Arctic Sea Ice Concentration from Operational Ice Charts
and Satellite Passive Microwave Data
or some other title
Stern et al., including Meier, Fetterer, Fowler
To be submitted to TGRS or JAOT
February 7, 2007
Abstract
Write this section last.
Introduction
Arctic sea ice varies annually from a minimum of about 6 million km2 in September to a
maximum of 15 million km2 in March. Sea ice is an important component of the climate
system, modulating the transfer of heat, moisture, and momentum between the ocean and
the atmosphere; affecting the salinity of the ocean mixed layer through growth and melt;
reflecting solar radiation back to space; and serving as an indicator of climate change.
The decline of Arctic sea ice over the past 27 years is well documented, both in summer
[Stroeve et al., 2005; ACIA Report, 2005; NSIDC reports on web, 2006; IPCC, 2007]
and winter [Meier, Eos article]. The primary data source for monitoring sea ice
concentration, extent, and trends since 1979 has been satellite passive microwave (PM)
observations [references]. PM-derived sea ice concentration is also used to validate sea
ice models [e.g. references] and to improve model fields through data assimilation [e.g.
Lindsay and Zhang, 2006].
Since 1972 the U.S. National Ice Center (NIC) has produced weekly or biweekly Arctic
and Antarctic sea ice charts. Data sources include various satellite sensors, aerial
surveys, ship reports, model output, information from foreign ice services, and
climatology. The charts are generated primarily for mission planning and safety of
navigation. Although the charts do incorporate PM data, this occurs only where all other
forms of data are not available [Partington et al., 2003]. Thus the NIC ice charts provide
an alternative (albeit not completely independent) source of information on sea ice
concentration.
Previous studies have compared PM-derived sea ice concentration with NIC ice charts
[Partington et al., 2003] during 1979-1994, and with Canadian ice charts [Agnew and
Howell, 2003] during 1979-1996. These studies, discussed in more detail in the next
section, generally found that the passive microwave products underestimated the sea ice
concentration compared to the ice charts. This underestimation is in agreement with
other studies that have examined passive microwave sea ice concentration [Comiso and
Kwok, 1996; Fetterer and Untersteiner, 1998; Kwok, 2002].
1
In the course of mapping the sea ice concentration using whatever source data, the
location of the ice edge is of particular importance. NIC analysts who created the ice
charts paid special attention to the ice edge and the marginal ice zone, as most military
and commercial operations take place there. The ice edge location is also important for
model validation, data assimilation, and understanding the physical controls on the sea
ice extent. For example, Lindsay and Zhang [2006] assimilate PM-derived sea ice
concentration into an ice/ocean model using a nudging technique that depends on the
difference in concentration between the model prediction and the observations. When the
difference is small, the model is weighted more heavily; when the difference is large, the
observations are weighted more heavily. This has the effect of assimilating the ice
extent, because large differences are likely to occur only near the ice edge. Thus an
accurately observed ice edge is critical to the success of the assimilation scheme. Francis
and Hunter [2006] examined the factors that drive ice edge variability, using the 50%
concentration contour from PM data as their dependent variable. An accurately observed
ice edge is important again, for correctly attributing the influence of the various driving
factors. As discussed below, the NIC ice charts depict the ice edge more accurately than
the PM data, offering the possibility of improved model validation, data assimilation, and
physical insight.
In this paper we analyze a new data set of Arctic sea ice charts spanning the period 19722004, which is available and fully documented at the National Snow and Ice Data Center
(NSIDC) web site [Fetterer and Fowler, 2006]. While the charts from 1972-1994 were
previously available on CD-ROM [Arctic Climatology Project, 2000], they have now
been combined with the charts from 1995 through 2004 and made available on-line,
including monthly climatologies of sea ice concentration at different levels, and browse
images. In the next section we describe the data sets used in this study in more detail.
This is followed by comparisons and results, and then concluding remarks.
Data Sets
NIC Operational Sea Ice Charts
Against a background of steadily improving satellite instrumentation and chart
production methods at NIC, two major efforts produced the data products that are the
foundation for this data set.
In 2000, NIC, in cooperation with the U.S./Russia Environmental Working Group
(EWG), released the 1972-1994 Arctic ice analyses in digital format [Arctic Climatology
Project, 2000]. This CD-ROM product, hereafter referred to as the EWG Atlas, is
distributed by NSIDC. The undertaking included reviewing and correcting all of the
historical weekly ice analyses in order to provide the most accurate products possible for
archiving and for creating a climatology. Russian ice chart data are included in the
product as well.
