Transcript Title of Presentation

Use of satellite water vapour data sets for
climate model evaluation & development
Mark Ringer & Viju John, Met Office Hadley Centre, Exeter, UK
GEWEX/ESA DUE GlobVapour Workshop, Frascati, 8-10 March 2011
© Crown copyright Met Office
Contents
• Background & motivation
• “Traditional” evaluation studies – total column
moisture in the new Hadley Centre climate
model, HadGEM2
• The forward modelling approach – simulation of
HIRS/AMSU radiances
• Climate model development & improvement
• Summary & conclusions
© Crown copyright Met Office
Why do we use satellite data?
• Evaluate the physical processes most relevant
to reducing uncertainty in climate predictions
• Inform & prioritise key areas for developing and
improving climate models
• Constrain climate change predictions – or at
least try and determine if this is possible
• Detection & attribution of observed variations to
natural and anthropogenic forcings
• Initialisation of models used for seasonal-todecadal prediction
© Crown copyright Met Office
Three key questions
• Can we use present-day observations to
constrain climate feedbacks?
• Can we use observations to improve the
processes which contribute most to the
range of uncertainty in climate
projections?
• Will future observations be suitable for
evaluating our climate projections for the
coming decades?
© Crown copyright Met Office
How do we use satellite data?
• “Traditional” method – compare high-level products (e.g. radiative fluxes,
total column moisture) with their model equivalents
• “Model-to-satellite” approach – simulate what is actually measured (e.g. IR
or microwave radiances, radar reflectivities, etc)
• Development of new, process-based, evaluation techniques for using the
data and combining with other information such as reanalyses – e.g.
compositing in terms of dynamics, clustering
• Apply similar techniques to analysis of climate change simulations and
feedbacks
• For “fast” processes such as clouds & precipitation we also use
comparisons with the global NWP model (e.g. CloudSat)
• In combination with in situ data from the global observation network, data
© Crown copyright Met Office
from aircraft and other field campaigns
Radiative feedbacks in the IPCC
AR4 models
Water Vapour Clouds
© Crown copyright Met Office
Albedo
Lapse Rate WV + LR
ALL
(Bony et al. 2006)
Title
…and our desire to
improve these processes
in climate models.
Images courtesy of the USGS (John M. Evans,
USGS, Colorado District) and lasp.colorado.edu
© Crown copyright Met Office
We clearly need to
remember the bigger
picture…
Evaluation of TCWV in HadGEM2
© Crown copyright Met Office
Annual Mean TCWV
July Mean TCWV
El Niño: Anomalies in January 1998
La Niña: Anomalies in January 2000
Equatorial Anomalies: 1989-2005
Multiple data sets: TCWV & Clear-Sky OLR
The forward modelling approach
© Crown copyright Met Office
• Allows direct comparison with
measured radiances, etc
• Avoids ambiguities associated
with comparing to retrieved
quantities
• Example shows HIRS
Channel 12 in previous version
of Hadley Centre model
• RTTOV is now part of the
COSP simulator
• For further details see:
http://cfmip.metoffice.com/COSP.html
Evaluation of HadGEM2 using AMSU-B
January
July
Model
OBS
• Direct simulation of AMSU-B Channel 3
• Converted to UTH in model and observations
• Further sub-sampled into times of large-scale descent
Long-term variability: Tropical mean
anomalies in large-scale descending regions
• Model reproduces observed variability very well
• Indicates no significant trend in UTH
Evaluation of the global forecast model
IR
Model
OBS
WV
Climate model development
© Crown copyright Met Office
Development of HadGEM3: Relative Humidity
New
Old-Obs
New-Old
New-Obs
Development of HadGEM3: UTH
New
New-Old
Old-Obs
New-Obs
Development of HadGEM3: Clear-sky OLR
New
Old-Obs
New-Old
New-Obs
Summary
© Crown copyright Met Office
Summary – 1
• TCWV data sets need to demonstrate
superiority compared to re-analyses, which are
continually improving
• Requirement for reliable vertical profiles of
moisture to assess model biases
• Increasing move towards forward modelling
and away from retrieved quantities
• Value of data is greatly enhanced in
combination with other information such as
radiative fluxes
© Crown copyright Met Office
Summary – 2
• Model resolutions – horizontal and vertical –
are continually increasing
• Increasing focus on seasonal-to-decadal
prediction
• Continued interest in understanding and trying
to place observational constraints on feedbacks
• Work on water vapour needs to be placed
within the wider context of interest in the
hydrological cycle: rainfall, clouds, etc.
© Crown copyright Met Office