On the Validation and
Intercomparison of Global Water
Vapor Climate Data Records (CDR’s)
from Satellites
John M. Forsythe and Thomas H. Vonder Haar
Cooperative Institute for Research in the
Atmosphere
Colorado State University
Fort Collins, CO USA
forsythe@cira.colostate.edu
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
1
Satellites are Essential for Climate Monitoring
The NASA Water
Vapor Project
(NVAP) dataset, was
originated at CIRA in
the early 1990’s to
employ satellites to
study water vapor.
Now covering 19882001, NVAP is a
blended satellite
dataset designed to
provide daily, global
views of layered and
total water vapor.
NVAP data available at http://eosweb.larc.nasa.gov/PRODOCS/nvap/table_nvap.html
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
2
Motivation
• Water vapor in the atmosphere is expected to increase with warming (~ 7 % /
K; Trenberth et al). Water vapor is a key greenhouse gas.
• Trends have been reported in surface and radiosonde observations (IPCC,
2001) and SSM/I over oceans (Wentz et al, 2000).
• What is the error in our water vapor CDR’s? Can we detect trends? How can
new sensors (Aqua, GPS) be used to refine the water vapor CDR?
Wentz and Schabel (2000):
Global Total Atmospheric Water Vapor
from 2xCO2 Model Run (Garratt et al., 1996)
+ 2.1 % (~0.5 mm) / decade
in Tropical Oceans (SSM/I)
29
Ross and Elliott (2001)
(radiosondes) :
28
~3 % / decade over N.
America
26
27
(mm)
CSIRO Model
25
24
NVAP
Observations
23
22
1
1950
50
200099
Years from 1900
Year
148
2050
The NVAP (1988 – 1999) global average of TPW
is 24.5 mm, with no significant trend
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
3
SSMI(3)
ATOVS
AMSU
Blended satellite
products are often
used to measure
climate variables:
(e.g. ISCCP, GPCP,
NVAP)
TOVS
Each sensor has
strengths and
weaknesses
TMI
SSM/T2
TPW 0
20
40
60
80mm
Multiple satellite
products are
blended to create
the NVAP dataset.
January 1, 2000
Total Precipitable
Water (TPW) shown
here.
NVAP covers 1988 2001
0
10
20
30
40
50
60
70mm
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
4
28
Time series of NVAP Global Mean TPW, 1988 - 1999
27
(mm)
26
25
24
23
22
21
20
1
1988
13
25
37
49
61
73
85
97
109
Months from January, 1988
121
133
1999
Annual Cycle ~ 10 % of Global Mean
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
5
Comparison of the Total Column Water Vapor, Sea Surface
Temperature, and Lower Tropospheric Temperature Anomalies
- Global Means
Three Independent Satellite Measurements – Highly Coupled
1.50
0.8
Mt. Pinatubo
Eruption
March 1991
1.20
0.90
Major El
Nino begins
May 1997
0.7
0.6
0.5
0.4
0.60
0.3
0.2
0.1
0.00
0.0
oC
(mm)
0.30
-0.1
-0.30
-0.60
-0.90
-1.20
-0.2
-0.3
Multiple,
unrelated CDR’s
can reinforce
each other
Total Column Water Vapor Anomalies - NVAP
Lower Tropospheric Temperature Anomalies MSU
Sea Surface Temperature Anomalies - Reynolds
-1.50
1988
-0.4
-0.5
-0.6
-0.7
-0.8
1989
1990
1991
1992
1993
1994
1995
1996
2nd NOAA/NESDIS Cooperative Research Program Symposium
1997
1998
1999
July 13-14 2005 Madison WI
6
Global TPW Anomalies
NOAA Operational TOVS seems to be biased low and
to have larger amplitude than SSM/I or radiosondes
3.00
2.00
1.00
NVAP TPW
0.00
(mm)
8-Radiosonde
only
6-SSMI only
-1.00
-2.00
3-TOVS only
-3.00
Jan-98
Jan-96
Jan-94
Jan-92
Jan-90
Jan-88
-4.00
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
7
NVAP SSM/I Instruments Usage
| 1988 | 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | 1995 | 1996 | 1997 | 1998 | 1999 | 2000 | 2001 |
F8
Every transition in this
record represents a
challenge for climate
monitoring
11/15
F10
4/99
F11
5/95
1/00, 2/00
F13
F14
5/97
3/00
F15
1988, 1989, 1990, 1991, 1992 All months- F8
1993, 1994 All months - F10, F11
