Towards a Satellite-Based Sea Ice Climate Data Record

boorishadamantΤεχνίτη Νοημοσύνη και Ρομποτική

29 Οκτ 2013 (πριν από 3 χρόνια και 10 μήνες)

68 εμφανίσεις



Towards a Satellite
-
Based Sea Ice Climate Data Record

Walter N. Meier
1
, Florence Fetterer
1
, Julienne Stroeve
1
, Donald J. Cavalieri
2
, Claire L. Parkinson
2
, Josefino C. Comiso
2
, and Ronald Weaver
1

1
National Snow and Ice Data Center, University of Colorado, Boulder, CO 80309
2
NASA Goddard Space Flight Center, Greenbelt, MD 20771

http://nsidc.org

Introduction

Sea ice plays an important role in climate by reflecting incoming solar radiation,
modifying the salinity of the upper ocean during ice growth and melt, and insulating
the ocean from the atmosphere. Thus a reliable, complete, and consistent climate
data record of sea ice extent and area is important for climate studies.

Good quality satellite records of sea ice from passive microwave sensors dating
back to 1978 already exist. These records have been used to track interannual
variability in both the Arctic and the Antarctic [1, 2]. In the Arctic, a significant
downward trend in summer ice cover has been detected [3].

These records are among the longest and most consistent from satellites. However,
there are estimates from several algorithms and no single algorithm has proven to
be superior under all ice conditions. Thus, there is still potential to create higher
quality estimates to obtain a unified sea ice climate data record.

U21A
-
0801

Background

Physical Basis for Retrieval:

The character of naturally emitted microwave
radiation is particularly sensitive to the phase of water. In particular, the solid phase
of sea ice is generally distinct from the liquid ocean waters allowing sea ice to be
distinguished from the open ocean using satellite passive microwave data. There
are several complicating factors such as emission from melting snow or meltwater
on the ice surface, wind
-
roughened ocean, and water in the atmosphere. These
factors lead to uncertainties in retrievals of sea ice cover.

Passive Microwave Sensors:

The first satellite
-
borne passive microwave
sensor was the Electrically Scanning Microwave Radiometer (ESMR), launched in
1972. However, this was a single
-
channel radiometer and had several major data
gaps, limiting the usefulness of its sea ice products for climate studies. The launch
of the Scanning Multichannel Microwave Radiometer (SMMR) in 1978 marked the
beginning of near
-
continuous coverage of sea ice that continued through the launch
of a series of Special Sensor Microwave/Imagers (SSM/I) beginning in 1987 through
to the present. Since 2002, the NASA EOS Advanced Microwave Scanning
Radiometer (AMSR
-
E) sensor, with more channels, increased spatial resolution, and
enhanced algorithms, has yielded improved sea ice estimates.

Sea Ice Algorithms:

Several algorithms have been developed to obtain
estimates of sea ice from the observed brightness temperatures. Two of the most
commonly used algorithms, the NASA Team [4] and Bootstrap [5], are presented
here. Both algorithms were developed at the NASA Goddard Space Flight Center
and the timeseries of sea ice conditions from the algorithms are archived at the
National Snow and Ice Data Center (NSIDC).

Sea Ice Products:

Time series of sea ice extent (total ocean area containing at
least 15% ice), sea ice area (total area covered by ice), and sea ice concentration
(percentage of area covered by ice) have been produced for both algorithms that
span the entire multichannel radiometer period, November 1978


present. These
products are already quite mature and contain numerous quality control
enhancements, including inter
-
sensor calibration to assure a consistent timeseries.
The products also filter contamination from mixed land
-
ocean pixels and weather
effects (from atmospheric emission and/or wind
-
roughening of the ocean). Missing
data are filled in through spatial and temporal interpolation to provide a complete
timeseries. The resulting daily and monthly estimates of sea ice cover, as well as
climatologies, browse imagery, and other ancillary data are available from NSIDC at:
http://nsidc.org/data/seaice/
. Browse images, animations, and monthly mean data
for the NASA Team algorithm are also available from NSIDC’s Sea Ice Index pages,
http://nsidc.org/data/seaice_index/
. Other sea ice data products include operational
ice charts from the U.S. National Ice Center [6] and other national ice services,
model input
-
based fields such as the Hadley Center climatology [7], and fused fields
such as a combined ESMR
-
SMMR
-
SSM/I
-
NIC timeseries [8]; these use a variety of
sources to provide complete fields for their given purposes.