2
NSIDC asked that the ice chart files on the CD-ROM be made available in EASE-Grid
(Equal Area Scalable Earth Grid) as well as in SIGRID (Sea Ice Grid) and GIS
compatible formats. In response, ERIM International created EASE-Grid files and
included them on the CD-ROM, following the convention established with the data set
AARI 10-Day Arctic Ocean EASE-Grid Sea Ice Observations [Fetterer and Troisi, 1997].
This convention groups the many possible stages of development in a SIGRID source
data file into five layers in the EASE-Grid file. For example, the EASE-Grid multiyear
concentration layer has the combined concentrations of the WMO categories of old ice,
second year ice, and multiyear ice stages of development from the source SIGRID file.
In 2005, NIC contracted with NSIDC to update the EWG Atlas climatology. The first
goal of the project was to create 5-year, 10-year, and 33-year total ice concentration
climatology products for the Arctic region using the weekly NIC ice charts. The second
goal was to make these climatology products, along with the entire series of weekly or
biweekly ice charts, available in EASE-Grid. The equal area projection and gridded
format of EASE-Grid data [Brodzik and Knowles, 2002], along with the existence of
many other polar data sets in EASE-Grid, makes EASE-Grid format more useful for
many researchers than the GIS format distributed by NIC. Thus the weekly or biweekly
NIC ice charts for 1995-2004 were converted from GIS format to EASE-Grid format and
combined with the 1972-1994 EASE-Grid files to create the complete data set. Since the
products on the EWG Atlas have a pixel size of 25 km, the 1995-2004 charts were
converted to 25-km pixel size also.
Data quality issues associated with the ice charts are as follows: (1) The number of
satellite data sources increased over the years and the resolution improved, resulting in
more information content in the later charts. Before 1980, analysts depended heavily on
visible, infrared, and single-channel passive microwave imagery. Multi-channel passive
microwave data were added in 1980; higher resolution visible and infrared imagery in
1991; synthetic aperture radar (SAR) imagery in 1995; and scatterometer data in 2004.
(2) The analysis systems evolved over time. In 1990, NIC's first computer-assisted
production system was fully installed. NIC transitioned to digital imagery analysis tools
in 1996-1997 and began to move toward a GIS production environment. (3) More
attention to detail was given to areas of operational importance such as the ice edge and
marginal ice zones of the Bering, Chukchi, and Beaufort Seas. (4) A large proportion of
the NIC analysts are military personnel on two to three year rotations. Analysts vary in
skill and level of training. (In order to mitigate this effect, the analyst training curriculum
was codified and standardized in the early 2000s). (5) Prior to 1976 and after 1986, pack
ice in the central Arctic was charted as 9 to 10 tenths concentration, if not known to be
otherwise. The digitized charts record this as 95% concentration. Between 1976 and
1986, such ice was labeled 10 tenths. The digitized charts record this as 100%
concentration. (6) Before 1995 only total ice concentration was reported. Starting in
1995 the charts recorded partial concentrations (multiyear, first-year, thin ice). The
partial concentrations do not always add up to the total concentration. [MORE on this
later, right?]. (7) Before 1995 NIC produced only Eastern and Western Arctic charts. By
1996 NIC was producing 39 non-overlapping regional charts. (8) In mid-June 2001, NIC
changed from producing charts every week to every other week. (9) The transformation
3
of chart data from paper or vector format to gridded format results in some unavoidable
loss of information and introduction of error. These have not been quantified. (10) The
digitized land pixels in the 1995-2004 charts did not exactly match those of the EWG
Atlas. We resolved this problem by applying a consistent land mask to all the charts
(1972-2004) consisting of pixels that were land in either the 1972-1994 charts or the
1995-2004 charts. (11) The procedure for converting the 1972-1994 charts into digital
EASE-Grid format for the EWG Atlas included several iterations and quality control
steps. The procedure for converting the 1995-2004 charts was not exactly the same. (12)
We believe that total ice concentrations before 1997 are biased low relative to those after,
due primarily to the introduction of high resolution SAR imagery around that time. (13)
The weekly charts are based on data that are not a snapshot of conditions, but rather were
assembled over several days. The analysts project information forward so the chart is
valid on a given day. This may introduce errors.
So—what can we say about the overall accuracy (or uncertainty) of the total ice
concentration numbers? We have to say something, and give reasons, even if they’re
basically guesses. We should also reference Partington’s (2003) estimate(s).
