Validation of the 2003 Antartic Groundling Line through the use of ENVI

Members: Joyce Bevins, Robyn Evans, Micheal Jefferson, and Ryan Lawerence
Mentor: Dr. Malcolm LeCompte
Key Words: ENVI, GloVis, ground control points, grounding line, hinge point, Landsat 7 ETM+, LIMA, PIG

Dynamics and mass balance of an ice sheet can be derived from an accurate measurement of its area. To measure the area of a continental ice sheet, the grounding line must first be accurately determined. The grounding line is the boundary between the ‘grounded’ ice resting on land and any associated floating ice comprising a retaining ice shelf. 

During a project entitled Antarctic Surface Accumulation and Ice Discharge or ASAID, Dr. Robert Bindschadler, lead an international team of glaciologists and computer scientists, including ECSU students, in an effort to obtain a more accurate measure of the area of the Antarctic ice sheet and determine its mass balance. That is, whether the amount of ice is growing or diminishing over long time intervals. Bindschadler’s team determined the grounding line using methods of photoclinometry with LANDSAT Enhanced Thematic Mapper (ETM) image brightness and surface elevation data from the Geoscience Laser Altimeter System (GLAS) aboard NASA’s Ice, Land and Cloud Elevation Satellite (ICESat). The ASAID grounding line (GL) was established using LANDSAT 7 and GLAS data obtained in 2003. However its accuracy and utility had not been tested.  

With the current ASAID 2003 Grounding Line (GL), the CERSER GL Validation Team was tasked by Dr. Bindschadler with determining its accuracy in two coastal regions and whether changes have occurred over long time intervals.  The team over-laid the 2003 GL on LANDSAT Seven ETM imagery temporally proximate to 2003. This modified image was then compared to decades older LANDSAT 4 & 5 Thematic Mapper (TM) imagery. GL validation and change determination were planned for two geographic areas known to exhibit rapid changes potentially due to climate warming: Pine Island Glacier (PIG) and Larsen Ice Shelf. However, due to time constraints, the team only examined a limited portion of the PIG.

The GL was tested along a portion of the Antarctic coast near the PIG. To accomplish the validation, LANDSAT 7 images from 2003 used in creating the GL, were obtained from the USGS archive (lima.usgs.gov).  Other LANDSAT images were obtained from the USGS GLOVIS on-line archive (glovis.usgs.gov). The oldest possible, cloud-free LANDSAT 4 and 5 TM images were obtained for the regions of interest. To facilitate data manipulation and image comparisons, the extremely large GL vector file, obtained from Dr. Bindschadler. was truncated to include only the geographic regions of interest. Truncation and image comparisons were accomplished using ITT Visualization System’s ENVI image processing software. Any departure from perfect geographic pixel registration was corrected by using the 2003 image as a reference and then warping the older image to conform to the common fixed control points visible on both images. The grounding line overlying the 2003 image was then examined and compared to the older image. The geographic coordinates and extent of any departures from coincidence were recorded and reported.

A possible deviation in the GL was found while comparing a 2001 LANDSAT 7 image to a 1986 LANDSAT 5 image, near a small glacier feeding into Pine Island Bay.  Comparison with a 2003 image of the same area revealed no GL inaccuracy; however a small ice shelf appeared to have progressively diminished over time until it disappeared in 2003.

Study Area :: Pine Island Glacier