1995 All months - F10, F11, F13 starts 5/95, except: Jul - F10, F11 only
1996 All months - F10, F11, F13
1997 F10 ends 11/15, F11, F13, F14 starts May, except: Feb - F10, F11 only, Dec - F11, F14 only
1998 All months - F11, F13, F14, except: Apr - F11, F14 only, Aug - F11, F14 only
1999 F11 ends Apr, F13, F14
2000 F11 Jan and Feb, F13, F14, F15 starts March
2001 All months – F13, F14, F15
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
8
NVAP Mode (Most Common) Data Source Changes Through Time
A reanalysis of NVAP
is needed to reduce
time-dependent
biases
1992
1988 - 1999
Dominated by SSM/I
over ocean
8 = Radiosonde data only
7 = TOVS and SSM/I combination
6 = SSM/I only
5 = SSM/I interpolated / TOVS combination
1996
4 = SSM/I interpolated
3 = TOVS only
2 = Space interpolated-filled
1 = Time interpolated-filled
0 = Missing data
More TOVS soundings
produced by NESDIS
after 1992
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
9
Ten GCOS Satellite Climate Monitoring Principles
(from Tom Karl, NCDC)
These have not been achieved for the water vapor CDR
Minimize orbit drift
Operational production of
priority climate products
Ensure sufficient overlap
Replace prior to failure
Facilitate access to products,
metadata, and raw data
Rigorous pre-launch
calibration
Need in situ baseline
observations
Adequate on-board
calibration
Real-time monitoring of
observing system performance
Continue baseline
instrument observations
on decommissioned satellites
The Unwritten Principle
Use of multiple observing
systems and multiple
analysis teams (for the same
variable)
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
10
Conclusions
• The blended NVAP global TPW anomalies have no
significant trend from 1988 – 1999.
• The operational TOVS record is discontinuous and has a
downward trend. SSM/I and radiosonde show upwards
global trends in TPW.
• There are some significant regional trends. Whether these are real
climate trends or instrument & algorithm effects requires further study
and NVAP reanalysis.
A climate dataset must be initially inspected for
time-dependent biases, then it can be reanalyzed
to sharpen it’s ability to detect trends.
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
11
Work in Progress at CIRA “Science Stewardship of Thematic Climate Data Records:
A Pilot Study with Global Water Vapor”
Supported by NESDIS/ORA
Goal: Demonstrate scientific stewardship (Bates, 2004, AMS
Satellite Conf.) applied to the water vapor CDR.
•Problem: Aqua water vapor products became available
in mid-2002, but NVAP only covers through 2001. Aqua
is the most capable water vapor-sensing spacecraft
flown. How does Aqua compare to NVAP? GPS?
• Solution: Create ~ 6 months of NVAP from 2003-2004
using heritage data and algorithms and compare to
Aqua (AIRS/HSB, AMSR) and GPS.
Expect to submit a journal paper with these results in December
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
12
Backup
Slides
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
13
1.9 % /
decade
Wentz and
Schabel, Nature,
Jan. 27, 2000
2.1 % /
decade
1.0 % /
decade
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
14
A similar trend analysis of NVAP is in progress at CIRA
Figure 2.26: Trends in annual mean surface water vapour pressure, 1975 to 1995, expressed as a
percentage of the 1975 to 1995 mean. Areas without dots have no data. Blue shaded areas have
nominally significant increasing trends and brown shaded areas have significant decreasing
trends, both at the 5% significance level. Biases in these data have been little studied so the
level of significance may be overstated. From New et al. (2000). [IPCC 2001].
2nd NOAA/NESDIS Cooperative Research Program Symposium
July 13-14 2005 Madison WI
15