Identifying Thermokarst Lakes in the Qinghai-Tibetan Plateau Using Discrete Wavelet Transform-Based Deep Learning By Andrew Li, Jiahe Liu, Olivia Liu, Xiaodi Wang
Published in IbPRIA 2023: Pattern Recognition and Image Analysis
Included in Springer's Lecture Notes in Computer Science book series (LNCS, Volume 14062)
Abstract: One of climate change’s less-visible consequences is rapid permafrost thaw (Lara, 2022). Not only does thermokarst terrain serve as a prominent indicator of permafrost thaw, but landforms such as thermokarst lakes emit significant amounts of greenhouse gases (GHGs), specifically carbon dioxide (CO2) and methane (CH4) (Zandt, Liebner, & Welte, 2020). Acquiring accurate predictions of these landforms would inform more precise global climate models. However, due to the diversity of their characteristics and the difficulty associated with collecting widespread field data, thermokarst terrain can be difficult to classify. To combat these difficulties, in this research, we explore the degradation of permafrost in the Qinghai-Tibetan Plateau (QTP) region through the classification of thermokarst lakes and develop a discrete wavelet transform (DWT) based dual input deep learning (DL) model with a convolutional neural network (CNN) to automatically classify and accurately predict thermokarst lakes with area between 0.2 and 0.5 km2, a range of lakes previously excluded from many assessments due to issues in satellite data. We created a new 3-way tensor dataset based on raw image data for more than 500 Sentinel-2 satellite lake images and decomposed those images using state-of-the-art M-band DWTs. We also incorporated non-image feature data for various climate variables. These methods significantly improved our model’s performance.
Cite: Li, A., Liu, J., Liu, O., Wang, X. (2023). Identifying Thermokarst Lakes Using Discrete Wavelet Transform–Based Deep Learning Framework. In: Pertusa, A., Gallego, A.J., Sánchez, J.A., Domingues, I. (eds) Pattern Recognition and Image Analysis. IbPRIA 2023. Lecture Notes in Computer Science, vol 14062. Springer, Cham. https://doi.org/10.1007/978-3-031-36616-1_38