U.S. Department of Energy Office of Biological and Environmental Research

BER Research Highlights


Machine-Learning-Based Measurement of Ice Wedge Polygon Properties
Published: January 21, 2019
Posted: April 22, 2019

A rapid assessment of wet-tundra arctic landscape conditions.

The Science
Ice wedge polygons are ubiquitous features in wet-tundra Arctic landscapes. Their topographic properties control the distribution of water, vegetation and biogeochemical processes. Measuring and counting these small-scale landscape features across the Arctic is an extremely difficult proposition, but necessary to assess the state and dynamics of the landscape. A new machine learning approach can now quickly quantify these small-scale features at regional scales, and enables improved estimates of ecosystem processes across large swaths of the Arctic landscape.

The Impact
This new capability now enables scientists to quickly assess the number, configuration, and state of ice wedge polygons across large swaths of the Arctic. With this technology, scientists will be able to quickly measure how these land forms are responding to rapid arctic warming and concurrent permafrost degradation that is reshaping local to regional topography. Products from this technology are informing models to project how changes in the structure of Arctic landscapes will influence feedbacks to the climate system.

Summary
Ice wedge polygons are the surface expression of ice wedges, or vertical veins of ground ice that divide tundra landscapes into a network of polygonal units, 10-30 m across. These polygons pervade the Arctic tundra and are categorized as low centered polygons, which are surrounded by rims of soil several tens of centimeter high, or high centered polygons, surrounded by troughs on the order of a meter deep. The spatial distribution of these two types of polygon controls important landscape processes, including redistribution of windblown snow, thermal regulation of the underlying permafrost, runoff and evaporation, and surface emissions of two important but very different greenhouse gasses, carbon dioxide and methane. Therefore, mapping polygon types across the Arctic is vital for understanding the hydrologic function of landscapes, as well as potential fluxes of carbon into the atmosphere. However, directly delineating each polygon across the Arctic is impractical. Scientists at the University of Texas in collaboration with Los Alamos National Laboratory have developed a new approach that utilizes machine learning algorithms to analyze high resolution digital elevation maps from airborne remote sensing. This approach has been shown to be fast and accurate at two test sites with complex polygonal terrain, near Prudhoe Bay and Utqiagvik (formerly Barrow), Alaska. The algorithm allows scientists to quickly and accurately inventory polygonal forms across broad tundra landscapes, which will ultimately inform projections of the fate of the large stock of organic matter stored in Arctic soils.

Contacts (BER PM)
Daniel Stover
SC-23.1
Terrestrial Ecosystem Science
Daniel.Stover@science.doe.gov

(PI Contacts)
Chuck Abolt
The University of Texas at Austin
chuck.abolt@utexas.edu

Adam Atchley
Los Alamos National Laboratory
Aatchley@lanl.gov

Funding
Funding is provided by DOE Biological and Environmental Research, Terrestrial Ecosystem Science, Next Generation Ecosystem Experiments Arctic (NGEE-Arctic) and NASA Earth and Space Science Fellowship program.

Publications
Abolt, C.J., M.H. Young, A.L. Atchley, and C.J. Wilson. “Brief communication: Rapid machine-learning-based extraction and measurement of ice wedge polygons in high-resolution digital elevation models.” The Cryosphere, 13(1), 234-245 (2019). [DOI: 10.5194/tc-13-237-2019]

Related Links
Code repository:
Abolt, C.J., M.H. Young, A.L. Atchley, and C.J. Wilson. “CNN-watershed: A machine learning-based tool for delineation and measurement of ice wedge polygons in high-resolution digital elevation models.” Zenodo repository. [DOI: 10.5281/zenodo.2554542]

Topic Areas:

  • Research Area: Terrestrial Ecosystem Science
  • Research Area: Next-Generation Ecosystem Experiments (NGEE)

Division: SC-23.1 Climate and Environmental Sciences Division, BER

 

BER supports basic research and scientific user facilities to advance DOE missions in energy and environment. More about BER

Recent Highlights

May 10, 2019
Quantifying Decision Uncertainty in Water Management via a Coupled Agent-Based Model
Considering risk perception can improve the representation of human decision-making processes in age [more...]

May 09, 2019
Projecting Global Urban Area Growth Through 2100 Based on Historical Time Series Data and Future Scenarios
Study provides country-specific urban area growth models and the first dataset on country-level urba [more...]

May 05, 2019
Calibrating Building Energy Demand Models to Refine Long-Term Energy Planning
A new, flexible calibration approach improved model accuracy in capturing year-to-year changes in bu [more...]

May 03, 2019
Calibration and Uncertainty Analysis of Demeter for Better Downscaling of Global Land Use and Land Cover Projections
Researchers improved the Demeter model’s performance by calibrating key parameters and establi [more...]

Apr 22, 2019
Representation of U.S. Warm Temperature Extremes in Global Climate Model Ensembles
Representation of warm temperature events varies considerably among global climate models, which has [more...]

List all highlights (possible long download time)