CERES Measurements

1
(No Transcript)
2
CERES Measurements

• CERES broadband measurements
• total, shortwave and longwave window
• Historical Data ERBS, ERBE
• CERES ERBE-Like
• Multiple Scanners programmable
• crosstrack, along-track and rotating azimuth
• Collocated with Imagers
• VIRS and MODIS for clouds, GOES for diurnal

3
How do more accurate CERES TOA fluxes differ from
ERBE?

• New ADM Impacts
• radiance to flux
• albedo (solar zenith)

New Geo 3-hourly sampling impact
- Zonal mean changes vary by 3 to 8 Wm-2 (CO2
doubling is 4 Wm-2) - Regional cloud radiative
forcing changes of 5 to 10 Wm-2. - Impacts
equator to pole heat transport, climate model
validation, cloud feedback, clear-sky fluxes,
surface/atmosphere radiative heating - 4 years
of global gridded Terra fluxes Fall, 2004.
IPCC climate model testing
4
CERES new Terra Angular Dependence Models (ADMs)
Results using TRMM ADMs and theory for snow/ice
(white is error gt 10 Wm-2)
Results for new Terra ADMs improves snow, ice,
and land
Terra ADM SW error 0.5 Wm-2 global bias 1.0 Wm-2
regional ?
Direct Integration Test Crosstrack Scanner
fluxes using ADMs minus Rotating Azimuth Scanner
Fluxes using hemispheric radiances (uses no ADMs
but only valid on 1000km monthly means)
5
New CERES Downward LW Surface Fluxes All-sky
Clear and Cloudy Errors
For BSRN sites equator to pole Bias lt 5
Wm-2, instantaneous sigma ranges from 15 to 25
Wm-2 Total of 60,000 comparisons Now
sufficient data to quantify by climate
region, cloud type, viewing angle
ARM central facility 700 satellite
overpasses within 1 minute day and night
CERES SSF Single FOV Downward LW Flux (Wm-2)
Bias lt 1 Wm-2, Sigma 15 Wm-2
ARM Ground Measured Downward LW Flux (Wm-2)
CERES Terra SSF data product over 3 years of
global Edition 2 available
6
Surface Flux Accuracy
7
New Ways to Compare Clouds, Radiation and
Models a cloud systems approach
50 km ECWMF Model cloud albedo too high by
0.1, SWCRF error of 30 Wm-2 1km Cloud
Resolving Model (CRM) improves a factor of
2 Examples of methods for A-train
super-parameterization clouds Better
cloud/dynamics cause and effect Observation
accuracy of CERES ensemble 1 Wm-2
Combines new CERES ADM TOA fluxes and ECMWF 4-D
Atmospheric State
8
New ERBS Decadal Variability Results
First Comparison of Radiation and Ocean Heat
Storage ERBS 60S - 60N (87 of earth area) Net
flux anomaly (annual mean error sigma of 0.5
Wm-2) Ocean heat storage flux from Willis et
al., 2004 submitted to JGR (annual mean error
sigma of 1.2 Wm-2)
- ERBS net heat flux and altimeter heat storage
flux agree within error - Ocean heat data shows
larger variability than radiation data, but
results depend on method in-situ limited to
700m depth altimeter responding to ocean thermal
changes at all depths, but needs corrections
for ocean currents - Ocean data more accurate
at decadal time scale, radiation at interannual -
ERBS data for late 1999 through present
available, but not processed (no funding)
9
The Iris ERBS Decadal Data Reject
Warmer Sea Surface Temperature
ERBS satellite tropical mean radiation Iris
simulated radiation anomalies No consistency
Thermal Emission Increase (Cooling)
Solar Increase (Warming)
Solar
Iris More efficient precipitation decreases upper
cloud anvil area
ERBS does not support the Iris relationship of
anvil and temperature
thermal
Lin et al. J. Climate, 2004
10
Climate Model Sensitivity UncertaintyA new
approach for IPCC

• Current Major Problem
• we cannot relate climate model errors versus
  observations to climate model error in predicted
  sensitivity or future climate change!
• instead we run N climate models and just use the
  range not rigorous
• uncertain how to weight different metrics of
  climate model error vs. observations
• IPCC Climate Sensitivity Workshop April 19-22,
  Exeter, U.K.
• new climate model Physics Perturbation Ensembles
  (PPEs) being runby UKMO and climateprediction.net
  like SETI_at_Home for climate
• run 1000s of climate model simulations, each
  varying uncertain physical parameterization
  coefficients (cloud, ocean, ice, etc) within
  range of uncertainty
• How relate climate model vs observation metrics
  to model sensitivity error?
• Use 1000s of Planets each similar to but
  not Earth, each different physics.
• we proposed to choose at random pairs of PPE
  climate simulations use one as a real Planet
  N, and the other as the model of Planet N.
• determine the model minus obs metrics between
  the two models and the climate sensitivity
  difference for CO2 doubling (global temp, U.S.
  precip, etc)
• use a neural net or other statistical tool to
  relate 500,000 such climate modelpairs with
  order 50 to 200 climate metrics to predict model
  sensitivity difference
• test against other independent climate models
  (including IPCC runs) and if successful, use
  neural net to relate actual climate model errors
  vs obs to each climate models sensitivity
  uncertainty.

11
Backup
12
Albedo during 2000 2003