1Cloud physics and cloud-climate feedbacks:Connecting this course to current researchRichard SomervilleSIO 217ANovember 14, 2006Start from two sections of Curry & Webster:Parameterization of Cloud Microphysical Processes(Section 8.6), pages 241 - 244. Understand the ideasbehind the equations on page 242 and note theremarks on page 244.Cloud-radiation Feedback (Section 13.4), especially thelast 2 paragraphs of Section 13.4.1 on pages 368, 369,and the last paragraph of Section 13.4 on page 374.What do we mean by “parameterization”?Some physical processes are too poorly understood,and/or they occur on too small space and time scales,so we cannot adequately represent them in globalnumerical climate models.For example, clouds and cloud-radiation interactionsare important to climate and to modeling climatechange.We don’t know how to include these processescorrectly, but we cannot afford to omit them.A working definition of “parameterization”:A parameterization is an algorithm or rule for obtainingthe statistical effect, of an ensemble of small-scaleprocesses (e. g., cloud processes), on the large-scaleprognostic fields computed explicitly in a model (e. g.,wind, pressure, temperature, humidity).In general, the parameterization must be explicit, in thesense that the statistical effect can be computed as afunction of the large-scale variables themselves.Example: cloud fraction might be a function of humidity.First, some background.We define “climate sensitivity” as the equilibriumchange in global average surface atmospherictemperature in response to doubling the presentatmospheric concentration of carbon dioxide. (Thereare many alternative definitions).Many (but not all) reputable models have sensitivitiesranging from about 1.5 deg C to 4.5 deg C. This rangeis an old result. It has not changed in nearly 30 years.Why?Cloud effects dominate climate sensitivity.In a typical comparison of coupled models, some ofwhich have the same atmosphere, and some of whichhave the same ocean, the clear result is that it is theatmospheric model which largely determines thesensitivity, and in particular, it is clouds.This is another confirmation of something known for along time. Changing the cloud-radiation scheme in onemodel replicates the sensitivity spread of many models.The problem is we don’t know which scheme is best.2Cloud algorithms require comprehensive approaches.Cloud radiative processes are determined BOTH bycloud macrophysics (cloud size, altitude, thickness,etc.) AND by cloud microphysics (water content,phase, particle shapes and size distributions, etc.).The old procedure of predicting clouds as a simplefunction of relative humidity, and then assigningradiative properties arbitrarily, is not good enough.Zeroth-order cloud effects are still unclear.Models don’t agree on even the simplest aspects ofcloud changes as climate warms. Some predict cloudamount increases; some predict it decreases.If in-situ and satellite interpretations can show anobserved large-scale secular trend in cloud amountin recent decades, our models must be able tosimulate that.To be believable, climate models must pass tests.It’s unacceptable that models with differentsensitivities all manage to reproduce the global mean20th-century temperature evolution by usingdramatically different assumptions on forcing.The temptation is to treat the 20th-centuryrecord of solar variability, volcanism, aerosols andgreenhouse gas changes as parameters that can betuned so the model produces the observed recordof surface temperature as a function of time.Climate sensitivity cannot be only 1 global numberThe annual cycle may be masked in average results orthrown away in perpetual-month simulations. In somemodels, cloud feedback in the longwave (terrestrial)is about the same in all seasons, but the shortwave(solar) cloud feedback actually changes sign over theannual cycle, due to cloud amount changes.The concept of cloud feedback should also includespatial as well as temporal variability.It may be a distraction to concentrate on global effects.We already know that aerosol effects are highlylocal. Perhaps cloud effects are largely local too.Even if cloud-radiation effects are not largeglobally, and we don’t know yet if this is true, theymay be very important locally.All this uncertainty about how clouds should bemodeled has motivated intense research, and onetheme of this research is to observe real clouds.3IPCC Synthesis ReportSimulated annual global mean surface temperaturesThere are two elements in this research (which is jointwork by Sam Iacobellis and Richard Somerville):1. We build parameterizations incorporating detailedand comprehensive cloud microphysics. For example,radiative properties of clouds depend on cloud watercontent, phase (liquid or ice or both), particle sizedistribution, ice crystal habit (shape), all of whichcan be treated explicitly as functions of knownvariables.2. We test these parameterizations against fieldobservations by using a special theoretical tool, thesingle-column model.Input:• Initial profiles of T and q• Advective fluxes of heat and moisture• Vertical velocityOutput:Large-Scale Variables• Temperature, T(t,z)• Humidity, q(t,z)• Cloud water/ice, qc(t,z)Diagnostics• Precipitation• Radiative Fluxes• Cloud propertiesSINGLE-COLUMN MODELSouthern GreatPlainsNorth Slope of AlaskaNorth Slope of AlaskaTropical Western PacificTropical Western Pacific“The Good, the Bad, and the Ugly”What is the resultof comparing cloudparameterizationsagainst detailed &very high-qualityobservational data?4150200250300350153 163 173 183 193 203 213 223 233 243(A) DOWNWELLING SURFACE SHORTWAVESCM (284)OBS (267)DWSO-SCMJULIAN DAY (2000)WATTS M-2-140-120-100-80-60-40-200153 163 173 183 193 203 213 223 233 243(B) TOA SHORTWAVE CLOUD FORCINGSCM (-32)OBS (-38)DWSO-SCMJULIAN DAY (2000)WATTS M-2220240260280300320153 163 173 183 193 203 213 223 233 243(C) OUTGOING LONGWAVE RADIATIONSCM (272)OBS (270)DWSO-SCMJULIAN DAY (2000)WATTS M-2010203040506070153 163 173 183 193 203 213 223 233 243(D) TOA LONGWAVE CLOUD FORCINGSCM (26)OBS (29)DWSO-SCMJULIAN DAY (2000)WATTS M-20.00.20.40.60.81.0153 163 173 183 193 203 213 223 233 243(E) CLOUD FRACTIONSCM (0.42)GOES (0.40)MMCR (0.49)DWSO-SCMJULIAN DAY (2000)FRACTION0510152025153 163 173 183 193 203 213 223 233 243(F) CLOUD OPTICAL THICKNESSSCM (4.6)OBS (6.3)DWSO-SCMJULIAN DAY (2000)OPTICAL THICKNESS5Cloud Thickness(Lowest 2km Only)Mean = 244mMean =
View Full Document
Unlocking...