Machine Learning-Aided Sea Ice Monitoring Using Feature Sequences Extracted from Spaceborne GNSS-Reflectometry Data

Yongchao Zhu,Xiaochuan Qu,Shuiping Li,Tingye Tao,Jens Wickert,Kegen Yu,Maximilian Semmling

doi:10.3390/rs12223751

Abstract

Two effective machine learning-aided sea ice monitoring methods are investigated using 42 months of spaceborne Global Navigation Satellite System-Reflectometry (GNSS-R) data collected by the TechDemoSat-1 (TDS-1). The two-dimensional delay waveforms with different Doppler spread characteristics are applied to extract six features, which are combined to monitor sea ice using the decision tree (DT) and random forest (RF) algorithms. Firstly, the feature sequences are used as input variables and sea ice concentration (SIC) data from the Advanced Microwave Space Radiometer-2 (AMSR-2) are applied as targeted output to train the sea ice monitoring model. Hereafter, the performance of the proposed method is evaluated through comparing with the sea ice edge (SIE) data from the Special Sensor Microwave Imager Sounder (SSMIS) data. The DT- and RF-based methods achieve an overall accuracy of 97.51% and 98.03%, respectively, in the Arctic region and 95.46% and 95.96%, respectively, in the Antarctic region. The DT- and RF-based methods achieve similar accuracies, while the Kappa coefficient of RF-based approach is slightly larger than that of the DT-based approach, which indicates that the RF-based method outperforms the DT-based method. The results show the potential of monitoring sea ice using machine learning-aided GNSS-R approaches.

Highlights

Sea ice monitoring shows significant importance because it has notable impacts on the Earth’s radiation balance, which affects the global climate significantly
The TDS-1 data collected over the Arctic and Antarctic regions with the latitude above 55◦N and 55◦S from January 2015 to December 2018 are analyzed in this study
The data unavailability from August to October 2017 is probably due to the scheduled shutdown of TDS-1 mission, which was originally set to the end of July 2017

Summary

Introduction

Sea ice monitoring shows significant importance because it has notable impacts on the Earth’s radiation balance, which affects the global climate significantly. Sea ice has been monitored with various approaches, such as field observations [2], numerical models [3] and remote sensing [4], the latter of which has been considered as the most efficient approach to detect sea ice. The Global Navigation Satellite System (GNSS) can be used for positioning, navigation and timing, and for sensing geophysical parameters through analyzing GNSS signals scattered from the Earth surface. The Global Navigation Satellite System (GNSS) can be used for positioning, navigation and timing, and for sensing geophysical parameters through analyzing GNSS signals scattered from the Earth surface This innovative remote sensing technology is termed GNSS Reflectometry (GNSS-R), which has been applied to ocean altimetry [5], wind field retrieval [6,7,8], tsunami detection [9,10], soil moisture estimation [11] and oil slick detection [12,13]. The two Chinese satellites called BuFeng-1 A/B, which are part of the first Chinese GNSS-R mission, were launched on 5 June 2019 [21]

Methods

Results

Conclusion