Show more...
Abstract:
Tidal glacier fronts in Svalbard are mapped using Landsat 8 orthorectified imagery. Fronts are determined for places where the ice is apparently in contact with seawater. This might not be 100% correct: there are some places where there may be sediment emerging, but if this can’t be seen clearly in the L8 image the glacier front is retained as being “in contact”, especially if it is only a short stretch.
CL_LENGTHS_2015, CL_LENGTHS_2000s: Glacier length data for two periods. I use centerline data from the Svalbard Glacier Database (Nuth et al. The Cryosphere 2013) to calculate glacier lengths in 2015 and in the 2000s coverage. These numbers should be used with caution. Note that this works OK for glaciers with a well-defined geometry in the lower glacier, and where the retreat is uniform across the glacier front. There are also glaciers with multiple centerlines; I generally use the one that makes most sense for the tidewater glacier front. 2. AREA_2015, AREA_2000s: I now include the total glacier area for both coverages. Note that with a few exceptions, I have only updated the front positions and not the outline; I did have to change boundaries on a few of the glaciers that surged between the two coverages. 3. DATE_2015, DATE_2000s: I now include the dates for L8 images. I still need to generate the dates for the 2000s front positions, but that will have to come in the next version. 4. Mean_fjord_depth, Mean_dist_to_fjord_pts: Fjord depths are calculated using an updated version of the S800 bathymetric data described in Hatterman et al 2016, which is being used in the TIGRIF modeling. Because there are few data close to tidewater fronts, I find the closest fjord depth in S800 for each front position point, and take then take the average of all these depths for each glacier. Again, this approach makes sense for some glaciers, narrow fronts in a fjord for example, but might be less realistically descriptive for a very wide “glacier”, such as Bråsvellbreen, which has a highly variable depth. Furthermore, the closest points to shore are likely influenced by the gridding method for generating S800. But until I can think of a more clever approach, this will have to be what I go with.
Quality
Fronts are determined for places where the ice is apparently in contact with seawater. This might not be 100% correct: there are some places where it looks like there may be sediment emerging, but if I can’t see it clearly in the L8 image I retain it as being “in contact”, especially if it is only a short stretch.
I also cleaned up the glacier boundaries close to the fronts, and in a very few cases, further upglacier.
Then I have tidied up the shapefile attributes to include just the following (exemplified by the first glacier):
NAME: 'Pedašenkobreen' IDENT: 11101 Geometry: 'Polygon' X: [1×316 double] Y: [1×316 double] BoundingBox: [2×2 double] NumLines: 316 TIDEWATER: 1 Comment: 'Fronts updated from SGD 2000-2010 (Nuth et al TC 2013) using 2015 L8 images.' ANALYZED: 'June 2017' ANALYST: '2000s: M. König, C. Nuth. 2015: G. Villaflor, J. Kohler' SOURCE_200X: 'SPOT5HRS-IPY_SPIRIT_PROJECT (Korona et. al.,2009)' SCENE_200X: 'SPI08-025-Svalbard' SOURCE_2015: 'LANDSAT 8 orthorectified image' SCENE_2015: 'LC82170032015252LGN00_B8' GLIMSID: 'G021039E78743N' xGLIMS: 631382.399872 yGLIMS: 8748098.834891 LENGTH_2015: 9822.8 LENGTH_200X: 9822.8 AREA_2015: 50520059 AREA_200X: 50864568 DATE_2015: '09-Sep-2015' DATE_200X: 2008 FJORD_DEPTH: 8 DIST_2_FD: 468
AREA is calculated from the X and Y, I will add LENGTH in a later version when I have cleaned up the centerline data. I removed attributes which either werent needed or which also needed to be updated, like the DEM stuff. These can come in a later version, as separate files.