The primary strength of the NIC charts is that they were created by specialists using
manual analyses of data from many sources. For example, if the ice concentration is
difficult to gauge in the summer using SAR imagery because of surface melt, the analyst
has the option of checking visible band imagery. If the ice edge cannot be located in
visible band imagery because of clouds, the analyst can use scatterometry. This (mostly)
manual form of multi-sensor data fusion produces ice information more accurately than
any single data source or automated approach. Also, the operational source data receives
additional quality control by NIC before being added to the final chart product, RIGHT?
To summarize: The weekly 1972-1994 charts are from the EWG Atlas in EASE-Grid
format. They consist of total ice concentration, with a separate designation for landfast
ice. The weekly 1995-2004 charts (biweekly starting mid-June 2001) were converted to
EASE-Grid by NSIDC. They consist of five layers: the concentrations of multiyear ice,
first-year ice, thin ice, total ice, and a binary (yes/no) landfast ice indicator. All charts are
361 x 361 pixels with a nominal pixel size of 25 km. SUMMARIZE THE ACCURACY
in one short sentence. See Fetterer and Fowler [2006] for the full documentation.
Landfast ice pixels are counted as 100% concentration for the purposes of calculating the
total ice concentration, ice area, or ice extent within a region. However, in this paper we
do not examine the landfast ice distribution or trends; see Yu et al. [2007].
Figure 1 is a sample digitized ice chart from DATE, showing the total ice concentration
for that date. DO WE even need this?
4
Satellite Passive Microwave Products
A few sentences about SMMR and SSM/I, what passive microwave is, what it detects.
Describe the NASA Team (NT) products [Cavalieri et al., 1996, updated 2006].
Describe the Bootstrap (BT) products [Comiso, 1995, updated 2005].
Give space/time coverage, how the algorithms differ, what they were designed for, and
the estimated uncertainties in sea ice concentration [references].
Some points are: (1) NT and BT use the same underlying passive microwave data. The
algorithms are designed for different purposes (say what). (2) NT and BT result in
different sea ice concentrations. The differences are seasonal. BT is generally higher
than NT. Maybe show a figure. (3) There are inconsistencies in the PM sea ice time
series. The transition from SMMR to SSM/I is evident when you look at BT minus NT.
There is seasonal variation. There are even differences within the same algorithm,
depending on tie points, weather filter, coastal contamination/extended land mask. (4)
Known underestimation of sea ice concentration esp. in summer by NT [references].
Mention other PM algorithms. NT2. NRTSI.
Walt writes:
BT and NT are applicable to both hemispheres. As I understand it, the motivation for BT
was to resolve deficiencies of NT in the Antarctic. But they both use different tiepoints
for each hemisphere. I’ve found that BT seems to be better than NT, in general, in the
Arctic; I haven’t looked at the Antarctic much.
The inconsistency issue for PM (point 3) is a good one. It’s something I’ve noticed, but
nothing has been published on it, so I think that’s a potential key result, especially if we
could provide some insight on (1) what caused it, (2) which algorithm it occurs in.
Previous Comparisons of Ice Charts and Passive Microwave Products
A study based on digital versions of the NIC charts covering the Arctic every week from
1972-1994 [Partington et al., 2003] shows that the charts consistently report about 4%
more ice per unit area than passive microwave retrievals from the NASA Team
algorithm. This holds for November through May. Beginning in June, the difference rises
to about 23%, and falls off gradually over the summer and into fall freeze-up. The
difference after freeze-up (which begins in September over most of the Arctic) is
probably due to the insensitivity of the passive microwave algorithm to thin ice. Both
chart data and passive microwave data show a negative trend in integrated arctic-wide
concentration over the period 1979-1994. The difference between the passive microwave
and chart trends is statistically significant only in the summer, where it is about 2% per
decade steeper in passive microwave data.
5
A comparison of ice-covered area from the NASA Team algorithm with 18 years of
Canadian Ice Service charts showed that passive microwave data markedly underestimate
ice area by 30% to 40% during spring melt and fall freeze-up, for the Hudson Bay and
East Coast regions [Agnew and Howell, 2003]. There is considerable scatter in the
differences rather than a consistent pattern. The difference between chart and passive
microwave-derived ice areas is greater for the Canadian charts than the U.S. charts. This
is likely a reflection of the fact that the U.S. National Ice Center uses passive microwave
data when other data are not available (which is often the case for the central Arctic and
other remote areas) while the Canadian Ice Service only rarely uses passive microwave
data, relying instead on airborne and satellite radar, satellite optical, and visual
observations for charts of the Canadian Arctic. These methods detect thin ice, lower
concentrations of ice, and flooded ice much better than passive microwave data allows
[personal communication, J. Falkingham, Chief of Operations, Canadian Ice Service,
December 2002].
