TY - JOUR
T1 - Semiautomatic mapping of permafrost in the Yukon Flats, Alaska
AU - Gulbrandsen, Mats Lundh
AU - Minsley, Burke J.
AU - Ball, Lyndsay B.
AU - Hansen, Thomas Mejer
N1 - Publisher Copyright:
©2016. American Geophysical Union. All Rights Reserved.
PY - 2016/12/16
Y1 - 2016/12/16
N2 - Thawing of permafrost due to global warming can have major impacts on hydrogeological processes, climate feedback, arctic ecology, and local environments. To understand these effects and processes, it is crucial to know the distribution of permafrost. In this study we exploit the fact that airborne electromagnetic (AEM) data are sensitive to the distribution of permafrost and demonstrate how the distribution of permafrost in the Yukon Flats, Alaska, is mapped in an efficient (semiautomatic) way, using a combination of supervised and unsupervised (machine) learning algorithms, i.e., Smart Interpretation and K-means clustering. Clustering is used to sort unfrozen and frozen regions, and Smart Interpretation is used to predict the depth of permafrost based on expert interpretations. This workflow allows, for the first time, a quantitative and objective approach to efficiently map permafrost based on large amounts of AEM data.
AB - Thawing of permafrost due to global warming can have major impacts on hydrogeological processes, climate feedback, arctic ecology, and local environments. To understand these effects and processes, it is crucial to know the distribution of permafrost. In this study we exploit the fact that airborne electromagnetic (AEM) data are sensitive to the distribution of permafrost and demonstrate how the distribution of permafrost in the Yukon Flats, Alaska, is mapped in an efficient (semiautomatic) way, using a combination of supervised and unsupervised (machine) learning algorithms, i.e., Smart Interpretation and K-means clustering. Clustering is used to sort unfrozen and frozen regions, and Smart Interpretation is used to predict the depth of permafrost based on expert interpretations. This workflow allows, for the first time, a quantitative and objective approach to efficiently map permafrost based on large amounts of AEM data.
KW - airborne electromagnetic data
KW - machine learning
KW - permafrost mapping
UR - http://www.scopus.com/inward/record.url?scp=85006476412&partnerID=8YFLogxK
U2 - 10.1002/2016GL071334
DO - 10.1002/2016GL071334
M3 - Article
AN - SCOPUS:85006476412
SN - 0094-8276
VL - 43
SP - 12,131-12,137
JO - Geophysical Research Letters
JF - Geophysical Research Letters
IS - 23
ER -