METHODS
Data Sources
Corrected BC Government DEMs for mapsheets 92F, 92G, 92J, 92N,
92O, 93D, derived from BC TRIM data
Google Earth SPOT satellite images of southern BC, 2003-2009
BC basemap from BC TRIM Program
Coordinates of post-LIA glacier extents documented by Koch et al. 2007, Koch et al. 2009, Lewis and Smith 2004, Osborn et al. 2007, Smith and Laroque 1996 (see references page for detailed source information)
The primary data source for this topographically-based analysis
is a set of
DEMs derived from BC TRIM data. Initial analysis involved identifying a
number
of potential sites over a large area, so DEMs covering most of the
extent
of the Southern Coast Mountains and a portion of Vancouver Island were
selected. Glaciers were digitized and stream networks delineated over
this
entire area, but time constraints limited slope-area analysis to 17
glaciers. This data has a resolution of 25
metres. The mapsheets were
combined into a raster mosaic, then clipped to each study area
(Strathcona, Spruce Lake, Pemberton, and Garibaldi). Glaciers and
proglacial lakes were digitized from Google Earth images, and the downstream
boundaries for channel evaluation were set based on information on LIA and post-LIA glacier
extents. All data was set to NAD 1983 BC Environment Albers,
the standard
projection of the BC Environment GIS Working Group.
Glacier Digitization
Glacial area can change drastically in a relatively short period of time. In order to obtain the most accurate representation of glacial area, Google Earth screenshot images (Google Earth satellite imagery is more recent than most topographic maps) were saved in JPEG format and opened in ArcMap. Two placemarkers were added to the image before it was saved, and placemarker coordinates were saved in a comma-separated value (csv) file. These placemarker coordinates were added to the ArcMap workspace in point form as XY coordinates after conversion into the correct spatial reference. Using the Georeferencing toolbar, the image was adjusted to the placemarkers. Glaciers were hand digitized as a polygon layer.
Both hillshade DEMs and contour maps (generated from DEMs) were used to verify orientation of glaciers and lakes.
DEM hillshade image of Colonel Foster Glacier with an overlay of the Google Earth georeferenced image, glaciers, lakes, and stream networks (network delineation process described below).
Contour image of Colonel Foster Glacier with an overlay of georeferenced image, glaciers, lakes and stream networks.
Once glaciers were digitized, two different methods of watershed delineation were explored.
Extracting Slope-Area Data: Contour Maps (Vector-Based)
The methodology for extracting slope-area data is based on Brardinoni and Hassan (2006). In order to quickly complete the process for each location, a model was developed as indicated in the following flowchart.
1. A fill operation was performed on the clipped raster layer in order to correct any data imperfections.
2. A contour map with 30 m intervals was generated from the initial raster (ideally, intervals would be smaller, but 30 m was selected for this project due to time constraints).
3. Flow direction and flow accumulation rasters were generated from the fill raster.
4. The Con operation was performed on the flow accumulation raster in order to delineate stream networks, using the expression streamnetwork = con (flowacc > 160, 1). This expression assigns the value one to all cells with more than 160 cells flowing into them, and NoData to all other cells. The minimum drainage of 160 cells was selected because it reflects a drainage area of approximately 0.1 square km, the size of the smallest glacier initially included in this study. When watershed delineation proved more time-consuming than originally expected and the number of glaciers included was reduced, the minimum glacier size became 1 square km. However, stream networks were compared to field data, Google Earth images, and topographic maps, and no major errors were detected.
5. Hand digitization was performed by tracing flow paths perpendicular to contour lines, beginning slightly downstream of the maximum LIA extent and continuing upwards to the glacier toe. The pour point for each watershed was defined as the point midway between each set of contour lines. The most effective method of performing this operation was to create and split a polygon.
6. Slope for each stream reach was determined by using the measure feature in ArcMap. Area was determined from the attribute table for each polygon and summed with each successive downstream reach.
DEM hillshade and contour overlay of Colonel Foster Glacier with glaciers, lakes, and hand-digitized watersheds.
Extracting Slope-Area Data: ArcMap Watershed Tool (Raster-Based)
The ArcMap Watershed tool offers a promising, although error-prone, alternate method for deriving slope-area data, described in the flowchart below. This project involved a brief exploration of the Watershed Tool method, but construction of watersheds in such a small area was fraught with errors. The vector-based contour method was used for this analysis because it is the accepted method in geomorphological research. However, it is presented here as a potentially useful alternate method.
Using
pour points at designated
locations along the stream, a series of watersheds was created and
the area determined for each. Slope data was obtained at each
point of area measurement by using information from a slope raster,
although the measurement tool is still necessary to obtain average
slope over each stream reach (slope has high variability from cell to
cell, so average slope is necessary for accurate analysis). The model
for the raster-based evaluation is presented below.
DEM hillshade and contour overlay of Colonel Foster Glacier with glaciers, lakes, and watershed delineated by ArcMap.
Glacier Type Classification
Pelto and Hedlund (2001), in a long-term study of glaciers in the North Cascades, identified three distinct glacier types based on glacier terminus activity. The synchronous LIA response over the region suggests that a similar type classification might be established for the BC Coast Mountain glaciers, but detailed 20th-century terminus behaviour information was not available for most of the glaciers studied in this project. However, although Pelto and Hedlund based their type classification solely on terminus behaviour, they discovered several a posteriori correlations between glacier type and certain characteristics, including slope, elevation, and accumulation.
This project explored the possibilities that: 1) southwestern British Columbia glaciers fall into similar categories of terminus behaviour as those defined by Pelto and Hedlund; 2) slope, elevation, and accumulation could be used to infer glacier category; and 3) proglacial stream morphology is related to glacier terminus behaviour.
In Pelto and Hedlund (2001), Type 1 glaciers have the highest mean elevations, largest mean slope, highest mean measured accumulation, most extensive crevassing and highest measured terminus-region velocity. Type 3 glaciers have the lowest slopes, least crevassing and lowest mean terminus velocity of any of the glacier types. Type 2 glaciers have a lower slope, a lower terminus-region velocity, less crevassing and a lower mean accumulation rate than type 1 glaciers. Type 1 glaciers underwent a period of sustained retreat after the LIA maximum, then a brief period of readvance during the mid-20th century, then retreat again in the late 20th century. Type 2 glaciers followed the same pattern, except instead of a mid-20th century advance, they underwent slow retreat or equilibrium. Type 3 glaciers have undergone continuous retreat from the LIA to the present.
For this project, glaciers were initially sorted based on existing terminus behaviour (available for Helm, Overlord, Lava, Black Tusk, and Diamond glaciers). Then, an attempt was made to identify clear linkages between a number of topographic and climatic factors, including: slope, elevation, glacier area, and mean precipitation as snow (PAS) for the past century (slope, elevation, and area obtained from ArcMap, mean PAS data obtained from Climate BC). Following Pelto and Hedlund, glacier elevation was defined as the elevation of the accumulation zone.
In order to identify potential correlations, a script was written in R.
Slope-Area Profiles
Slope-area information obtained from the DEMs was used to generate plots in Excel.
Glacier Type-Channel Morphology Correlation Analysis
Complexity in slope-area profiles indicates complexity of channel morphology. As a basic qualitative assessment of complexity in the channel long profile, linear regressions were run for each slope-area profile using R and the residual standard error was measured. A higher residual standard error was interpreted as a higher degree of channel complexity. As a measurement of cross-correlation, residual standard error (indicated on the figure below as variance), slope, and glacier type were plotted in R.