COMPARISON OF CLOUD AMOUNTS FROM MODIS, TRMM, ISCCP AND SAGE

1
COMPARISON OF CLOUD AMOUNTS FROM MODIS, TRMM,
ISCCP AND SAGE with comments on satellite
intercalibration Patrick Minnis, Louis
Nguyen NASA Langley Research Center Sunny
Sun-Mack, Yan Chen SAIC Dave Doelling ASM,
Inc. Pi Wang STX 2006 Meeting of GEWEX Cloud
Assessment, Madison, WI June 6-7, 2006
2
Cloud Products - ISCCP - SAGE - solar
occultation (250 km path length) any
cloud betwixt sensor sun will cause a cloud
response - sparse measurements (1 for each
terminator passage) - subvisible (? lt 0.03)
opaque (? gt 0.03) - MODIS Atmosphere Team
(1-km Terra Aqua MODIS) - Collection 4,
February 2000 - December 2005 - CERES (1-km
Terra Aqua MODIS, sampled to 2 km) -
Collection 4, February 2000 - December 2005
3
Summary of SAGE Tropical Measurements, 1985-1999
Opaque cloud frequency decreasing, z gt 12 km
Subvisible cloud frequency increasing, z gt 12 km
4
Another view of SAGE Tropical Measurements,
1985-1999
Opaque Cloud decreases above 12 km
Subvisual cloud increases
Wang et al., JClim, submitted
5
Comparisons with SAGE-II
6
RH300 Trends from NCEP Reanalysis
- Remember Warrens decreasing cirrus trend!
7
(No Transcript)
8
Ice/High cloud amount Comparison from
1998-2003 37N-37S
Ocean
I - ISCCP T - Terra CERES A - Aqua CERES V - VIRS
Land
Good agreement in magnitude over land Divergent
trends over ocean
Combo
Time in months since December 1997
9
Conclusions from Last Meeting (2005)
ISCCP high cloud trends appear consistent over
land with surface and other recent datasets
(should use 22 or more yrs) - decent correlation
with humidity ISCCP high cloud trends are
variant over ocean - divergent from sfc and
recent clouds - weak correlation with NCEP RH
Decreasing humidity suggests decrease in cirrus
cloudiness - SAGE analysis suggests that the
response includes thinner clouds in Tropics
drop in mean height rather than simple decrease
in cloud amount  Cause of UTH drop? Is it real?
10
Other Cloud Amount Comparisons
11
CERES GLOBAL SEASONAL CYCLE OF CLOUD AMOUNT, 2003
Open symbols refer to cloud fraction with
retrievable pixels
0.634
0.609
0.625
0.606
Very similar cycles except for June July when
Aqua lt Terra 4 of cloudy pixels do not
accommodate retrieval models!
12
CERES day
CERES vs ISCCP Cloud Amount Jul. 2000
ISCCP mean
13
Monthly mean cloud fractions from surface, ISCCP,
and CERES Terra MODIS
14
CERES MODIS CLOUD PRODUCTS ARE DIFFERENT THAN THE
MODIS TEAM PRODUCTS
 Different masks (use different channels,
thresholds, etc.) Different radiative transfer
- different ice/water models - different
atmospheric properties - different interpretive
models Different processing systems gt
differences in products
15
CERES vs MODIS Team interpretation of MODIS data
16
NIGHT CLOUD AMOUNTS, Terra, October 2003
CERES Ed2
MOD08 MODIS Team
17
DAYTIME CLOUD AMOUNT DIFFERENCE, MOD08 -
CERES Terra, October 2003
CERES fewer clouds over ITCZ eastern Antarctica
more clouds in Arctic western
Antarctic
18
Total cloud amount Comparison from
1998-2003 37N-37S
Time in months since December 1997
I - ISCCP T - Terra CERES A - Aqua CERES V -
VIRS M08- MOD08
19
DAYTIME MONTHLY MEAN CLOUD AMOUNTS, MOD08 -
CERES 30N - 30S
Trends CERES 0.017/decade MODIS 0.048/decade
20
DAYTIME MONTHLY MEAN CLOUD AMOUNTS, MOD08 -
CERES 30N - 60N 30S - 60S
Trends CERES 0.005/decade MODIS 0.017/decade
21
DAYTIME MONTHLY MEAN CLOUD AMOUNTS, MOD08 -
CERES 60N - 90N 60S - 90S
Trends CERES 0.079/decade MODIS 0.059/decade
22
TOTAL CLOUD TOP PRESSURE, Terra, October 2003
mb
CERES Ed2
MOD08 MODIS Team
23
MOD08 cloud pressures generally larger than CERES
especially in Tropics where difference is 75 mb
