Vol. 82, No. 7, February 13, 2001

Visualizing Laurentian Great Lakes Ice Cycles

Raymond A. Assel and David C. Norton

For additional information, contact Raymond Assel, Great Lakes Environmental Research Laboratory, 2205 Commonwealth Blvd., Ann Arbor, MI 48105, USA. E-mail: Assel@glerl.noaa.gov

Copyright 2001 American Geophysical Union

The availability of improved information on lake ice is timely because of the concern over global warming and its potential impacts. The cryosphere is an important indicator of climate and climate change [Fitzharris, 1996; Magnuson et al., 2000]. A computer animation of long-term median ice cover for the Laurentian Great Lakes of North America was developed in a previous study [Schneider et al., 1993]. In this article, we present information about the development and availability of computer animations of discrete annual ice cycles. These animations portray the spatial and temporal patterns of ice cover for individual winter seasons and provide an index of the winter regional climate and climate variability of the Great Lakes during the last 3 decades of the 20th century. In 1994, under the auspices of the National Oceanic and Atmospheric Administration's (NOAA) Earth System and Data Information Management (ESDIM) Program, a project was initiated to update a 20-winter digital ice concentration data base [Assel, 1983] and climatology [Assel et al., 1983]. The data reduction phase of that project is now complete, and 812 historic "composite" Great Lakes ice charts have been digitized and quality controlled. The National Ice Center (NIC) and the Canadian Ice Service (CIS) contributed the historic ice charts used in this project and continue to act in an advisory role. The first analysis product from the updated data base is a set of computer animations of the seasonal progression of ice cover extent and concentration for 23 individual winter seasons.

Historical Ice Chart Data

The ice charts digitized by Assel [1983] in the earlier study were heterogeneous in spatial observations because most were based on discrete airborne observations. Only a limited portion of the entire surface of the Great Lakes was usually observed on any given chart. The "composite" ice charts digitized in this update contain ice cover information for the entire surface area of the Great Lakes for a specific date. These "composite" ice charts are based on an analysis of all available information: ship reports, shore reports, satellite imagery, and airborne reconnaissance (visual and side-looking radar). In areas not observed on the date of the composite ice chart, ice conditions are estimated based on an analysis of persistence, climatology, and weather conditions since the last observation. Composite ice charts provided a relatively homogenous temporal and spatial data set from which to produce computer animations for individual winter seasons.

The composite ice charts were digitized using an electronic digitizer and ARC/INFO geographic information system software. Ice charts (Figure 1) portray polygons that delimit areas of uniform ice attributes such as ice concentration, ice thickness, and ice floe size. Ice attribute codes are given in the upper left panel of Figure 1. See Norton et al. [2000] for details. The original paper copy composite ice charts are in an Albers Equal Area, a Lambert Conformal Conic, or a Mercator map project (depending upon the year and data source).

Fig. 1. Composite ice chart modified from Canadian ice chart for February 22, 1978. The ice concentration, ice age class, and ice thickness class are shown for each ice polygon in a numeric code (see insert in upper right corner). This follows the World Meteorological Organization ice code for freshwater ice. If some ice information is not known, it is omitted from the numeric code as illustrated in this ice chart.

Fig. 2. Image of total ice concentration abstracted from Figure 1.

Before digitizing an ice chart, a continuity check for geographic integrity is made on the digitizing tablet by positioning the cursor over a given set of latitude and longitude coordinates on the ice chart and registering their locations. The maximum allowable root mean square error for the set of positions was arbitrarily set at 2.5 km. This precision likely exceeds the position accuracy of polygon boundaries in mid-lake areas but perhaps not nearshore, where geographic features along the shoreline can be used to make better estimates of polygon boundary locations. The 2.5-km precision is also compatible with the NOAA Great Lakes CoastWatch surface water temperature data set [Schwab et al., 1999] and will expedite further analysis of these data.

Individual ice polygons were digitized and joined to a vector shoreline coverage of the Great Lakes. The ice attribute records, which were entered as a separate computer file, were then linked with the appropriate ice polygon records to produce a linked vector coverage of polygons and associated attributes. The vector coverage was converted to raster coverages in a Mercator projection at a 2.5-km resolution. The raster coverages were converted to ASCII files, and ASCII files were corrected for inconsistencies among the ice charts in the shoreline grid cells. The corrected ASCII files are the final form of these data and are used in all subsequent analysis. The details of the methods, procedures, and algorithms used to digitize and quality control the ice charts and to produce the final corrected ASCII grids are not given here for the sake of brevity, but they can be found in Norton et al. [2000].

