Semi-automated tracking of iceberg B43 using Sentinel-1 SAR images via Google Earth Engine

Younghyun Koo,Alberto M Mestas-Nuñez,Grant J Macdonald,Chang-Uk Hyun,Hongjie Xie,Stephen F Ackley

doi:10.5194/tc-15-4727-2021

Younghyun Koo, Alberto M Mestas-Nuñez + Show 4 more

Open Access

https://doi.org/10.5194/tc-15-4727-2021

Copy DOI

Abstract

Abstract. Sentinel-1 C-band synthetic aperture radar (SAR) images can be used to observe the drift of icebergs over the Southern Ocean with around 1–3 d of temporal resolution and 10–40 m of spatial resolution. The Google Earth Engine (GEE) cloud-based platform allows processing of a large quantity of Sentinel-1 images, saving time and computational resources. In this study, we process Sentinel-1 data via GEE to detect and track the drift of iceberg B43 during its lifespan of 3 years (2017–2020) in the Southern Ocean. First, to detect all candidate icebergs in Sentinel-1 images, we employ an object-based image segmentation (simple non-iterative clustering – SNIC) and a traditional backscatter threshold method. Next, we automatically choose and trace the location of the target iceberg by comparing the centroid distance histograms (CDHs) of all detected icebergs in subsequent days with the CDH of the reference target iceberg. Using this approach, we successfully track iceberg B43 from the Amundsen Sea to the Ross Sea and examine its changes in area, speed, and direction. Three periods with sudden losses of area (i.e., split-offs) coincide with periods of low sea ice concentration, warm air temperature, and high waves. This implies that these variables may be related to mechanisms causing the split-off of the iceberg. Since the iceberg is generally surrounded by compacted sea ice, its drift correlates in part with sea ice motion and wind velocity. Given that the bulk of the iceberg is under water (∼30–60 m freeboard and ∼150–400 m thickness), its motion is predominantly driven by the westward-flowing Antarctic Coastal Current, which dominates the circulation of the region. Considering the complexity of modeling icebergs, there is a demand for a large iceberg database to better understand the behavior of icebergs and their interactions with surrounding environments. The semi-automated iceberg tracking based on the storage capacity and computing power of GEE can be used for this purpose.

Highlights

When a large ice mass breaks off from an ice shelf or glacier into the ocean, it forms an iceberg
This study demonstrates for the first time the potential of Google Earth Engine (GEE) for iceberg tracking in the Southern Ocean using Sentinel-1 synthetic aperture radar (SAR) images
It presents a cloud-based computational method for tracking iceberg B43 from when it calved off the Amundsen Sea coast until it broke up in the Ross Sea

Summary

Introduction

When a large ice mass breaks off from an ice shelf or glacier into the ocean, it forms an iceberg. Since the trajectories and speeds of icebergs depend on multiple and complex environmental variables (ocean, atmosphere, sea ice, etc.), icebergs provide important insights for the interaction of these variables (Schodlok et al, 2006). The formation and melting of icebergs influence global climate (Romanov et al, 2008; Mackie et al, 2020), ocean flux (Silva et al, 2006; Rackow et al, 2017; Starr et al, 2021), sea ice production (Martin et al, 2007; Merino et al, 2016), dissolved iron concentration (Lin et al, 2011; De Jong et al, 2015), and ecosystems and biology (Wilson et al, 2016; Schwarz and Schodlok, 2009; Biddle et al, 2015). Detecting and tracking icebergs is extremely important to understand the changing sea ice, ocean, and atmosphere in the polar regions

Objectives

Methods

Results

Conclusion