Using Satellite Observations and Reanalyses to Evaluate Climate and Weather Models Richard Allan Environmental Systems Science Centre, University of Reading Thanks to: Tony Slingo and Mark Ringer

1
Using Satellite Observations and Reanalyses to
Evaluate Climate and Weather ModelsRichard
AllanEnvironmental Systems Science Centre,
University of ReadingThanks to Tony Slingo
and Mark Ringer
2
INTRODUCTION

• Evaluation of Weather and Climate Prediction
  Models (some examples)
• Climate prediction uncertainty dependent on
  feedback processes
• What time/space-scales are important for climate
  change
• Feedbacks generally operating on shorter
  time-scales
• but diagnosis of feedbacks may only be possible
  on longer time-scales

3
OVERVIEW OF TALK

• 1) Evaluating simulated radiation budget
• dynamical regimes, climate model, reanalysis
• 2) Clear-sky radiation and sampling
• 3) Interannual Variability
• Water vapour, cloud radiative effect, reanalyses?
• 4) Geostationary Earth Radiation Budget
• GERB, Met Office NWP model, surface radiation

4
1) Evaluating model simulations of top of
atmosphere radiation budget

• Important for the radiative/convective balance of
  model
• Valuable diagnostic of model clouds, water
  vapour, etc

5
OLR (Wm-2) (colours) Omega, hPa/day
(contours) April 1998
Model
Obs
Model - Obs
6
Ggg
Omega hPaday
SST (K)
Ringer Allan (2004) Tellus A
7

• Climate models must simulate adequately the
  properties of cloud within each dynamic regime
  and how they respond to warming
• See also, e.g.
• Bony et al. (2003) Clim. Dyn
• Williams et al. (2003) Clim. Dyn.
• Tselioudis and Jakob (2002) JGR
• Chen et al. (2002) Science

8
2) Clear-sky radiation

• Longwave cooling important for determining
  subtropical subsidence
• Clear-sky OLR important diagnostic for water
  vapour and temperature
• Difficulties in observing clear-sky radiation
• Monthly mean clear-sky radiation over convective
  regions
• Satellite will sample highly anomalous situations

9
Using ERA-40 Daily data to illustrate clear-sky
sampling bias of CERES data
10
Model-obs differences Clear-sky Sampling
Type II HadAM3-OBS Type-I
DT6.7 DOLRc
11
DOLRc (Wm-2)
12
Using ERA40 clear-sky OLR to evaluate dynamical
regimes
ERA40-CERES similar ERA40 lt CERES
ERA40 minus CERES clear-sky OLR (January-August
1998)
Allan Ringer 2003, GRL
13

• Need to account for clear-sky sampling
  differences between satellite and models
• Reanalyses offer one alternative
• Especially important where clear-sky situations
  are rare
• e.g. monthly mean clear-sky OLR differences of
  about 15 Wm-2 for tropical convective regimes

14
3) Interannual variability in water vapour and
clouds

• How do clouds and water vapour respond to global
  warming?
• Interannual variability one example of range of
  tests of climate models
• e.g. paleo, century, decadal, ENSO, seasonal,
  diurnal, etc
• Water vapour variation
• Boundary layer, free tropospheric RH, reanalyses?
• Decadal changes in cloud radiative effect

15
Evaluation of HadAM3 Climate Model

• AMIP-type 1979-1998 experiments
• Explicitly simulate 6.7 mm radiance in HadAM3
• Modified satellite-like clear-sky diagnostics

16
Interannual variability of Column Water vapour
(Allan et al. 2003, QJRMS, p.3371)
SST
CWV
1980 1985
1990 1995
See also Soden (2000) J.Clim 13
17
CWV Sensitivity to SST

• dCWV/dTs 3.5 kgm-2 K-1 for HadAM3 and
  Satellite Microwave Observations (SMMR, SSM/I)
  over tropical oceans
• Corresponds to 9K-1 in agreement with Wentz
  Schabel (2000) who analysed observed trends
• But what about moisture away from the marine
  Boundary Layer?

18
Can we use reanalyses?
Allan et al. 2004, JGR, accepted
Reanalyses are currently unsuitable for detection
of subtle trends associated with water vapour
feedbacks BUT Climatology from ERA40 is good.
Variability from 24 hr forecast from ERA40 is
much better than above.
19
Clear-sky OLRInterannual monthly anomalies
tropical oceansHadAM3 vs ERBS, ScaRaB and CERES
ga1-(OLRc/sTs4)
1980 1985 1990 1995
(Allan et al. 2003, QJRMS, p.3371)
20
dOLRc/dTs2 Wm-2 K-1 doesnt indicate consistent
water vapour feedback?
HadAM3
GFDL
HadAM3
GFDL
dTa(p)/dTs dq(p)/dTs
Allan et al. 2002, JGR, 107(D17), 4329.
21
Sensitivity of OLRc to UTH
22
Interannual monthly anomalies of 6.7 micron
radiance HadAM3 vs HIRS (tropical oceans)
(Allan et al. 2003, QJRMS, p.3371)
Small changes in T_6.7 (or RH) in model and obs
(dUTH/dTs 0 ?)
23
(additional forcings)
(Allan et al. 2003, QJRMS, p.3371)
24
Small changes in RH but apparently larger changes
in tropical cloudiness? (Wielicki et al, 2002)
25
Altitude and orbit corrections (40S-40N)
Clear LW
LW
SW
Following Wielicki et al.
(2002) Allan Slingo (2002)
26

• Water vapour changes in models and satellite data
  consistent with constant RH
• Variability in cloud radiative effect in models
  appears underestimated compared to ERB data even
  after recent corrections
• Reanalysis are at present unsuitable for looking
  at subtle changes and trends in water vapour and
  cloud

27
4) Comparisons between Geostationary Earth
Radiation Budget (GERB) data and Met Office NWP
model (SINERGEE)

• Similar spatiotemporal sampling
• model time step GERB time 15-20 minutes
• Spatial resolution 60 km
• Near real time comparisons
• http//www.nerc-essc.ac.uk/rpa/GERB/gerb.html

28
OLR
SINERGEE comparisonof Met Office NWP Model with
GERB data
GERB
Model
Example comparison 31st March 2004, 12h00
Albedo
29
Combining GERB and BSRN radiation data
30
CONCLUSIONS

• Radiation budget as function of dynamical
  regimes evaluate cloud radiative effect in
  models
• Need to account for different clear-sky sampling
  between models and data
• Interannual variability
• Decadal variations of RH small in models and data
• Variations in cloud radiative effect appear to be
  underestimated by models
• Comparisons of GERB with NWP model shorter
  timescales closer to details of parametrizations