The number of ice charts digitized for each winter season is summarized in Table 1. The CIS composite ice charts start in the early 1970s, and the NIC composite ice charts start in the late 1980s. Both the CIS and NIC ice charts were digitized for winters from 1989 to 1994. Even though this seems to be repetitious at first glance, it is not redundant because CIS and NIC ice charts for winters when both agencies produced ice analysis are usually for different dates during a given winter. Differences in analyzed ice conditions on CIS and NIC ice charts for winters when both agencies produced ice analysis are attributed to differences in available data, analysis procedures, and methods. No attempt was made to compensate for these differences in the construction of the annual ice cover computer animations. The evaluation of these differences is left to the reader.

Computer Animations

Total ice concentration, the percent of a unit of lake surface area covered by ice, is used to make a computer animation. Animations of ice thickness and ice floe size were not made because these data are in discrete classes, which makes interpolating problematic; also, the data are often sporadic and given on some ice charts but not others or given on only a portion of an ice chart.

Ice concentration was digitized to the nearest 10% from 10% to 90% concentration and to the nearest 5% from 0% to 10% and from 90% to 100%. A linear interpolation of ice concentration, rounded to the nearest 1% concentration, was performed at each grid cell location between consecutive ice charts for a given winter season. A series of daily grids of ice concentration was produced in this manner between the dates of the first and last ice charts for each winter season.

Most winters less than 10% of the surface area of the Great Lakes was covered with ice on the date of the first ice chart, with the exceptions of winters 1973, 1976, 1977, 1990, and 1991 for CIS ice charts (Table 1a). Less than 5% of the surface area of the Great Lakes was ice covered on the date of the last ice chart. The average percentage of the total surface area of the Great Lakes covered by ice on the date of the first ice chart is 5.5% (CIS) and 2.2% (NIC). The average percentage of lake surface area covered by ice on the date of the last ice chart is 1.3% (CIS charts) and 0.2% (NIC charts). Given these facts, ice cover concentration was not extrapolated prior to the first ice chart or after the last ice chart each winter.

The daily grids of ice concentration created from the linear interpolation procedure were used to create Graphics Interface Format (GIF) files. The animations were made in FLC format and later were converted to the Audio Video Interleave (AVI) format. The animation files can be viewed and downloaded at ftp://ftp.glerl.noaa.gov/ice/animations/. Information on software to view the FLC files is available under Animation Viewers (Aaplay) at http://coastwatch.glerl.noaa.gov/software/software.html.

Twenty-three animations were produced for the CIS ice charts, and six animations were produced for the NIC ice charts, one for each annual ice cycle, respectively. Each winter's computer animation has a standard length of 182 days. Day 1 is December 1 and day 182 is May 31 (May 30 on leap years). The grids prior to the date of the first ice chart and after the date of the last ice chart each winter are coded as zero ice concentration. The file name of the computer animation for each winter season and each data source (CIS and NIC) is given in Table 1.

Information included on each frame of each animation (see Figure 2) includes a month, day, and year counter, a color legend for the ice concentration, and a data indicator (black bar positioned in front of one of three labels on the right side below the color legend: Observed Data, Interpolated Data, or No Data). Ice concentration is color coded into one of ten discrete 10% ice concentration ranges from dark blue (for 0 to 9%) to red (for 90 to 99%). A complete ice cover, that is 100% ice concentration, is also color coded as red. Note that not all the ice polygons on Figure 1 are distinguishable on Figure 2. This is because adjacent polygons in the same 10% ice concentration range in Figure 1 will have the same color in Figure 2.

The computerized animations represent an estimate of the daily average ice concentration at each 2.5-km grid cell between observed ice concentration values on ice chart dates. The average number of days between ice charts is approximately 6.6 days for CIS charts and 2.3 days for NIC ice charts. The assumption for making linear interpolation is that ice changed in a linear manner between consecutive ice charts. Of course, this is not always the case. Ice cover can undergo discrete or nonlinear changes between ice charts during episodic events-for example, storms or extended cold outbreaks--particularly during the initial ice formation period or final ice dissipation period, when rapid changes in ice conditions are more likely to occur. However, analysis of nonlinear changes between ice charts is beyond the scope of this report.

