We present Calving Front Machine (CALFIN), an automated method for extracting calving fronts from satellite images of marine-terminating glaciers. The results use Landsat imagery from 1972 to 2019 to generate 22,678 calving front lines across 66 Greenlandic glaciers. The method uses deep learning, and builds on existing work by Mohajerani et al., Zhang et al., and Baumhoer et al. Additional post-processing techniques allow for accurate segmentation of imagery into Shapefile outputs. This method is uniquely robust to the impact of clouds, illumination differences, ice mélange, and Landsat-7 Scan Line Corrector errors. CALFIN provides improvements on the current state of the art. A model inter-comparison is performed to evaluate performance against existing methodologies. CALFIN's ability to generalize to SAR imagery is also evaluated. CALFIN's fronts are often indistinguishable from manually-curated fronts, deviating by 2.25 pixels (86.76 meters) from the true front on a diverse set of 162 testing images. The current implementation offers a new opportunity to explore sub-seasonal trends on the extent of Greenland's margins, and supplies new constraints for simulations of the evolution of the mass balance of the Greenland Ice Sheet and its contributions to future sea level rise.

We collect our source data from Landsat NIR band images spanning from 1972 to 2019. We provide 1997 manually-masked calving fronts, which we use for training, validating, and testing our automated algorithm. We also provide over 17912 automatically generated fronts, along with estimated mean errors calculated for each basin. This dataset contains a total of 19909 calving fronts for Helheim, Kangerlussuaq, Kong Oscar, Hayes, Rink Isbrae, Upernavik, Jakobshavn, Petermann, Kangiata Nunaata, and 62 other nearby glaciers along East/West Greenland.

We process our source data by utilizing deep learning, in the form of the Google DeeplabV3+ Xception derived CALFIN Neural Network. Additional post-processing techniques allow our method to achieve accurate and useful segmentation of raw Landsat subsets into Shapefile outputs. 

We provide two levels of data products. 

• Level 0 products consist of fjord boundary GeoTiff masks, the domain Shapefiles used for subsetting, a glacier names reference Shapefile, and the Landsat scene name ID lists.
• Level 1 product consists of a LineString Shapefile with 22678 features, and a Polygon Shapefile with 17771 features, grouped by glacial basin. Polygons are ocean masks that are constructed from merged calving front Line Strings, fjord boundaries, and domain boundaries. Both
  • Shapefiles share a feature schema derived from the MEaSUREs glacial terminus positions dataset (NSIDC-0642), as detailed in Table S2:
  • Properties
    • Spatial extent: 66 Greenlandic glacial basins, including Petermann, Upernavik, Rink Isbrae, Jakobshavn, Helheim, Kangerlussuaq, Kangiata Nunaata, Kong Oscar, Hayes, and other nearby basins (See External Spatial Figure)
    • Time Series: Sept. 1972 - June 2019 (See External Temporal Figure)
    • Temporal resolution: sub-seasonal
    • Spatial resolution: 30 meters
    • Spatial accuracy: <90 meters
    • Projection: EPSG:3413 (WGS 84 / NSIDC Sea Ice Polar Stereographic North)
    • Usage Example: (See External Usage Figure)

    • Shapefile Feature Schema Attribute Table

      Data Field	Description	Format (Values)
      GlacierID	Numerical ID assigned to each glacier (as derived from MEaSUREs NSIDC-0642)	# ([1, 246])
      Center_X	Mean X coordinate in EPSG:3413.	# ([-463626, 682313])
      Center_Y	Mean Y coordinate in EPSG:3413.	# ([-2821269, -906747])
      Latitude	Latitude of center.	# ([64.29, 81.24])
      Longitude	Longitude of center.	# ([-63.17, -28.21])
      QualFlag	Quality flag to indicate digitization conditions	# (0 - Manually digitized, 3 - Manually digitized, w/ L7 SCE, 10 - Automatically digitized, 13 - Automatically digitized, w/ L7 SCE. See MEaSUREs NSIDC-0642)
      Satellite	Satellite/sensor of the digitized source image	LXSS ([LM01, LC08]) See usgs.gov/faqs/what-naming-convention-landsat- collections-level-1-scenes
      Date	Date of the digitized source image	YYYY-MM-DD ([1972-09-06, 2019-06-25])
      ImageID	Source image file name.	LXSS_LLLL_PPPRRR_YYYYMMDD_yyyymmdd_CC_TX (LC08_L1TP_026006_20170702_20170715_01_T1, etc.)
      GrnlndcN	Greenlandic glacier name	NAME (New_Greenl names from Bjørk et al., 2015 database of Greenland glacier names)
      OfficialN	Officially recognized glacier name	NAME (Official_n names from Bjørk et al., 2015 database of Greenland glacier names)
      AltName	Alternative, Foreign, Old Greenlandic, or other glacier names	NAME (Foreign_na, Old_Greenl, Alternative names (Bjørk et al., 2015), or other names)
      RefName	Reference glacier name, non-authoritative names used in CALFIN to denote grouped/unnamed glaciers	NAME (New_Greenl, Official_n, Foreign_na, Old_Greenl, Alternative names (Bjørk et al., 2015), or other names)
      Author	Digitization author’s name	LastName_FirstInitial (Cheng_D)

  • level-1_shapefiles-greenland-termini.zip
    • LineString Shapefiles from 1972-2019, containing all features in one file
  • level-1_shapefiles-greenland-termini-closed.zip
    • Polygon Shapefiles from 1972-2019, containing all features in one file
  • level-1_shapefiles-domain-termini.zip
    • LineString Shapefiles from 1972-2019, separated by glacial domain
  • level-1_shapefiles-domain-termini-closed.zip
    • Polygon Shapefiles from 1972-2019, separated by glacial domain.

Reference for MEaSUREs glacial terminus positions dataset:

Joughin, I., T. Moon, J. Joughin, and T. Black. 2015, 2017. MEaSUREs Annual Greenland Outlet Glacier Terminus Positions from SAR Mosaics, Version 1. [NSIDC-0642]. Boulder, Colorado USA. NASA National Snow and Ice Data Center Distributed Active Archive Center. doi: https://doi.org/10.5067/DC0MLBOCL3EL.