (bright => eastward ; dark => westward -- Northern hemisphere summer)
Ocean model currently in use Simulation of Surface Currents Ocean model under development -- Hallberg and Gnanadesikan, JPO 2004
Climate models spontaneously produce variability on all time scales Including the year-to-year variability known as El-Nino
temperature anomalies (10yr running means) averaged around latitude circles over 20th century: -- Observations (big) -- 5 realizations of a model simulation (small) A lot of variability on decadal time scales; especially in high northern latitudes; but not nearly enough to explain trend over century Source: Delworth and Knutson, Science (2000).
Climate sensitivity = temperature change per unit change in radiative forcing Nature has conspired to make the problem of constraining climate sensitivity very difficult: 1) The uncertainty in aerosol forcing over the 20th century limits our ability to use 20th century warming to determine sensitivity empirically 2) The difficulty in simulating clouds prevents us from developing a satisfying quantitative theory of climate sensitivity
GFDL s CM2.1 with well-mixed greenhouse gases only Global mean temperature change Observations (GISS) It is likely that increases in greenhouse gas concentrations alone would have caused more warming than observed because volcanic and anthropogenic aerosols have offset some warming that would otherwise have taken place. (AR4 WG1 SPM).
(GFDL CM2.1 -- Includes estimates of volcanic and anthropogenic aerosols, as well as estimates of variations in solar irradiance) Models can produce very good fits by including aerosol effects, but models with stronger aerosol forcing and higher climate sensitivity are also viable (and vice-versa)
Model simulated clouds (actually infrared radiation escaping to space)
Clouds (especially in the tropics) are influenced by small scales in the atmospheric circulation 100kms simulation of a 100km x 100km area of the tropics
Courtesy of Brian Soden Change in Low Cloud Amount (%/K) GFDL and NCAR/CAM models
Various estimates of climate sensitivity (global mean warming for doubling carbon dioxide) (from IPCC AR4 WG1)
IPCC AR4 WG1 Summary for Policymakers
Need to combine many sources of information to constrain sensitivity: -- sensitivities obtained by world s climate models -- 20th century warming (using estimates of aerosol cooling) -- paleoclimates (especially last glacial maximum ~20K years ago) -- response to volcanoes
Global mean cooling due to Pinatubo volcanic eruption Observations with El Nino removed Range of Model Simulations GFDL CM2.1 Courtesy of G Stenchikov Relaxation time after abrupt cooling contains information on climate sensitivity
http://data.giss.nasa.gov/gistemp/
??
3 aspects of the global response of the hydrological cycle to warming the dry get drier and the wet get wetter the semi-arid subtropics expand polewards the tropical rain belts move towards the hemisphere that warms the fastest
Total Column Water Vapor Anomalies (1987-2004) We have high confidence in the model projections of increased water vapor. Held and Soden J.Clim. 2006
Precipitation and evaporation Aqua_planet climate model (no seasons, no land surface) Instantaneous precip (lat,lon) Time means
Precipitation and evaporation Aqua_planet climate model (no seasons, no land surface) Instantaneous precip (lat,lon) Precipitation moves north and south with the seasons Time means
dry wet dry wet wet
dry wet dry wet wet
Average over year Dec-Jan-Feb June-Jul-Aug NDVI vegetation index Anomaly in June 2006 (J. Allan NASA Earth Observatory)
.. 2006 Drought Headlines
Latitude of maximum surface westerlies in Southern Hemisphere Ensemble of climate model runs Roaring 40 s are moving polewards observations
Was part (most?) of the poleward displacement of the atmospheric circulation in the Southern Hemisphere over the last few decades due to the ozone hole?
Latitude of maximum surface westerlies in Southern Hemisphere Ensemble of climate model runs (about half of shift due to ozone hole) Roaring 40 s are moving polewards observations
Zonal wind as a function of latitude and height (pressure) -- averaged around latitude circles and over time Dec-Jan-Feb Jun-Jul-Aug simulation observation Models do a good job of simulating atmospheric circulation, => confidence in their predictions of poleward expansion
Poleward shift of circulation in Northern Hemisphere (much larger than simulated by models!) Reichler et al, Nature Geosciences, 2008
The Sahel drought of the 1970 s and 1980 s was NOT primarily due to desertification African drought It was, rather, the response to changing ocean temperatures! models can simulate the observed long term variations in Sahel rainfall if given the observed ocean surface temperatures
Response of GFDL/CM2.1 to range of SRES scenarios (B1/A1B/A2) Catastrophic drought In Sahel
GFDL/CM2.1 is an outlier 129 different models: (courtesy of Matthew Collins, Hadley Center) histogram of % Sahel rainfall response to 2xCO2
Regional model forced at boundaries and on large scales in interior by relaxation to observations, and at surface by observed ocean temperatures Knutson, et al, BAMS, 2007
Observations - 2000 Simulation - 2000 Knutson, et al, BAMS, 2007
Knutson, et al, BAMS, 2007
Number of Atlantic hurricanes each year from 1980-2005 -- observed (black) and simulated (red) by downscaling model Knutson, et al, BAMS, 2007
Hurricanes and global warming: Ongoing work (at GFDL and elsewhere) is tentatively suggesting that global warming will bring -- more intense hurricanes -- with more rainfall -- but not necessarily an increase in the number of Atlantic storms (perhaps even a decrease in number) The recent increase in the number of Atlantic storms is probably related to changes in circulation associated with the rapid warming of the tropical North Atlantic with respect to the rest of the tropical oceans. (progress on this problem is relatively rapid, so stay tuned)
Global ocean heat content Water vapor 1955 1980 2005 -- there are some very important aspects of this problem about which we are still very uncertain -- and there are other very important aspects about which we are very confident IPCC process helps governments and the public distinguish between the two Sahel rain Latitude of Roaring 40 s Atlantic hurricanes