Regional Assessments of Surface Ice Elevations from Swath-Processed CryoSat-2 SARIn Data

• Published in: Remote Sensing
• Main Authors: Natalia Havelund Andersen, Sebastian Bjerregaard Simonsen, Mai Winstrup, Johan Nilsson, Louise Sandberg Sørensen
• Format: Text
• Language: English
• Published: Multidisciplinary Digital Publishing Institute 2021
• Subjects: swath processing; ice elevations; CS2; validation; Arctic; Climate change; ice covered areas
• Online Access: https://doi.org/10.3390/rs13112213

The Arctic responds rapidly to climate change, and the melting of land ice is a major contributor to the observed present-day sea-level rise. The coastal regions of these ice-covered areas are showing the most dramatic changes in the form of widespread thinning. Therefore, it is vital to improve the monitoring of these areas to help us better understand their contribution to present-day sea levels. In this study, we derive ice-surface elevations from the swath processing of CryoSat-2 SARIn data, and evaluate the results in several Arctic regions. In contrast to the conventional retracking of radar data, swath processing greatly enhances spatial coverage as it uses the majority of information in the radar waveform to create a swath of elevation measurements. However, detailed validation procedures for swath-processed data are important to assess the performance of the method. Therefore, a range of validation activities were carried out to evaluate the performance of the swath processor in four different regions in the Arctic. We assessed accuracy by investigating both intramission crossover elevation differences, and comparisons to independent elevation data. The validation data consisted of both air- and spaceborne laser altimetry, and airborne X-band radar data. There were varying elevation biases between CryoSat-2 and the validation datasets. The best agreement was found for CryoSat-2 and ICESat-2 over the Helheim region in June 2019. To test the stability of the swath processor, we applied two different coherence thresholds. The number of data points was increased by approximately 25% when decreasing the coherence threshold in the processor from 0.8 to 0.6. However, depending on the region, this came with the cost of an increase of 33–65% in standard deviation of the intramission differences. Our study highlights the importance of selecting an appropriate coherence threshold for the swath processor. Coherence threshold should be chosen on a case-specific basis depending on the need for enhanced spatial ...