Assessment of Global Cloud Climatologies
project of GEWEX Radiation Panel / WCRP
Claudia Stubenrauch
+
GEWEX cloud assessment group
(S. Ackerman, R. Eastman, A. Evans, A. Heidinger, B. Maddux, P. Minnis,
J. Norris, W. B. Rossow, S. Sun-Mack, P.-H. Wang, S. Warren, D. Wylie…)
http://cimss.ssec.wisc.edu/cloud_climatology/2006
Co-chairs: B. Baum, SSEC, University of Wisconsin, USA
C. Stubenrauch, CNRS/IPSL - Laboratoire de MétéorologieDynamique, France
April 2007
EGU 2007
1
Satellite radiometers measure:
emitted, reflected, scattered
radiation
INVERSION
cloud detection
inverse radiative transfer
cloud properties
GEO (3hrs)+polar
ISCCP
IR,VIS
polar satellites (12/6 hrs)
PATMOS-x
IR,NIR,VIS
MODIS
April 2007
EGU 2007
UW HIRS, TOVS Path-B
IR Vertical Sounder:
CO2-band
AIRS
2
http://cimss.ssec.wisc.edu/cloud_climatology/2006
workshop: 6-7 July 2006 in Madison, USA: 20 presentations, 50 participants
Longterm cloud climatologies:
ISCCP GEWEX cloud dataset
PATMOS-x AVHRR
UW-HIRS 13h30/1h30
TOVS Path-B 7h30/19h30
1983-2006
(Rossow et al. 1999)
1981-2006
(NESDIS/ORA; Heidinger)
1985-2001
(Wylie et al. 2005)
1987-1995
(Stubenrauch et al. 2006)
SAGE limb solar occultation
SOBS (Surface Observations):
1984-1991,1993-2005 (Wang et al. 1996, 2001)
1952-1996(sea), 1971-1996(land)
(Hahn & Warren 1999; 2003)
EOS cloud climatologies (since 2000, 2002):
MODIS (Ackerman et al.) CERES-MODIS (Minnis et al.)
AIRS (Susskindet al.; Stubenrauch et al.)
Evaluation & analysis of cloud properties:
average, regional, seasonal and interannual variations, diurnal cycle
Climate monitoring: trends and where they can originate from
April 2007
EGU 2007
3
ISCCP (Rossow & Schiffer BAMS, 1999)
Evaluation
night: +75 hPa pcld bias (Stubenrauch et al. 1999)
uncertainties depend on cloud type:
Stratus (tcld>5): pcld 25-50 hPa within radiosonde meas., ~ -65 hPa bias; err Tcld<1.5 K
high clouds (tcld>5, with diffuse top): pcld 150 hPa (trp)/ 50 hPa (midl) above top
isolated thin Cirrus: difficult to detect
thin Cirrus above low clouds: often identified as midlevel or lowlevel cloud
15% tcld decrease for doubling droplet size
TOVS Path-B (Stubenrauch et al. J. Clim. 2006)
pcld uncertainty 25 hPa over ocean, 40 hPa over land (2nd c2 solution)
pcld = mid-cloud pcld,: 600m/ 2 km below cloud-top (low/high clouds) (LITE, Stubenrauch et al. 2005)
Sensitivity study for De of Ci (Rädel et al. 2003)
UW HIRS (for Wylie & Menzel J. Clim. 1999, not yet for Wylie et al. J. Clim. 2005)
pcld 70 hPa above top (lidar, Wylie & Menzel 1989)
100 hPa above
for transmissive cloud overlying opaque cloud (Menzel et al. 1992)
April 2007
EGU 2007
4
Geographical distribution of HCA
DJF
winter
strom tracks
January
SAGE
ISCCP
July JJA
TOVS Path-B
ITCZ
UW-HIRS
April 2007
EGU 2007
5
Average CA
ISCCP(84-04) day/night TOVS-B(87-95) UW-HIRS(85-01) SAGE(85-99) PATMOS-x(81-06) CE-MODIS(03-05) WI-MODIS(02-06) SOBS(84-04)
CA (%)
glo bal
oce an
all
66
73
75
Thick Ci
3
2
Cirrus
19
HCA/ CA
95
66
61
67
61
64
la nd
70
74
77
2
3
2
27
31
18
33
41
44
44
38
42
25
21
23
MCA / CA
27
16
16
20
19
16
32
33
LCA / CA
39
42
37
36
44
44
46
46
95
72
66
73
65
69
58
69
70
97
50
50
59
51
54
1
3
4
5
27
33
21
27
29
30
39
44
44
35
37
21
18
17
41
45
49
45
47
56
34
29
43
44
26
14
14
18
17
14
28
29
42
31
25
17
25
25
20
39
43
48
72
41
47
42
38
49
51
50
52
80
29
30
34
29
29
26
27
27
48
diurnal sampling, time period for ISCCP / TOVS-B: 1% effect; low-level over land: 2%
can be more important if using afternoon satellites (D. Wylie, A. Evan)
 70 % (±5%) cloud amount: 5-15% more over ocean than over land
CE-MODIS low, PATMOS low over land, SAGE CA (200km, clds t>0.03) 1/3 higher
40% single-layer low clouds: more over ocean than over land; SOBS
40% high clouds: only 3% thick Ci; more over land than over ocean
IR sounders ~ 10% more sensitive to Ci than ISCCP
SAGE cloud vertical structure in good agreement with IR sounders
