Mapping Changes in Glacier Extent from Satellite
Landsat Thematic Mapper (TM) imagery was used to map the extent of debris-free glacier ice on Mt. Rainier and Mt. Hood. We created a multi-decadal time series of images from 1987 to 2005. Landsat TM scenes were selected minimize seasonal snow cover and maximize exposure of debris-free glacial ice. Images from selected years with little snow and minimal cloud cover were used. Snow and temperature data from in situ stations on Mt. Hood and Mt. Rainier were used to determine year with the minimum snow cover for each decade. This minimized the potential of misclassifying remnant snow as glacial ice. Debris-free glacier ice was delineated using a normalized difference glacier index (NDGI) involving Landsat TM bands 4 & 5:
NDGI = ρ 4 − ρ 5 /ρ 4 − ρ 5
where, ρ is the atmospherically corrected surface reflectance in Landsat TM band 4 (0.75-0.90 μm) or 5 (1.55-1.75 μm). A pixel was deemed to be debris-free glacier ice if it had an NDGI value of at least 0.05. This threshold was selected using high-resolution orthorectified aerial photographs of Mt. Hood from August 2005 to delineate the glaciers and a Landsat TM NDGI image from the same year. We calibrated the threshold on Eliot Glacier and validated it on Reid Glacier, both on Mt. Hood. Estimated glacier areas were within 1% of the areas determined from the aerial photographs. This threshold was then
applied to all NDGI images for Mt. Hood and Mt. Rainier. Next, we used recently acquired high resolution LiDAR elevation data to compute the slope for each glacier and then performed a slope correction to the Landsat-derived glacier area estimates.
Contradicting earlier studies that say the glaciers on Mt. Hood are receding faster than the glaciers on Mt. Rainier, it was found that from 1987 to 2005 Mt. Rainier and Mt. Hood lost similar amounts of debris-free ice extent at 14.0% and 13.9% respectfully. For both Mt. Hood and Mt. Rainier the change in true debris-free ice area was greater than that of
the projected area change due to the steep slopes of both mountains. For Mt. Rainier an increase in recession rate was shown from 1992-2005 compared to 1987-1992 while on Mt. Hood the opposite is seen.
On Mt. Rainier we noted that highly fragmented glaciers at lower elevations such as the Inter, Pyramid, and the Van Trump Glaciers lost the highest percent of their original 1987 ice extent and were also shown to be associated with new debris flows in 2006. All debris flows on Mt. Rainier initiated within the 1913 historical glacial outline but only the new producing glacial drainages had debris flows occurring within or near the 1994 glacial boundary. On Mt. Hood none of the 2006 debris flows initiated within zones of debris-free ice glacial recession from 1987 to 2005 or within the 1972 historical outline, however all debris flows from 2006 originated from streams with a direct connection to glaciers. The Newton Clark Glacier having lost the most coverage of debris-free ice from 1987 to 2005 is also associated with the highest number of debris-flows in it’s drainage
since 1980. Precipitation data for both mountains show no future trend but an increasing trend in summer temperatures for Mt. Hood is seen. This suggests that glaciers play a significant role in the initiation of debris flows but due to a lack of accurate historical debris flow initiation locations there is not enough evidence to indicate that glacier recession rates are solely responsible for producing debris flows.
