NSIDC Cryosphere Seminar
Mapping Abrupt Permafrost Thaw from Space Using Machine Learning by Dr. Lingcao Huang, CIRES
Permafrost is undergoing strong warming and thawing in recent decades, which can potentially cause unprecedented environmental consequences as it stores a large amount of carbon and underlays about 25% of the exposed land in the northern hemisphere. Abrupt thawing of ice-rich permafrost may result in landforms such as retrogressive thaw slumps and thermo-erosion gullies on the Earth’s surface. As reported by many local studies, the occurrence and frequency of abrupt permafrost thaw have increased significantly. However, its spatial distribution and development in most permafrost areas are poorly quantified, which hinders the understanding of permafrost degradation in large areas. The reason for this is the difficultly of analyzing big data and extracting small and subtle features on images. In the past decades, the accumulation of satellite data and the advance of machine learning algorithms, especially convolutional neural networks, make it possible to tackle this problem. In this talk, I will present my research of using high-resolution () satellite data and machine learning algorithms to map these landforms in Tibet and Alaska north slope.
Lingcao Huang is a CIRES visiting Postdoctoral Fellow at the University of Colorado Boulder. His current research focuses on mapping the spatial distribution and development of ice-rich permafrost abrupt thawing in Alaska north slope from various remote sensing data using machine learning. He received Bachelor and Master degrees in remote sensing & photogrammetry at Wuhan University and a Doctoral degree in earth and atmospheric sciences at the Chinese University of Hong Kong.