Sea Ice Detection Using U.K. TDS-1 GNSS-R Data

Abstract

1A sea ice detection algorithm developed using UK TDS-1 Global Navigation Satellite Systems (GNSS)- newly formed sea ice, the scattering is mostly coherent which is typical for a mostly flat surface. In order to 7 measure the degree of coherence of the received waveform or DDM three different estimators are presented: 8 the normalized DDM average, the Trailing Edge Slope (TES), and the matched filter approach. Here we present 9 a probabilistic study based on a Bayesian approach using two different and independent ground-truth datasets. 10This approach allow us thoroughly assessing the performance of the estimators. The best results are achieved 11 for both the TES and the matched filter approach with a probability of detection of 97%, a probability of false 12 alarm of ∼ 2%, and a probability of error of 2.5%. However, the matched filter approach is preferred due to 13 its simplicity. The measurement of the sea ice concentration is also assessed in this work, but the nature of the 14 UK TDS-1 data (lack of calibrated data) does not allow us to make any specific conclusions about the sea ice 15 concentration. 16 Index Terms 17Sea Ice, coherent scattering, incoherent scattering, GNSS-R, UK TDS-1 18

Sea Ice Detection Using GNSS‐R Data From TechDemoSat‐1

“…This was calculated in all four directions on the DDM (up and down: dy ; left and right: dx ); dy in the down direction can be considered another method of calculation of the TES. DDMA (DDM average) . As used by Alonso‐Arroyo et al (). The average value in a box around the maximum amplitude for the box sizes below.…”

“…Previous studies (Alonso‐Arroyo et al, ; Rivas et al, ; Yan et al, ; Yan & Huang, ; Zhu et al, ) have been shown to correctly distinguish between sea ice and open ocean in up to 98.4% of cases in the detection of sea ice compared to colocated passive microwave data. However, these studies were limited in scope, examining small subsets of TDS‐1 observations only.…”

“…A possible cause of these differences in detection is the various sensitivities of the different sensors to the sea ice and the differing behavior of microwave emissions from the surface captured by passive microwave radiometers and used in the CCI product as compared to the L‐Band reflections used by GNSS‐R. It is believed that the GNSS‐R sensors may be more sensitive to the presence of sea ice (Alonso‐Arroyo et al, ), and it is for this reason that the 0% concentration threshold for sea ice assignment was used in training the algorithm. The use of the more widely used 15% concentration threshold with these parameters yielded agreement of 97.7% and 95.4% with the CCI product in the Antarctic and Arctic respectively.…”

First spaceborne phase altimetry over sea ice using TechDemoSat‐1 GNSS‐R signals

“…However, carrier phase information can be only retrieved from coherent reflections of GNSS signal. The DDMs collected over sea ice have been also analyzed for sea ice detection and sea ice concentration retrieval [e.g., Yan and Huang, 2016;Yan et al, 2017;Alonso-Arroyo et al, 2017]. The presence of sea ice at the water surface significantly shifts the diffuse reflection limit and improves the phase coherence of L band observations.…”

Measuring Greenland Ice Sheet Melt Using Spaceborne GNSS Reflectometry From TechDemoSat‐1

“…In contrast, diffuse scattering from the open ocean can significantly reduce the SNR. The use of GNSS-R power to detect and characterize sea ice has been widely studied (e.g., Alonso-Arroyo et al, 2017;Cartwright et al, 2019;Nghiem et al, 2019;Yan & Huang, 2016), and it is not further investigated here. The DDMs collected over the GrIS show moderate SNR, and it is hypothesized that the reflected signal from the GrIS is a combination of coherent, volumetric, and incoherent scattering components.…”