Sea Ice Products

Figure 2:

Timeseries of Arctic (left) and Antarctic (right) monthly sea ice extent from the Bootstrap
and NASA Team (Nov. 1978


Sep. 2005) algorithms; Jan. 2004


Sep. 2005 NASA Team values
from the Sea Ice Index, all other values from the NASA Goddard SMMR
-
SSM/I Timeseries. Tick
marks represent January of each year.

Arctic

Antarctic

March

February

September

September

100

80

60

40

20

0

Concentration (%)

The NASA Team (NT) and Bootstrap (BT) algorithms
provide complete information on the ice cover
throughout the year (Figure 1). While care is taken to
insure each algorithm is internally consistent via inter
-
sensor calibration and quality control techniques, the
two algorithms yield different results, as illustrated in
Figure 2. The NT algorithm typically shows less ice
than the BT, particularly during summer. Each
algorithm uses a different approach to estimate sea ice
cover, including different passive microwave channels
and different reference brightness temperatures (T
b
).
For example, the NT algorithm uses brightness
temperature ratios to remove the effect of physical
temperature; the BT algorithm uses seasonally
-
adjusted reference T
b
s to improve performance during
summer melt. Each method has advantages and
disadvantages and compromises must be made
because the spatial, temporal, and spectral resolution
cannot resolve all features of the complex sea ice
surface. This is particularly true for thin ice, whose
emissivity varies with thickness depending on the
frequency. This can cause considerable uncertainties
in such regions because the algorithms are based on
general conditions over the whole hemisphere, which
is dominated overall by thicker ice. Each algorithm
makes different compromises, as discussed in a joint
study by the NASA Goddard authors [9].

The key
features of each algorithm product is that they both
use internally consistent methods, consistent
microwave frequencies, and consistent quality control.

Differences in sea ice records are further illuminated when looking at other sea ice data
records for September monthly means (Figure 3) from a variety of time series, described in
Table 1. The Goddard and NSIDC data sets are pure passive microwave products from the
NASA Team or Bootstrap and are more consistent. The NIC [6], Hadley [7], and ESMR
Merged [8] datasets are combined products using different sources through time to create
longer time series, but this leads to differences in data quality through time and possible
inconsistencies in the time series. This is most notable in the NIC and Hadley data sets after
1996. The addition of Radarsat SAR as a source allowed more thin ice to be detected and
increased estimates from the NIC charts compared to earlier years. The Hadley time series
is based on passive microwave fields, adjusted by the 1973
-
1994 NIC climatology to correct
the passive microwave summer low
-
bias; however, after 1996, the Hadley data set switched
from the Goddard NT product to NCEP sea ice fields. Thus, the NIC and Hadley post
-
1996
fields are possibly more accurate than the pure passive microwave, but they are not
consistent with their earlier period. The change in the Hadley product is particularly evident
where the 1987
-
2002 decadal trend in the Arctic is actually positive (Table 1), inconsistent
with all other time series.

A climate data record of sea ice cover should be one consistent, quality
-
controlled data set
that is able to track climate variability and change [10]. As of now there is no single clear
-
cut
superior sea ice data product. Data fusion approaches, similar to that used in the NIC,
Hadley, and ESMR Merged timeseries, can be used to produce an optimal data product. The
newer AMSR
-
E, with higher resolution and higher
-
quality sensor properties, can further
improve the SMMR
-
SSM/I timeseries as we approach the NPOESS era.

Great care must be
taken to avoid inconsistencies due to differing data sources and different processing methods,
as demonstrated by the NIC and Hadley time series. However, intelligent data fusion of the
data sets, with strict standards for consistency, could yield high quality time series. Another
important component of a climate data record is a quality assessment. There have been
several studies of the errors in the various passive microwave algorithm products, e.g. [11],
but there is not yet a data product with complete quality flags. This is the next step toward a
satellite
-
based sea ice climate data record.

Conclusion

Figure 1:

Bootstrap 1978
-
2003 monthly
mean sea ice concentration fields for the
Arctic (top row) and Antarctic (bottom row)
for the month of annual maximum (March
for Arctic, September for Antarctic) and
minimum (September for Arctic, February
for Antarctic) ice cover.

Figure 3:

Arctic
September minimum sea
ice extent from various
sources. Goddard NT
2004 & 2005 estimated
from the NSIDC Sea Ice
Index.