24
WATER DROPLET EFFECTIVE RADIUS, Terra, October
2003
µm
CERES Ed2
MOD08 MODIS Team
25
TOTAL OPTICAL DEPTH DIFFERENCE, MOD08 -
CERES Terra, October 2003
Differences large only in polar regions -
different retrieval methods over snow
26
MOD08 cloud optical depth less than or equal to
CERES Midlatitude difference is 2
27
SUMMARY
Significant differences between CERES MODIS
interpretation of clouds  Some areas of
agreement (opt depths over ocean)  Each dataset
still under evaluation  Many changes ahead for
Collection 5 (CERES Edition 3)
28
CALIBRATION MONITORING
Calibration  May or may not affect cloud
fraction  Will affect cloud optical depth  Can
produce misleading trends Critical for
producing physical quantities - modelers can be
misled by biased parameters
Proposed method for consistent intercalibration
Use self-calibrated instruments as references -
MODIS on Aqua Terra, VIRS on TRMM
Intercalibrate with moderately tight matching
constraints over ocean  Determine trends in each
satellite with deep convective cloud targets
Provide correction factors for differences in
filter functions
29
APPROACH
EXAMINE RELATIVE TRENDS IN CALIBRATED IMAGER
CHANNELS - Terra-VIRS - Aqua-VIRS - MODIS vs
CERES SW Transfer calibration to other
satellites - use methods of Minnis et al. (2002,
JTech)
30
EXAMINE RELATIVE TRENDS IN IMAGER
CHANNELS Terra-VIRS, VISIBLE
Compute slope for each month
VIRS Version 5a Version 6
31
EXAMINE RELATIVE TRENDS IN IMAGER CHANNELS VISIBLE
VIRS vs Terra
VIRS vs Aqua
Aqua brighter (1-2) than Terra VIRS V5a
appears to be ok, V6 seems to have added a trend!
32
MORE VISIBLE CHANNEL COMPARISONS Terra vs Aqua
Aqua brighter (1-2) than Terra in direct
comparison Confirms VIRS V5a conclusion
33
INTERCALIBRATIONS
Comparison of CERES SW and MODIS 0.635 µm, Jan
2000 - Mar 2005
Slope of CERES vs MODIS SW vs 0.64 µm
Aqua
Terra
No trend for Terra apparent trend for Aqua
Terra Aqua MODIS may trend relative to each
other Terra darker than Aqua by 1.2 at start of
2003
34
Comparison of CERES SW and Terra MODIS 0.635 µm,
Jan 2000 - Jul 2005
Terra discontinuity at day 1407, gains were
tweaked. Any trends caused by sudden change in
gain
35
(No Transcript)
36
Deep Convective Cloud Radiances Corrected to
Overhead Sun, VIRS V5a, 1998-2003
Monthly pdf
SZA correction factor
T11i lt 205.0 K, SZA lt 40, VZA lt 40, 10 lt RAA lt
170, ?(T11) lt 1.0 K, and ?(?) lt 0.02 ?i
37
Aqua MODIS Trend 0.635 µm, July 2002 - Jul 2005
Aqua vs CERES
Aqua Deep Convective
Aqua stable as a rock! CERES is degrading!
38
Terra MODIS Deep Convective Trend 0.635 µm, Jan
2000 - Oct 2005
Terra shows apparent trend Discontinuity
causing it CERES is not degrading!
39
TRMM VIRS Deep Convective Trend 0.64 µm, Jan 1998
- Oct 2005
V5a stable as Aqua VIRS lunar calibration
used for V6 - not helpful!
40
Back to VIRS vs MODIS 0.64 µm, Jan 2000 - Jul 2005
VIRS vs Terra
VIRS vs Aqua
Theoretical slope over ocean is 1.048
V6 DCC correction aligns VIRS with both MODIS
Aqua -VIRS in nearly perfect agreement with
theory - Terra too dark!
41
Transfer of MODIS to NOAA-16 Vis Calibration
Using Polar Crossings
42
Transfer of AVHRR Vis Calibrations Back in Time
Use DCC approach to confirm/adjust relative
trends after anchoring to MODIS/VIRS
from Doelling et al., 2001 (AMS Sat Met Conf)
43
Summary Comments
 Careful intercalibration supplemented by deep
convective cloud relative calibrations can
provide a reliable calibration record - that
will link historical and current satellites -
easily applied to both polar geosynchronous
satellites  Need to account for spectral
differences - theory is probably most practical
way - empirical, more difficult, is more accurate