River ice jams which form during spring breakup have been responsible for the largest recorded floods at many villages in Alaska. River ice breakup forecasts are based on numerous factors, including ice thickness, snowpack, spring air temperatures, and local ice condition reports. To augment observer reports during breakup, Alaska River Forecast Center (AKRFC) personnel join Alaska Division of Emergency Services personnel in ice reconnaissance flights along sections of the Yukon and Kuskokwim Rivers. Observations made during the flights are reported to the AKRFC for use in forecasts and warnings. Reconnaissance flights are expensive, can be delayed or grounded by poor weather, and are generally restricted to these two large rivers. It is very desirable to have an additional method of monitoring rivers for ice jams and associated flooding.

To study the usefulness of Synthetic Aperture Radar (SAR) imagery for breakup monitoring and forecasting, the AKRFC received near-real-time SAR images of selected rivers in Alaska during the 1997 spring breakup season. Most of the SAR data used by the AKRFC has come from the Canadian RADARSAT satellite. This SAR has a resolution of 25x28 meters (82x92 ft) in the mode used most often by the AKRFC.

Figure 1 - May 8 SAR image of the Yukon River between river mile 373 (top of image) and river mile 345 (bottom of image). The ice reconnaissance flight reported that ice in this reach was intact and rotten. (c) Canadian Space Agency 1997

SARs, like other radars, can collect data any hour of the day. In addition, the wavelength of SARs used to image the earth’s surface is selected to see through clouds. Thus, SAR images are not limited by weather or darkness, which is very useful for monitoring floods and river breakup.

The amount of energy backscattered from a surface depends on many factors, including surface roughness, terrain geometry, satellite-to-surface geometry and satellite look angles, and properties of the surface material. In general, smooth surfaces backscatter little energy, while rough surfaces backscatter a significant amount of energy; but this can be greatly complicated by changes in surface terrain and materials. Thus interpretation of SAR imagery is not always straight-forward.

Figure 2 - May 12 SAR image of the Yukon River between river mile 373 and 345. The main channel appears to be open, with some running ice. Ice in the upper slough may be jammed chunk ice. (c) Canadian Space Agency 1997

Qualitative analysis of the images we have studied indicate that, in general, open rivers and lakes appear dark (reflect little energy) and have good contrast with the surrounding land, while ice cover or ice in the water appear lighter and have less contrast with the surrounding land in SAR imagery. Wind-driven waves on large, open river reaches or on lakes create a rougher surface and can reflect more energy, and may be confused with ice. Sequences of images fairly close in time, and a knowledge of the meteorological conditions, can help differentiate ice and wind effects. Sufficient debris in the river can also lead to enhanced, ice-like reflections. In addition, lakes that were reported to be ice covered sometimes appeared dark and uniform, as if open, in SAR images. This may be due to partial melting of the snowpack on the ice and/or deterioration of the ice itself.

The AKRFC request for 1997 spring breakup SAR images was placed several weeks in advance of the expected breakup period. We focused primarily on the Yukon and Kuskokwim Rivers, in order to compare SAR images with observations from the ice reconnaissance flights. We also requested images for the Yentna River near Anchorage, since short reconnaissance flights could also be made to this river.

Breakup on the upper portions of the Yukon and Kuskokwim Rivers was nearly coincident with the first SAR images received. SAR images were compared with reports from the ice reconnaissance flights and to observer reports.

Examples of SAR images are presented in Figures 1, 2, and 3. These figures show the Yukon River approximately 345 to 373 miles upstream of the mouth (about 15 miles north of Grayling to south of Johnson Camp) on May 8, 12, and 19, 1997. An ice reconnaissance flight on May 8 reported that ice on this part of the river was intact, rotten, and barely lifted, this days SAR image is shown in Figure 1. The May 12 SAR image (Figure 2) shows that the main channel of the Yukon River is mostly free of ice. By May 19 (Figure 3), the Yukon River appears to be ice free in the SAR image.

Figure 3 - May 19 SAR image of the Yukon River between RM 373 and 345. The river appears to be ice free. (c) Canadian Space Agency 1997

Initial analysis of the 1997 spring breakup images indicate that there is a good correlation between reconnaissance reports and features interpreted from the SAR imagery during breakup on large rivers. In many cases image interpretation was enhanced by a sequence of images over time. SAR imagery was useful in identifying reaches on larger rivers where ice was present, where ice was running, or where ice had been cleared from the reach. SAR images which capture ice runs in progress allow the location, coverage, and length of the ice run to be determined. It appears that floating debris can also be located using SAR images.

No strong ice jams or extensive ice jam flooding were observed on the Yukon and Kuskokwim Rivers during the 1997 spring breakup season. Analysis of a few images taken in August 1997 of flooding on the upper Tanana River indicated that it may be possible to monitor the extent of floods using SAR data.

There are a few things which currently limit the use and usefulness of SAR data. The resolution of the SAR imagery limits its usefulness on small rivers. The SAR satellite is in a near-polar orbit, and there are times that the SAR can not see Alaska. Although images are now available less than a day after the data is taken, during the spring 1997 breakup study images were not available until one to three days after they data was taken. And finally, the current method of scheduling SAR images requires data requests to be placed several weeks ahead of time.

The AKRFC will continue to request spring breakup images for comparison with ice reconnaissance flights and breakup observations, in order to refine our techniques. The initial requests for the 1998 spring breakup season were submitted in late February 1998. A larger and earlier time window for images was requested for 1998, in the hope that some pre-breakup images would be obtained.

In addition, the AKRFC plans to study the usefulness of SAR data for monitoring glacier dammed lakes. Glacier dammed lake outburst floods (jokulhlaups) are a recurring threat at some locations in Alaska. After a glacier dammed lake has emptied, the lake slowly fills by collecting snow melt and rain runoff. As a lake fills, it surface area increases, and we hope that SAR images can be used to determine an approximate level of the lake by tracking changes in the area covered by the lake over time.

Back to Contents