creation of more accurate, detailed, and consistent re-
analyst relies primarily on SAR imagery with a re-
ports (Page 1997). He has developed a network of re-
solution of 10300 m to begin developing his fore-
liable observers upon whom he regularly calls to verify
casts and analyses (Fig. 14). If the SAR imagery
remote sensing information. In addition to five-day
doesn't exist or is ambiguous, he can also draw upon
forecasts, the NWS in Anchorage issues year-round
AVHRR imagery or the Defense Meteorological Sat-
graphic analyses of current sea ice conditions and sea
ellite Program Operational Linescan System's (DMSP
surface temperatures. Scheduled sea ice analyses and
OLS) visible and infrared imagery (with a resolution
five-day sea ice forecasts are produced Monday,
of 0.53 km). NIC ice reports are generally more con-
Wednesday, and Friday, and a sea surface tempera-
servative than the NWS reports, reporting somewhat
ture chart of Alaskan waters is produced each Tues-
higher concentrations and older, thicker ice. The NIC's
day and Thursday. Annotated satellite analyses of sea
entire archive (123 digital images by the end of the
surface temperatures and sea ice are produced when
1999-2000 season) is available for viewing and down-
clear skies allow these features to be observed. These
loading at their website,* whereas only the single lat-
latest products are available for viewing at their
est NWS chart is similarly available.
Internet website.* Visual and infrared images from Ad-
Because of its longevity the NWS ice analysis
vanced Very High Resolution Radiometer (AVHRR)
archive was selected over the NIC database for the
sensors aboard several satellites are received directly
mapping of the sea ice in Cook Inlet. Despite some
by the Anchorage NWS. Images from satellite-borne
irregular gaps in the record, 673 NWS charts through
Special Sensor Microwave/Imager (SSM/I) and syn-
the 1998-99 season were available for analysis, an ex-
thetic aperture radar (SAR) are also available through
ample of which is shown in Figure 15. For example,
the National Ice Center or the University of Alaska
no charts were on file for the 1985-86 season until
Fairbanks after signal processing routines for ice ap-
January, there were only two charts for the entire 1986-
87 season, and there were none for all of 1996. Minor
Winter's extended darkness and low sun angles dur-
gaps are apparent during other years as well, and it is
ing daylight hours make visual images difficult to in-
not known whether reports have been lost or if none
terpret. The most reliable imagery for interpreting ice
were issued during those periods. The record is par-
conditions during the winter when sun angles are low
ticularly irregular for November and April, but this
is infrared. SSM/I and SAR images are not available
was not unexpected for the months of freeze-up and
directly and require special processing and interpreta-
melt-out. Because of this uncertainty in the data record
tion, which makes them inconvenient for regular op-
at the beginning and end of the ice season, only the
erational use. Ground-level inspection of Turnagain
NWS ice charts that were issued between the begin-
Arm conditions, for example, has on several occasions
ning of December and the end of March were used.
revealed open water when these images implied an
Accordingly no attempt was made to estimate the time
ice cover. The majority of NWS winter analyses are
of onset or termination of the ice cover.
based on AVHRR infrared images of about 1-km re-
Currently the NWS Cook Inlet ice chart archive is
solution. Conditions perceived from this imagery are
maintained on microfiche at the Alaska State Climate
verified whenever possible with ground-level inspec-
Center in Anchorage. For this atlas, each archived chart
tion and observer reports at the Port of Anchorage,
was digitally scanned to create an image that could
Nikiski, and occasionally oil platforms and vessels in
then be analyzed quantitatively using GIS software.
Cook Inlet.
Within the GIS architecture, each image was rectified
The National Ice Center (NIC) in Washington, D.C.,
and registered to a standard base map that portioned
also produces routine ice analyses for Cook Inlet on a
Cook Inlet into equal-area grid cells. Numerical values
semi-weekly schedule (Tuesdays and Fridays)
representing the particular ice concentration and stage
throughout the ice season. Appendix B shows an ex-
of development were assigned to each cell. The data
ample of an ice analysis that was produced by the NIC,
from the images were accumulated into biweekly in-
along with their explanatory material for interpreting
tervals, and statistical operations were conducted to
the image. These analyses generally differ only slightly
show the mean ice conditions in terms of concentra-
from those produced by the NWS†. Indeed the NIC
tion and stage of development, and the zones of ice
and NWS ice analysts maintain at least weekly con-
tact to share data sources and interpretation. The NIC
lated data were used to produce the ice maps shown
on the following pages. The calculation procedure is
more fully described in Appendix C.
*http:// www.Alaska.net/~nwsfoanc/
†Craig Evanego, National Ice Center, Washington, D.C., per-
*http://www.natice.noaa.gov/
sonal communication, October 24, 2000.
22