AN ASSESSMENT OF THE GOES-VISSR FOR DETERMINING LAKE ICE BREAKUP DATES: 
IMPLICATIONS FOR OPERATIONAL MONITORING 
D.C. Rodman, R.H. Wynne, and T.M. Lillesand 
Environmental Remote Sensing Center (ERSC) 
Institute for Environmental Studies 
1225 West Dayton Street 
University of Wisconsin-Madison 
Madison, Wisconsin, U.S.A. 
ABSTRACT 
We analyzed 1830 visible band (0.54-0.70 um) Geostationary Operational Environmental Satellite Visible 
and Infrared Spin-Scan Radiometer (GOES-VISSR) scenes to derive lake ice breakup dates for 81 selected lakes 
and reservoirs from 1980 to 1994 in the U.S. upper Midwest and portions of Canada (60°N, 105°W to 40°N, 
85?W). This image processing was intrinsic to a study of lake ice breakup patterns as indicators of climate change 
in the mid- to high latitudes, the spatiotemporal domain in which temperature increases due to enhanced greenhouse 
warming are generally predicted to be greatest. Visual interpretation was performed in a temporal context, in which 
ice-off dates were determined relative to ice conditions on proximate days. Major constraints were cloud cover, poor 
spatial resolution and spectral information of the GOES-VISSR, and difficulty in distinguishing bare ice from open 
water. Comparison to available ground-derived ice breakup dates revealed a mean absolute difference of 3.2 days 
and a mean difference of -0.4 days. A metadata data base was compiled containing summary information and reduced 
versions of the GOES-VISSR images for future reference. Our results indicate that the date of lake ice breakup can 
be reliably determined from GOES-VISSR data and is inherently amenable to operational monitoring. 
1. INTRODUCTION AND OVERVIEW 
With increasing concern yet continued 
uncertainty about global climate change and its 
impacts, there is a need for new and different ways to 
attempt to quantify any change. It was with this 
understanding that we undertook a study of lake ice 
thaw dates as a robust climate proxy. As noted in 
Wynne et al. (in press), lake ice cover fills a 
significant niche in climate change monitoring both 
geographically and temporally, and is also 
measurable from spaceborne sensors. 
For this study we analyzed 1830 visible band 
(0.54-0.70pm) Geostationary Operational 
Environmental Satellite Visible and Infrared Spin- 
Scan Radiometer (GOES-VISSR) satellite images 
acquired from 1980 to 1994 to determine ice breakup 
(“ice-off”) dates for 81 lakes in the upper Midwest 
and south-central Canada (Figure 1). Lake ice-off 
dates for each year were determined by visual 
interpretation of the images, and the resulting data set 
was analyzed to detect both temporal and 
spatiotemporal trends. 
The primary focus of this paper is the image 
processing and interpretation process for this study. 
Wynne et al. (in press) have reported on the initial 
analyses of the resulting data set. Ongoing 
geostatistical analysis will allow rigorous modeling of 
86 
  
  
  
  
  
  
  
  
  
Ae MP 
o > ae, > 
s 
4 - a * 
* 
^ 
-— - T > 
& 
* 4 F 
3 T 
Figure 1 Study Lakes. 
ice breakup dates in the region in an explicitly 
spatiotemporal framework. These results are included 
here to the extent that they relate to the interpretation 
process. T] 
beginning w 
processing 
image ana 
storage. W 
for operatic 
required tim 
images, anc 
dates as an 
2.1 Hardw: 
We 
Extreme® | 
preparation 
MIPS R400 
internal clo 
configured v 
disk, a 2GE 
audio SCSI 
CD-ROM 
software uti 
ARC/INFO 
under IRIX 
The 
includes nu: 
tape drives, 
recorders. 
Wisconsin-M 
provided to ] 
a local netw 
Engineering 
building as | 
reception of 
ingestion in 
Access Sys 
national G( 
connection | 
the archival 
We 
resolution vis 
VISSR for e: 
(1980-1994) 
images acqu 
approximate 
(UTC), tran 
the image. S 
Midwest as 
Manitoba, an 
(Figure 1, p