Spot checks of the ice edge position using a 15% concentration cutoff against NIC ice
charts show that when there is a broad, diffuse ice edge, the NRTSI and NASA Team
products sometimes do not detect ice where the concentration can be as high as 60%.
When the ice edge is more compact, the 15% concentration cutoff reflects its location
fairly well. The large footprint of the 19 GHz channel means that a compact ice edge will
be smeared out in passive microwave imagery.
A study comparing passive microwave sea ice concentration data with 1-km resolution
imagery from the Advanced Very High Resolution Radiometer [Meier, 2005] focuses on
the ice edge. Four SSM/I algorithms are used. The work illustrates how algorithms often
underestimate concentration. The NASA Team products underestimate concentration by
about 10% on average, and by much more in some circumstances.
Newer algorithms have been developed that perform better than the NASA Team
algorithm. An enhanced version of the NASA Team algorithm, "NT2," incorporates the
SSM/I 85 GHz channel and applies a forward-radiative transfer model to correct for
weather effects that are exacerbated by use of the 85 GHz channel. This algorithm is the
standard algorithm for arctic sea ice concentration retrievals with the AMSR-E
instrument [Markus and Dokken, 2002].
So… summarize, and say what is new and different about our comparisons in the next
section.
Comparisons and Results
Sea Ice Area and Extent
In the first outline, I had separate sections for the hemispheric comparison and the
regional comparisons. But now I think they should be combined, because the sum of the
regions is the hemisphere, and all the numbers can go in one table (see Table 1).
6
Let Ak be the area of pixel k, with sea ice concentration Ck. The sea ice area within a
region is the sum of AkCk over all pixels in the region. The sea ice extent is the sum of Ak
over all pixels for which Ck is above some threshold, such as 15%. In this section we
compare sea ice area and extent from the NIC ice charts, NT algorithm, and BT algorithm
for nine regions of the northern hemisphere (Fig. X), and the hemisphere as a whole, in
different seasons (?). OK, what exactly are we going to do?
First we create monthly-average ice charts by averaging together all the weekly charts
within a given month.
Then we take the monthly-average NT and BT products and convert them to EASE-Grid
by nearest-neighbor interpolation (right?).
Then we have to reconcile the land masks: EASE-Grid vs. SSMI-to-EASE-interpolated.
So we assign land to any pixel that is land in either case (right?).
We also have to convert the region mask from SSMI to EASE.
Then we can do comparisons by region (or even by pixel – i.e. difference images).
Walt wrote:
We have a paper coming out in Annals Glaciol., where we looked at regional trends
(using enhanced regions) for NT, so could run for the others fairly easily. We also did
significance testing, which is probably worth trying to do as well.
Great! Did you use an EASE-Grid version of NT?
Let’s back up a little bit and consider that we have 3 data sets (NIC, NT, BT). For each
data set and each region (10 regions, including the whole northern hemisphere) we have a
time series of sea ice area (and repeat all of this for sea ice extent). What do we want to
show (in the way of plots) and report (in the way of numbers in a table)?
Some possibilities for time series plots:
 10 panels (one for each region) with 3 plots per panel (for NIC, NT, BT). But the
time series will often be almost on top of each other, I think.
 Difference plots (NIC-NT, NIC-BT, BT-NT) to show seasonal variation.
 Plots by season (winter, spring, summer, fall averages).
 Plots by month (maybe just September and March).
 Include trend lines on some or all of the plots.
Table(s) should report:
 Means and standard deviations of sea ice area.
 Means and standard deviations of the differences in sea ice area between data sets.
 Means and standard deviations of differences by season and maybe by month.
 Overall trends and seasonal trends.
See Table 1 below.
7
Some points to make, not necessarily in this order:
(1) In Walt’s sea ice extent plot, we see an increase in NIC extent in 1995 when NIC
started using SAR data, which allowed better identification of the ice edge. This is also
around the time when NIC transitioned to digital chart production, and began creating
regional analyses rather than hemispheric. Also the chart data for 1995-2004 were
processed differently than the pre-1995 chart data. So several factors here. (Why do we
think the SAR factor is the most important one?) Also the switch to SAR caused rather
significant changes in the trend, depending on time period, whereas the PM data sets are
more consistent and the changes in trends over the years are due to changes in the ice
cover, not changes in the data quality.