HCA/CA: SAGE,TOVS/HIRS > CE-MODIS > PATMOS > ISCCPday > MODIS > ISCCPnght
April 2007
EGU 2007
6
CA seasonal cycle over land
30°-60°N
0°-30°N
UW-HIRS TOVS-B ISCCP PATMOS-x
0°-30°S
MODIS CE-MODIS SOBS
Seasonal (& diurnal) cycles:
• stronger over land than over ocean and strongest in subtropics (ITCZ)
• seasonal cycles similar, exception: SH polar land and newer datasets (PATMOS, MODIS)
• TOVS/HIRS CA 5-12% larger than ISCCP & PATMOS
April 2007
EGU 2007
7
HCA seasonal cycle over land
30°-60°N
0°-30°N
0°-30°S
H
C
A
UW-HIRS TOVS-B ISCCP PATMOS-x
HCA seasonal cycle : 7-15%,
MODIS CE-MODIS SOBS
10-27%,
18-45%
ISCCP underestimates seasonal cycle of HCA by up to 20%
UW-HIRS slightly smaller seasonal cycle than TOVS-B
PATMOS & MODIS HCA too low in NH midlatitudes
SOBS HCA seasonal cycle modulated by clouds underneath
April 2007
EGU 2007
8
TOVS Path-B: diurnal cycle of high clouds
Stubenrauch et al. J. Climate 2006
NOAA10/12 7h30 AM&PM, NOAA11 2h00 AM&PM(1989-90) NOAA11 4h30 AM&PM(1994-95)
strongest diurnal cycles over land, in tropics (& in midlat summer)
max. thin cirrus in early afternoon
max Cb (ISCCP) in early evening
max. cirrus and thick (large-scale) cirrus in evening
cirrus occurrence continues during night and decreases during day
TOVS-B extends ISCCP during night
April 2007
EGU 2007
9
Global CA anomalies
E
3 sat
4 sat.
5 sat.
global CLA within ±2.5%
UW-HIRS: more or less stable
ISCCP: ~5% decrease from 1987 to 2000
related to increasing nb of GEO satellites ?
SOBS: increasing over ocean, stable over land >1985 (Warren et al. J. Clim. 2006)
April 2007
EGU 2007
10
CA change : cluster analysis correcting for artefacts
J. Norris
http://meteora.ucsd.edu/~jnorris/isccp_artifacts.html
6 clusters out of 7 related to artificial satellite features:
sea-ice: Oct-Dec 2004, high lat land: > Oct 2001(NOAA16)
nb of GEOs - view angle, GEO view area
April 2007
EGU 2007
also
(Evan et al. GRL 2007)
11
Anomaly per cloud type
W. B. Rossow
CUMULUS
Changes in Cloud Property Distribution : decreasing t of low clouds -> below detection
(Tselioudis et al. 1992: t decreases with T)
April 2007
EGU 2007
12
Trend analysis of high clouds:
synergy of different variables
Tropics (Wang et al., J. Clim., 2007, in rev.):
• decreasing NCEP humidity suggests decrease in HCA
SAGE: thinner & lower high clouds in tropics
rather than simple HCA decrease
• cause of UTH drop? Is it real?
Midlatitudes (Stubenrauch & Schumann, GRL 2005):
23
23
clear + thin cirrus
potential contrails
20,5
eff. cloud amount
20,5
eff. cloud amount
RHice < 70% & RHwat > 0.4 RH*crit
18
15,5
13
North Atlantic
South Atlantic
10,5
8
18
15,5
13
10,5
North Atlantic
South Atlantic
8
87,5 88 88,5 89 89,5 90 90,5 91 91,5 92 92,5 93 93,5 94 94,5 95
87,5 88 88,5 89 89,5 90 90,5 91 91,5 92 92,5 93 93,5 94 94,5 95
time (years)
increase of thin Ci in both hemispheres
April 2007
time (years)
stronger increase related to contrails in NH
EGU 2007
13
Satellite observations:
 unique possibility to study cloud properties over long period
Intercomparisons:
average cloud properties: in general good agreement
70% (±5%) clouds: ~ 40% high clouds & ~40% single-layer low clouds
HCA/CA: SAGE > TOVS/HIRS > CE-MODIS > PATMOS > ISCCPday > MODIS > ISCCPnght
PATMOS-x & EOS datasets still in validation process
seasonal cycle: CA good agreement (except SH polar land)
 ISCCP HCA cycle in tropics underestimated
 SOBS LCA cycle over ocean smaller; absolute value 18% larger
regional differences (latitudinal, ocean/land):
linked to cirrus sensitivity: IR sounders & SAGE : HCA +4% (midl) to +20 % (trp)
atmos. profiles: TOVS B less HCA in SH compared to SAGE & HIRS
diurnal cycle: TOVS-B extends ISCCP during night
 Trend analysis: careful of satellite drifts, calibration etc.
synergy of different variables important !
Evaluation continues & WMO report in preparation
April 2007
EGU 2007
14