|Paper title||From glacier to icebergs|
|Form of presentation||Poster|
The volume of freshwater (solid and liquid) exported from glaciers to the ocean is important for the global climate system, as an increase in the freshwater content can slow down the large-scale thermohaline circulation and change the mass balance of the glaciers. The solid part of the freshwater is exported as icebergs that breaks off the marine terminating glaciers. Because of this, the iceberg density around Greenland is linked to the glacial surface velocities. As a direct consequence of the climate response, Arctic sea ice has experienced a rapid reduction in extent and thickness in recent decades, opening up the opportunity for increased shipping activities in the Arctic and and adjacent Seas. The increase in shipping is expected to continue as the ice-free season becomes longer, and new routes opens up. Icebergs are a large hazard to ships, especially in the near coastal areas of Greenland. Thus, detection of icebergs in all sizes, even growlers and bergy bits is important.
This presentation aims at linking the observed outflow of glacial ice to the total iceberg volume and based on this predict the iceberg density and the solid freshwater contribution from the glaciers. Here we will compare glacial outflow from observations of ice surface velocities through a defined fluxgate near Upernavik, which is located in the north-western Greenland and compare it to the iceberg volumes estimated from Copernicus Sentinel-1 SAR images using the Danish Meteorological Institute iceberg detection algorithm CFAR and high-resolution SPOT images using a semi-automatic classification algorithm.
The flux of glacial ice is calculated using a processor developed within the Polar Thematic Exploitation platform (P-TEP). It estimates time-series of glacial ice fluxes through a pre-defined fluxgate at a selected glacier, given a user defined input of surface velocity (vsurf). The solid ice discharge through a flux gate (F) of length (L) and ice thickness (H) is given by F=f*vproj*H*L , where f=0.93 is the mean ratio of surface to depth-averaged velocity and vproj is vsurf projected to the gate perpendicular velocity. We here use Morlighem bedrock model to estate the ice thickness at the fluxgate, and PROMICE and MEaSURE’s glacial surface velocities based on Copernicus Sentinel-1 SAR images.
The iceberg detection method based on the Copernicus Sentinel-1 SAR images is the so-called CFAR (Constant False Alarm Rate). This method detects the icebergs based on an assumed background intensity defined by the backscatter and an assumption that targets (icebergs) are detected as a backscatter value with a signal above this background intensity. The procedure utilized by the CFAR algorithm examines each pixel in the SAR imagery using a “sliding window”. The pixel in question is the pixel in the centre of the window and the background is represented by the window’s outer edge of pixels. The statistical distribution (Probability Density Function [PDF]) of the edge pixels is derived, and if the pixel in question is extremely unlikely to belong to the background intensity it will be classified as a target pixel.
The iceberg detection algorithm is known to be challenged within near coastal areas for several reasons. It overestimates the iceberg area and its volume, in particular for smaller icebergs, due to the relative coarse resolution of the SAR images. At the same time small growlers and bergy bits are not captured by the CFAR algorithm. At last in near coastal areas where the volume and number of iceberg covered pixels are large the background intensity will include icebergs, which then remain undetected.
For this reason, we will validate the CFAR algorithm with high resolution SPOT Images, and quantify the exported freshwater based on both methods. This will be useful not only for this study but also for estimates of the volume of ice that is not detected by the CFAR algorithm.
To summarize this presentation will:
1/ provide estimates of the solid freshwater discharge based on a fluxgate near Upernavik glacier outflow
2/ Correlate the high resolution SPOT images with the CFAR detections of icebergs near Upernavik.
3/ Compare volumes of ice based on the freshwater discharge to iceberg volumes based on estimates from both SPOT images and the CFAR algorithm.