(2) NIC assigned 10/10 ice concentration (hence 100% in the digitized charts) to the
central Arctic during 1976-1986, but 9/10 to 10/10 ice concentration (hence 95% in the
digitized charts) before 1976 and after 1986. This accounts for NIC ice area being
slightly higher than NT and BT before1986, and then lower from 1987 to 1995 (when the
SAR boost kicked in).
(3) In general, NIC shows the most ice, then BT, then NT. The biggest differences of
NIC minus BT are in the regions where the ice edge is present. These results motivate
the analysis in the next section on the ice edge.
(4) BT-NT shows the differences in the PM algorithms. How do they compare to the
differences NIC-BT and NIC-NT? Can see the switch from SMMR to SSM/I at the end
of 1987.
(5) Can we explain the seasonal differences, why there is a seasonal cycle in NIC-BT for
example?
Should we also show some comparisons of weekly NIC charts with corresponding daily
PM products, as in Walt’s poster (his Figure 3)? Walt wrote:
The charts are produced to be valid on a given day – they interpolate/extrapolate earlier
data sources to that day. However, using daily PM isn’t the best, especially for area,
because there can be day-to-day changes that aren’t real but are artifacts of weather,
surface anomalies, etc. I initially was going to use a 3 or 5 day mean centered on the
chart day for PM. But that started to get to be a hassle when you take into account
SMMR because SMMR doesn’t have daily data. I’m sure it could be done, I just ran out
of time to solve those.
OK. If we think some case studies would be a good idea, we could just do them during
the SSM/I era when there was daily data. It would take a bit of work though. For the
SMMR era we could pick a chart date that coincides with a SMMR daily product, and
average the SMMR daily product with the ones from plus-or-minus 2 days. Again it
would take some work.
Comparison of Ice Edge
Note that the definition of the ice edge depends on an arbitrary concentration cutoff in the
chart data and PM data, because the pixels are so large that they contain a mix of ice and
water, unlike, say, SAR pixels of 100m.
8
This would attempt to quantify what I illustrated in the bottom half of my poster: The ice
concentration is generally higher in NIC than in BT near the ice edge. (Again this is
monthly average data). Here is the idea. First pick a month, say March. For March the
ice edge falls naturally into geographic segments: Sea of Okhotsk; Bering Sea; Baffin
Bay / Labrador Sea; south tip of Greenland to Svalbard; Svalbard to Barents Sea. These
are the regions 1, 2, 4, 5, 6 of the regional comparison above. For each region, look at
the “extent map” like I show in my poster: yellow is where NIC and BT are both > 15%;
blue is where NIC > 15% but BT is not; red is where BT > 15% but NIC is not.
Calculate the blue area and the red area. Also calculate the length of the ice edge (details
needed here). Then divide the blue area and the red area by the length of the ice edge.
This gives an average “extent anomaly”, i.e. the average distance by which the NIC ice
edge is “farther out” than the BT ice edge (for the blue area), or the BT ice edge is farther
out than the NIC ice edge (for the red area). For each region we then have a time series
(of Marches, 1979-2004) of extent anomalies relative to the 15% concentration cutoff.
We can repeat for other concentration cutoffs. We could do 15%, 20% (suggested by
Walt), and 50%. As for other months: we should do September, but of course the ice
edge no longer exists in regions 1, 2, and 4. We would use regions 5, 6, and the central
Arctic (region 7). I suppose we could do December and June too. And we could repeat
the whole thing for the NT data. Then we make some summary statistics and a plot or
two to illustrate the results.
Walt’s good ideas:
One thing to consider is that the charts map the actual ice edge, i.e., the 0% contour. For
PM, 15% is often used because it agrees best with the ice edge. Due to the sensor
footprint, etc., you can’t define a 0% contour. So it might be worth comparing PM 15%
(or other %) to NIC 0% - which PM % provides the best agreement with NIC?
Another thing, which may be too much to try to do, but is there a way we might be able
to compare diffuse edges with sharp edges? – this probably has a lot to do with what % is
best and how accurate in general the PM algorithms are. Maybe use the charts as a
baseline (e.g., ice conc within 1-2 pixels of edge of 80-100% is a sharp ice edge, but ice
conc within 1-2 pixels of 50% or less is a diffuse ice edge)?
Multiyear Ice, First-year Ice, Thin Ice
This section is the third column of Walt’s poster, his Figures 3, 4, 5. Describe how the
partial concentrations were corrected so they add up to 1. Result: Relative to the charts,
maximum PM underestimation of total ice concentration occurs between late June and
early August, minimum underestimation at onset of freeze-up. This is explained in terms
of PM response to MY, FY, and thin ice as in final paragraph of poster and Fig 5.