Concluding Remarks

The animations for individual winters portray many different spatial patterns for any given date or period in the winter season over the 23 winters of record. Comparing animations for different winters, it is possible to gain insight into the interannual variation of ice cover during the past 3 decades. Viewing the ice cycle for a given winter provides insight into the intra-seasonal progression of ice cover for that winter. Intraseasonal variation in ice cover is of interest for analysis of other concurrent lake-related phenomena such as the frequency and extent of lake-effect snowfall, whitefish year class size [Brown et al., 1993], and the timing of the spring coastal processes associated with the loss of ice cover [Eadie et al., 1996]. These animations should also be useful in placing the future regional climate of the Great Lakes basin in a historical perspective. For example, will large inter-annual variations in ice cover that exist under the current climate continue into the 21st century, or will they be attenuated by a milder climate? Will changes in the ice cover regime produce significant changes in the lake ecosystem [Magnuson et al., 1997]?

This report represents only a first look at the updated digital ice cover data base and is provided to make the reader aware of the availability of the computer animations. A statistical analysis of total ice concentration (maximum, minimum, median, etc.) on a weekly base period is being performed on each grid cell over the 23 winter period for 1973-1995. Summary weekly ice charts will be prepared and published as a new Great Lakes Ice Atlas. We hope to make the data base and climatology available to the public at large within a year. More studies on various aspects of these data are also planned over the next several years. The daily animation grids are being used to calculate lake-averaged ice cover over annual and monthly time periods for an analysis of the spatial and temporal characteristics of the ice cycle. We plan to publish the results in the future. An effort to augment our data base through the 1999-2000 winter season is under way with the National Ice Center and the Canadian Ice Service, and we are also looking into the feasibility of making annual updates of the ice concentration data base after that.

Acknowledgments

This work would not have been possible without the contributions of the following organizations and individuals. The NIC and CIS provided the ice charts. The NOAA ESDIM Program provided the initial funding support. Most of the data reduction and data quality control was done by staff from the Cooperative Institute for Limnological and Ecological Research (CILER). Deborah Lee (formerly GLERL) and Ned Morse (formerly CILER) provided support and insight in using ARC/INFO algorithms. Mathew Rubens (formerly CILER) provided invaluable computer programing support. Brent Lofgren and David Schwab provided internal (GLERL) review of the manuscript. This is GLERL contribution number 1187.

References

Assel, R. A., A Computerized Ice Concentration Data Base for the Great Lakes, NOAA DRERL GLERL-24*, Great Lakes Environmental Research Laboratory, Ann Arbor, Mich., 1983.

Assel, R. A., F. H. Quinn, G. A. Leshkevich, and S. J. Bolsenga, Great Lakes Ice Atlas, NOAA Atlas No. 4., Great Lakes Environmental Research Laboratory, Ann Arbor, Mich., 1983.

Brown, R., W. Taylor, and R. A. Assel, Factors affecting the recruitment of lake whitefish in two areas of northern Lake Michigan, J. Great Lakes Res., 19, 418-428, 1993.

Eadie, B. J., et al., Development of Recurrent Coastal Plume in Lake Michigan Observed for First Time, Eos, Trans. AGU, 35, 337-338, 1996.

Fitzharris, B. B. (Ed.), The cryosphere: changes and their impacts, in Climate Change 1995: Impacts, Adaptations, and Mitigation of Climate Change: Scientific-Technical Analysis, Contribution of Working Group II to the Second Assessment Report of the Intergovernmental Panel on Climate Change, Cambridge University Press, Cambridge, United Kingdom, 1996.

Magnuson, J. J., et al., Potential effects of climate changes on aquatic systems: Laurentian great lakes and Precambrian shield region, J. Hydrological Proc., 11, 825-871, 1997.

Magnuson, J. J., et al., Historical trends in lake and river ice cover in the Northern Hemisphere, Science, 289, 1743-1746, 2000.

Norton, D. C., et al., Great Lakes Ice Cover Data Rescue Project, 2000 NOAA TM ERL-GLERL-117*, Great Lakes Environmental Research Laboratory, Ann Arbor, Mich., 2000.

Schneider, K., R. A. Assel, and T. E. Croley, II, Normal Temperature and Ice Cover of the Great Lakes: A Computer Animation, Data Base, and Analysis Tool, NOAA/GLERL TM ERL GLERL-81*, Great Lakes Environmental Research Laboratory, Ann Arbor, Mich., 1993.

Schwab, D. J., G. A. Leshkevich, and G. C. Muhr, Automated mapping of surface water temperature in the Great Lakes, J. Great Lakes Res., 25, 486-481, 1999.

* Available on the Internet at http://www.glerl.noaa.gov/pubs/techrept/techrept.html.