Product


Source


Time
Period

Total

% Decadal
Trend

1987
-
2002

% Decadal
Trend


Comments

Goddard NT

SMMR
-
SSM/I

1979
-
2005

-
8.65

-
8.93

Sensor inter
-
calibration and QC

Goddard BT

SMMR
-
SSM/I

1979
-
2005

-
7.55

-
8.36

Sensor inter
-
calibration and QC

NSIDC NT

SSM/I

1987
-
2004

-
9.22

-
7.50

Limited QC

NSIDC BT

SSM/I

1987
-
2004

-
9.32

-
7.56

Limited QC

NIC

Operational charts

1972
-
2004

-
3.01

-
4.46

Manual analysis based on a variety of
satellite imagery and other sources

Hadley

SMMR
-
SSM/I
-
NIC

1972
-
2003

-
6.16

0.78

NIC climatology used to adjust passive
microwave NT estimates

ESMR
Merged

ESMR
-
SMMR
-
SSM/I
-
NIC

1972
-
2002

-
5.40

-
8.81

1972
-
1978 NIC charts used to fill gaps
and inter
-
calibrate with ESMR

Table 1:
Data sources for
Figure 4. NT = NASA Team,
BT = Bootstrap. Time period
varies for each, but all
products overlap for
1987
-
2002. Decadal trends
are for September means.

References

[1] Gloersen, P., C.L. Parkinson, D.J. Cavalieri, J. Comiso, and H.J. Zwally, 1999. Spatial distribution of trends and season
ali
ty in the hemispheric sea ice covers: 1978
-
1996,
J. Geophys. Res.
, 104(C9), 20,827
-
20,836.

[2] Parkinson, C.L., D.J. Cavalieri, P. Gloersen, H.J. Zwally, and J.C. Comiso, 1999. Arctic sea ice extents, areas, and tre
nds
, 1978
-
1996,
J. Geophys. Res.
, 104(C9), 20,837
-
20,856.

[3] Stroeve, J.C., M.C. Serreze, F. Fetterer, T. Arbetter, W.N. Meier, J. Maslanik, and K. Knowles, 2005. Tracking the Arcti
c’s

shrinking ice cover: Another extreme minimum in
2004,
Geophys. Res. Lett.
, 32, L04501, doi: 10.1029/2004GL021810.

[4] Cavalieri, D.J., P. Gloersen, and W.J. Campbell, 1984. Determination of sea ice parameters with the Nimbus 7 SMMR,
J. Geophys. Res.
, 89(C3), 5355
-
5369.

[5] Comiso, J., 1986. Characteristics of Arctic winter sea
-
ice from passive microwave and infrared observations,
J. Geophys. Res.
, 91(C1), 975
-
994.

[6] Dedrick, K., K. Partington, M. Van Woert, C. Bertoia, and D. Benner, 2001. U.S. National/Naval Ice Center Digital Sea Ic
e D
ata and Climatology,
Can. J. Remote Sensing
,
27(5), 457
-
475.

[7] Rayner, N.A., D.E. Parker, E.B. Horton, C.K. Folland, L.V. Alexander, D.P. Rowell, E.C. Kent, and A. Kaplan. 2003. Global

an
alysis of sea surface temperature, sea ice, and
night marine air temperature since the late nineteenth century.

Journal of Geophysical Research
. 108 (D14). doi:10.1029/2002JD002670.

[8] Cavalieri, D.J., C.L. Parkinson, and K.Y. Vinnikov, 2003. 30
-
year satellite record reveals contrasting Arctic and Antarctic

variability,
Geophys. Res. Lett.
, 30(18), 1970, doi:
10.1029/2003GL018031.

[9] Comiso, J.C., D.J. Cavalieri, C.L. Parkinson, and P. Gloersen, 1997. Passive microwave algorithms for sea ice concentrat
ion
: A comparison of two techniques,
Rem. Sens.
Env.
, 60, 357
-
384.

[10] National Research Council, 2004. Climate data records from environmental satellites, National Academies Press, Washingt
on,

DC, 116 pp.

[11] Meier, W.N., 2005. Comparison of passive microwave ice concentration algorithm retrievals with AVHRR imagery in the Arc
tic

peripheral seas,
IEEE Trans. Geosci. and
Remote Sens.
, 43(6), 1324
-
1337.

Potential Climate Data Records

http://polynya.gsfc.nasa.gov