How to assess the accuracy of MY, FY, and thin ice? For MY ice we can cite the recent
GRL paper by Nghiem et al. [2006] where they used QuikScat (QS) data. To actually
acquire and use QS data ourselves is probably too much extra work.
9
Walt adds:
I didn’t calculate any trends for the partial ice conc., but it doesn’t look like there are any
significant trends.
This is another area that is new, but there are some issues I think in terms of the
consistency of the partial fields. For example, as I think I mentioned, they didn’t
generally label a lot of FY as FY, so the sum of the partials doesn’t add up to the total – I
had to add the “unclassified” ice to the FY category. I think there’s also potentially some
inconsistency in how they transition ice between categories.
Conclusions
Recap of the chart data set (with reminders about its limitations), say what we did again,
and give the main results. Mention that there will be updates to the chart data set.
Acknowledgments
We acknowledge the far-sighted officers, enlisted personnel, and civilians at the National
Ice Center who have made it possible for NIC charts to be published as a research data
set. Production of the EASE-Grid climatology was supported by contract N00600-05-P0169 from NIC. Other support came from the National Science Foundation, grant
ARC0454912, and by funding from NOAA's National Environmental Satellite, Data, and
Information Service (NESDIS) and the National Geophysical Data Center (NGDC).
References
These are directly cut-and-pasted from the on-line documentation, so we will have to
check them carefully and delete those that we don’t actually use.
Arctic Climatology Project. 2000. Environmental Working Group joint U.S.-Russian sea
ice atlas. Edited by F. Tanis and V. Smolyanitsky. Ann Arbor, MI: Environmental
Research Institute of Michigan, in association with the National Snow and Ice Data
Center. CD-ROM.
Agnew, T., and S. Howell. 2002. Comparison of digitized Canadian ice charts and
passive microwave sea-ice concentrations. IGARSS 2002, Toronto, Ontario, Canada,
Conference Proceedings. http://ieeexplore.ieee.org/iel5/7969/22036/01024996.pdf.
Accessed 25 September 2006.
Agnew, T., and S. Howell. 2003. The use of operational ice charts for evaluating passive
microwave ice concentration data, Atmosphere-Ocean 41 (4), 317-331.
Bertoia, C., J. Falkingham and F. Fetterer. 1998. Polar SAR data for operational sea ice
mapping. In Analysis of SAR Data of the Polar Oceans. Edited by C. Tsatsoulis and R.
Kwok, 201-234. Springer-Verlag, Berlin.
10
Bertoia, C., M. Manore, and H. S. Andersen. 2001. Mapping ice covered waters from
space. Backscatter 12 (3) , 14-22.
Brodzik, M. J. and K. W. Knowles. 2002. "EASE-Grid: a versatile set of equal-area
projections and grids." In M. Goodchild, ed. Discrete Global Grids. Santa Barbara, CA,
USA: National Center for Geographic Information & Analysis.
Canadian Ice Service. 1992. Ice thickness climatology, 1961 - 1990 normals. Minis.
Supply Serv. Can. En. 57-28. Can. Commun. Group, Ottawa, Ont. 277 pp.
Canadian Ice Service, 1971-2000: Canadian Digital Regional Ice Charts, Ottawa, Ont.,
unpublished.
Carrieres, T. 2000. Documentation for the Canadian Ice Service digital sea ice database.
Can. Ice Serv., Ottawa, Ontario. 49 pp.
Comiso, J. C., and R. Kwok. 1996. Surface and radiative characteristics of the summer
Arctic sea ice cover from multi-sensor satellite observations. Journal of Geophysical
Research 101 (C12): 28, 397-28, 416.
DeAbreu, R. J. Yackel, D. Barber, M. Arkett. 2001. Operational satellite sensing of
Arctic first-year sea ice melt. Canadian Journal of Remote Sensing 27(5), 487-501.
Dedrick, K., K. Partington, M. Van Woert, C. A. Bertoia, and D. Benner. 2001. U. S.
National/Naval Ice Center Digital Sea Ice Data and Climatology. Canadian Journal of
Remote Sensing 27(5), 457-475.
Environment Canada, 2005. Manual of Standard Procedures for Observing and
Reporting Ice Conditions (MANICE). Issuing authority: Assistant Deputy Minister,
Meteorological Service of Canada. Available online. (http://iceglaces.ec.gc.ca/App/WsvPageDsp.cfm?ID=172&LnId=23&Lang=eng)
Fetterer, F., compiler. 2006. A selection of documentation related to National Ice Center
sea ice charts in digital format. NSIDC Special Report 13. Boulder, CO, USA: National
Snow and Ice Data Center.
