cloud-climate feedbacks: what we think we know and why we think we know it

Download Cloud-climate feedbacks: what we think we know and why we think we know it

Post on 12-Jan-2016




2 download

Embed Size (px)


Cloud-climate feedbacks: what we think we know and why we think we know it. David Mansbach 14 April 2006. T 2 4. T 2 4. T 2


  • Cloud-climate feedbacks:what we think we know and why we think we know itDavid Mansbach14 April 2006T1
  • Clouds, in general and todayGreenhouse effect, albedo effect, and satellite measurementsCloud changes in a perturbed climateVery wide range of scales and processes involvedModelingParameterize clouds and take GHG-forced runs as a predictionsParameterizations inspired by physical principles but lead to errors compared to validationObservationsUnderstand observed cloud changes to variability of specific conditionsMerge observed cloud tendencies with modeled large-scale changes or conceptually suggested changesTrade-offs

  • Longwave forcing: cloud greenhouse effect and cloud albedo effectAs a cloud gets thicker, it acts like a blackbody: absorbing at all wavelengths and emitting according to T4Higher clouds -- in cold upper atmosphere -- emit less IR/longwave radiation to space, and keep more energy in the Earth system

    Thickness, water/ice distribution, sun angle affect how much cloud reflects sunlight (its albedo)T1

  • Tropical and extratropical cloudsBony et al. 2004/Emanuel 1994Bony et al. 2006/Cotton 1990Area of ascent is small; area of decent is largeCloud cover is greater in areas of descent, and lower in altitudeFrontal systems form a variety of cloudsNature strength of storms determines clouds

  • from Randall, 2004So now we just need to decide how clouds will change in the future... Define cloud radiative forcing at any point as the difference in outgoing radiation with a cloud present minus that with clear sky Satellite data such as ERBE show net effect of cloud forcing is dominated by SW effect; CRF ~ -20 W m-2

  • Changing clouds, changing cloud radiative forcingCloud processes operate on some small scales -- think of a thunderstorm in the distance or wispy clouds overheadMore condensed water generally means more optically thick clouds -- ie, more absorption and emission of longwave -- and affects refletionShortwave reflectivity also depends on number of droplets -- sunlight will be reflected more if there are many small droplets (also leads to interplay with aerosols!)Overall effects of clouds depend on myriad processes -- ie, thermodynamic, microphysical, optical, convective, dynamicMany effects can be hypothesizedie CO2x2 -> more evaporation, -> more cloud liquid water -> more SW reflectivity -> negative feedback (ie Somerville and Remer 1984)cf: CO2x2 -> warmer SST -> breakup of SC, greater areas of deep convection -> positive feedbackfrom NASA

  • Thats why we have modelsto look at global CRF changes, try using a global modelalthough scales of individual clouds might be ~100m or ~1 km, climate model resolution ~100kmparameterizations link large-scale climate to cloud properties based on observations and theoryalso conserve important properties, such as moisture, energy, etc.easier said than done -- larger-scale conditions do not necessarily fully determine actual cloud fields; radiative impacts and feedbacks could be considerableGCM-simulated current cloud climatology is often obviously unrealisticSchmidt et al 2006CTP

  • Different cutting-edge models also dont agree preciselycloudsBony et al. 2006water vaporaerosolslapse rate w/water vaporlapse ratetotalAlthough spread is large, modern models predict a positive cloud feedback to global warming, meaning that future cloud forcing is less negative (clouds will not cool the Earth system as much as today)

  • Concntrating on different models cloud response to forcingSCRF SCRF LCRF LCRFSCRF SCRF LCRF LCRFWilliams et al. 2006

  • Norris + Iacobellis (2 slides) - compositing methods and diff regimesfinal estimates in most likely scenariosUsing observations to inform discussion of clouds in future climate

    Using years of satellite and reanalysis data, plot average cloud properties as functions of temperature advection and vertical velocityThese data are for conditions of SST and lower-tropospheric static stability in normal/moderate conditionsThis allows for a sort of empirical partial derivative of various cloud propertiesNorris and Iacobellis 2005

  • even if GCM clouds are unrealistic, dynamical predictions can be combined with CRF observationsGeneral inferences from past polar amplification and known storm dynamics, as well as a GCM (Dai et al. 2001), suggest storm track weakening (less extreme vertical velocity) along with warmer SST and little change in vertical stability -> less cloudiness, thinner cloudsmodeled temperature changes would lead to less broad marine stratocumulus and less marine fog -> less SW CRF, positive forcing

    for surface temperature advection for mid-troposphere vertical velocity Norris and Iacobellis 2005for stronger storm track

  • Other observations relevant to midlatitude CRF changesTselioudis and Rossow, 2006

  • Implications of observed CRF tendenciesa GCM (Carnell and Senior 1998) predicts fewer weak and moderate storms, but more strong stormsimplied additional SW cooling is 0 to 3.5 W m-2 in different areas (fewer clouds, but more reflective)implied additional LW warming is 0.1 to 2.2 W m-2 (fewer clouds, but higher)overall increase in strength dominates, leads to global cloud COOLING of ~1 W m-2analysis of cloud response to circulation and temperature changes is consistent with other study, but choice of modeled circulation changes are differentif these midlatitude changes were factored into Norris & Iacobelliss figures, total CRF would still be positive, but less so, because of thermodynamic response

  • TradeoffsGlobal feedbacks of clouds unknown; depends on myriad processes on various scalesPhysical mechanisms can be hypothesized to support SW and LW feedbacks of any signThe latest round of models predict positive cloud feedback; some observational analysis shows consistent physical reasoning for thisModel spread is large; model clouds still have many errorsHow predictable are clouds really?


View more >