AI4SeaIce: Toward Solving Ambiguous SAR Textures in Convolutional Neural Networks for Automatic Sea Ice Concentration Charting
Open Access
- 7 February 2022
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 60 (01962892) , 1-13
Abstract
Automatically producing arctic sea ice charts from Sentinel-1 Synthetic Aperture Radar (SAR) images is challenging for Convolutional Neural Networks (CNN) due to ambiguous backscattering signatures. The number of pixels viewed by the CNN model in the input image used to generate an output pixel, or the receptive field, is important to detect large features or physical objects such as sea ice and correctly classify them. In addition, a noise phenomenon is present in the Sentinel-1 ESA Instrument Processing Facility (IPF) v2.9 SAR data, particularly in the subswath transitions, visible as long vertical lines and grained particles resembling small sea ice floes. To overcome these two challenges, we suggest adjusting the receptive field of the popular U-Net CNN architecture used for semantic segmentation. It is achieved by symmetrically adding additional blocks of convolutional, pooling and upsampling layers in the encoder and decoder of the U-Net, constituting an increase in the number of levels. This show great improvements in performance and in the homogeneity of predictions. Secondly, training models on SAR data noise corrected with an enhanced technique has demonstrated a significant increase in model performance, and enabled better predictions in uncertain regions. An 8-level U-Net trained on the alternative noise corrected SAR data is presented capable of correctly predicting many ambiguous SAR signatures, and increased performance by 8.44% points compared to the regular U-Net trained on the ordinary ESA IPF v2.9 noise corrected SAR data. This is the first installment of this multi-series installment of articles related to AI applied to sea ice (in short AI4SeaIce).Keywords
This publication has 16 references indexed in Scilit:
- Textural Noise Correction for Sentinel-1 TOPSAR Cross-Polarization Channel ImagesIEEE Transactions on Geoscience and Remote Sensing, 2019
- Efficient Thermal Noise Removal for Sentinel-1 TOPSAR Cross-Polarization ChannelIEEE Transactions on Geoscience and Remote Sensing, 2017
- Melting ice Caps and the Economic Impact of Opening the Northern Sea RouteThe Economic Journal, 2017
- DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFsIEEE Transactions on Pattern Analysis and Machine Intelligence, 2017
- Sea Ice Concentration Estimation during Freeze-Up from SAR Imagery Using a Convolutional Neural NetworkRemote Sensing, 2017
- Sea Ice Concentration Estimation During Melt From Dual-Pol SAR Scenes Using Deep Convolutional Neural Networks: A Case StudyIEEE Transactions on Geoscience and Remote Sensing, 2016
- Predicting Depth, Surface Normals and Semantic Labels with a Common Multi-scale Convolutional Architecture2015
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015
- A Comparison Between High-Resolution EO-Based and Ice Analyst-Assigned Sea Ice ConcentrationsIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015
- Sentinel-1 system overview and performance2012