Fetterer, F., and V. Troisi. 1997. AARI 10-Day Arctic Ocean EASE-Grid sea ice
observations. Boulder, CO, USA: National Snow and Ice Data Center. Digital media.
Fetterer, F., and N. Untersteiner. 1998. Observations of melt ponds on Arctic sea ice.
Journal of Geophysical Research 103 (C11): 24, 821-24, 835.
Geiger, C. 2006. Propagation of uncertainties in sea ice thickness calculations from
basin-scale operational observations: A report prepared for the International Ice Charting
Working Group and the National/Naval Ice Center. ERDC/CRREL TR-06-16. 39 pp.
11
Jeffers, S., T. A. Agnerw, B. T. Alt, R. De Abreu and S. McCourt. 2001. Investigating the
anomalous sea ice conditions in the Canadian High Arctic (Queen Elizabeth Islands)
during summer of 1998. Annals of Glaciology 33, 507-512.
Intergovernmental Oceanographic Commission. 2004. SIGRID-3 : a vector archive
format for sea ice charts / developed by the International Ice Charting Working Group's
ad hoc format team for the WMO Global Digital Sea Ice Data Bank Project. (Geneva) :
World Meteorological Organization ; (Paris) : Intergovernmental Oceanographic
Commission of UNESCO.
Knight, R. W. 1984. Introduction to a new sea-ice database. Annals of Glaciology 5.
(Reproduced in NSIDC Special Report 13).
Kokaly, R. F. 1996. Methods for regridding AARI data in SIGRID format to the EASEGrid projection. NSIDC Special Report 9.
Kokaly, R.F. 1996. NSIDC Special Report - 6: An evaluation of the results from using
nearest neighbor and area min/max resampling methods to regrid AARI digitized sea ice
charts to EASE-Grid.
Kwok, R. 2002. Sea ice concentration estimates from satellite passive microwave
radiometry and openings from SAR ice motion. Geophysical Research Letters 29 (9):
10.1029/202GL014787.
Maillard, P. and D. A. Clausi. 2005. Operational map-guided classification of SAR sea
ice imagery. IEEE Transactions on Geoscience and Remote Sensing, Vol 43, No. 12, pp
2940-2951, doi 10.1109/TGRS.2005.857897
McKenna, P., and W. N. Meier. 2002. SSM/I sea ice algorithm inter-comparison:
Operational case studies from the National Ice Center, IGARSS Proceedings,
INT_A32_04, Toronto, 24-28 June 2002.
Meier, W. N, M. L. van Woert, and C. Bertoia. 2001. Evaluation of operational SSM/I ice
concentration algorithms, Ann. Glaciology, 33, 102-108.
Meier, Walter, 2005. Comparison of Passive Microwave Ice Concentration Algorithm
Retrievals With AVHRR Imagery in Arctic Peripheral Seas. IEEE Transactions on
Geoscience and Remote Sensing, Vol 40, No. 6, pp 1324-1334.
Meier, W. N, M. L. van Woert, and C. Bertoia. 2001. Evaluation of operational SSM/I ice
concentration algorithms, Ann. Glaciology, 33, 102-108.
Meier, W. N., T. Maksym and M.L. Van Woert, Evaluation of Arctic operational passive
microwave products: a case study in the Barents Sea during October 2001. Ice in the
Environment: Proceedings of the 16th IAHR International Symposium on Ice, Dunedin,
12
New Zealand, 2nd-6th December 2002, International Association of Hydraulic
Engineering and Research
Melling, H. 2002. Sea ice of the northern Canadian Arctic Archipelago. J. Geophys. Res.,
107, C11, 3181, doi:10.1029/2001JC001102.
Partington, K., T. Flynn, D. Lamb, C. Bertoia, and K. Dedrick., 2003. Late twentieth
century Northern Hemisphere sea-ice record from U.S. National Ice Center ice charts, J.
Geophys. Res., 108, C11, 3343, doi:10.1029/2002JC001623.
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 analysis 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.
Ropelewski, C. F. 1983. Spatial and temporal variations in Antarctic sea ice (1973-82).
Journal of Climate and Applied Meteorology, vol 22, 470-473.
Singarayer, J. S., P. J. Valdes, and J. L. Bamber. 2005. The atmospheric impact of
uncertainties in recent Arctic sea ice reconstructions, Journal of Climate, 18, 3996-4012.
van Woert., M. 2002. U. S. Navy operational sea ice remote sensing. IGARSS
Proceedings, INT_A32_04, Toronto, 24-28 June 2002.
Willis, Z. M, Foster, C. Bertoia, and K. Dedrick. 200. National Ice Center/Naval Ice
Center support to submarine operations Undersea Warfare: Issue 8, pp. 18-21.
World Meteorological Organization (WMO). 1970. WMO sea-ice nomenclature=
Nomenclature OMM des glaces en mer= (Nomenklatura VMO po morskomu l«du)=
Nomenclatura de la OMM del hielo marino. Terminology, codes and illustrated glossary.
Geneva: Secretariat of the World Meteorological Organization.
World Meteorological Organization (WMO). 2000. Sea-ice information services in the
world. Geneva : Secretariat of the World Meteorological Organization. WMO-574.
Add these references:
Cavalieri, D., C. Parkinson, P. Gloersen, and H. J. Zwally. 1996, updated 2006. Sea ice
concentrations from Nimbus-7 SMMR and DMSP SSM/I passive microwave data, 19782004. Boulder, Colorado USA: National Snow and Ice Data Center. Digital media.
Comiso, J. 1999, updated 2005. Bootstrap sea ice concentrations for NIMBUS-7 SMMR
and DMSP SSM/I, 1978-2004. Boulder, CO, USA: National Snow and Ice Data Center.
Digital media.
13
*** Shouldn’t we write this reference as “Fetterer and Fowler 2006”? ***
National Ice Center. 2006. National Ice Center Arctic sea ice charts and climatologies in
gridded format. Edited and compiled by F. Fetterer and C. Fowler. Boulder, Colorado
USA: National Snow and Ice Data Center. Digital media.
Francis, J. A., and E. Hunter, 2006: New Insight into the Disappearing Arctic Sea Ice,
Eos, Transactions, 87, #46, 509-511.
Lindsay, R. W. and J. Zhang, 2006: Assimilation of ice concentration in an ice-ocean
model. J. Atmos. Ocean. Tech., 23, 742-749.
Nghiem et al 2006 GRL
Yu, Y., H. Stern, C. Fowler, and F. Fetterer, 2007: Interannual variability of arctic
landfast ice between 1976 and 2004, submitted.
14
Table 1. Regions, Areas, Etc.
#
Name
Area*
1
2
3
4
5
6
7
8
9
Okhotsk/Japan
Bering Sea
Hudson Bay
Baffin/Labrador
Greenland Sea
Barents/Kara
Central Arctic
Canadian Islands
Gulf of St. Law.
2.15
2.19
1.10
2.65
2.85
2.46
6.90
0.54
0.58
Northern Hem.
X.XX
Field 1
Field 2
Etc
* millions of square kilometers
Parkinson et al. [1999] has 2 tables, one for sea ice extent and one for sea ice area. Each
table has the nine regions above, plus the Northern Hemisphere. Each table reports the
trends for:
Yearly
Winter
Spring
Summer
Autumn
where each of the above headings has sub-headings:
Slope ± standard deviation (103 km2 / year), Significance, %/decade
where the significance is either 99, 95, or blank, based on the F-test.
For our comparison, we will have table sections for NIC, NT, and BT.
For each region and each data set, we could report things like:
Mean and standard deviation of sea ice area, 1979-1986 (8 years, decade I).
Mean and standard deviation of sea ice area, 1987-1994 (8 years, decade II).
Mean and standard deviation of sea ice area, 1995-2004 (10 years, decade III).
Mean and standard deviation of differences, NIC-NT, NIC-BT, BT-NT, all decades.
Mean and standard deviation of differences by season (winter, spring, summer, fall).
Mean and standard deviation of differences for September and March.
Trends for the whole time period 1979-2004.
Trends for each decade.
Repeat all of the above for sea ice extent.
15
FIG 1 HERE
Figure 1. Sample digitized ice chart from DATE, showing the total ice concentration for
that date.
FIG 2(a) HERE
FIG 2(b) HERE
Figure 2. (a) Ice chart showing monthly-average ice concentration for MONTH/YEAR.
(b) SMMR or SSM/I ice concentration product, NT or BT, the monthly average for the
same month as the ice chart. Use the same color scale for both (a) and (b).
Figure X. This kind of figure (NT minus BT sea ice area in the central Arctic Ocean)
could be used to illustrate a couple of points: (1) there are consistent seasonal differences
between NT and BT; (2) the transition from SMMR to SSM/I is evident in 1988.
16
1
2
7
8
6
5
3
4
9
Figure X. Regions used in analysis of sea ice concentration. See Table 1 for names and
statistics.
17