Search

A multifunctional scatterometer system and optimized radar signal processing for simultaneous observation of various physical oceanographic parameters are described in this paper. Existing observation methods with microwave remote sensing techniques generally use several separate systems such as scatterometer, altimeter, and Doppler radar for sea surface monitoring, which are inefficient in system operation and cross-analysis of each observation data. To improve this point, we integrated separate measurement functions into a single observation system by adding a measurement function of Doppler frequency to the existing system. So it enables to simultaneously measure the range and polarimetric responses of backscattering as well as movements of the sea surface. Here, the simultaneous measurement function of Doppler frequency was implemented by sampling an FMCW (frequency modulated continuous wave) radar signal as 2D raw data consisting of fast- and slow-time samples, i.e., the range and backscattering of radar target signals are analyzed from the fast-time samples while the Doppler frequency by the radar target’s movement extracts from the slow-time samples. Through the Fourier transformed-based range-Doppler signal process, distance (R), backscattering (σ°), and Doppler frequency (fD) are sequentially extracted from the 2D raw data, and a correlation to the physical oceanographic parameters is analyzed. Operability of the proposed system was examed through total 3 times of field campaigns from June 2017 to August 2020 and the observation data retrieved by the radar measurement data (R, σ°, fD) was also cross-analyzed with in-situ data: e.g., tide, significant wave height, and wind speed and direction. Differences in the comparative results as an observational accuracy are as follows. Tidal level (Root Mean Square Error 0.169 m (R)), significant wave height (RMSE 0.127 m (R), 0.362 m (σ°)), wind speed (RMSE 1.880 m/s (fD), 2.094 m/s (σ°)) and direction (18.84° (fD)).

Deep learning is a promising method for image classification, including satellite images acquired by various sensors. However, the synergistic use of geospatial data for water body extraction from Sentinel-1 data using deep learning and the applicability of existing deep learning models have not been thoroughly tested for operational flood monitoring. Here, we present a novel water body extraction model based on a deep neural network that exploits Sentinel-1 data and flood-related geospatial datasets. For the model, the U-Net was customised and optimised to utilise Sentinel-1 data and other flood-related geospatial data, including digital elevation model (DEM), Slope, Aspect, Profile Curvature (PC), Topographic Wetness Index (TWI), Terrain Ruggedness Index (TRI), and Buffer for the Southeast Asia region. Testing and validation of the water body extraction model was applied to three Sentinel-1 images for Vietnam, Myanmar, and Bangladesh. By segmenting 384 Sentinel-1 images, model performance and segmentation accuracy for all of the 128 cases that the combination of stacked layers had determined were evaluated following the types of combined input layers. Of the 128 cases, 31 cases showed improvement in Overall Accuracy (OA), and 19 cases showed improvement in both averaged intersection over union (IOU) and F1 score for the three Sentinel-1 images segmented for water body extraction. The averaged OA, IOU, and F1 scores of the ‘Sentinel-1 VV’ band are 95.77, 80.35, and 88.85, respectively, whereas those of ‘band combination VV, Slope, PC, and TRI’ are 96.73, 85.42, and 92.08, showing improvement by exploiting geospatial data. Such improvement was further verified with water body extraction results for the Chindwin river basin, and quantitative analysis of ‘band combination VV, Slope, PC, and TRI’ showed an improvement of the F1 score by 7.68 percent compared to the segmentation output of the ‘Sentinel-1 VV’ band. Through this research, it was demonstrated that the accuracy of deep learning-based water body extraction from Sentinel-1 images can be improved up to 7.68 percent by employing geospatial data. To the best of our knowledge, this is the first work of research that demonstrates the synergistic use of geospatial data in deep learning-based water body extraction over wide areas. It is anticipated that the results of this research could be a valuable reference when deep neural networks are applied for satellite image segmentation for operational flood monitoring and when geospatial layers are employed to improve the accuracy of deep learning-based image segmentation.

The relative sea-level changes from tide gauges in the Korean peninsula provide essential information to understand the regional and global mean sea-level changes. Several corrections to raw tide gauge records are required to account for coastal vertical land motion (VLM), regional and local coastal variability. However, due to the lack of in-situ measurements such as leveling data and the Global Navigation Satellite System (GNSS), making precise assessments of VLM at the tide gauges is still challenging. This study aims to address the above limitation to assess the VLM in the Korean tide gauges using the time-series Interferometric Synthetic Aperture Radar (InSAR) technique. For 10 tide gauges selected in the Korean peninsula, we applied the Stanford Method for Persistent Scatterers (StaMPS)&mdash, Small Baseline Subset (SBAS) method to C-band Sentinel-1 A/B Synthetic Aperture Radar (SAR) data acquired during 2014/10&ndash, 2020/05, with the novel sequential interferograms pair selection approach to increase the slowly decorrelating filtered phase (SDFP) pixels density near the tide gauges. Our findings show that overall the tide gauges in the Korean peninsula are stable, besides the largest VLM observed at Pohang tide gauge station (East Sea) of about &minus, 26.02 mm/year, also, higher rates of uplift (&gt, 1 mm/year) were observed along the coast of Yellow Sea (Incheon TG and Boryeong TG) and higher rates of subsidence (&lt, &minus, 2 mm/year) were observed at Jeju TG and Seogwipo TG. Our approach estimates the rate of VLM at selected tide gauges with an unprecedented spatial and temporal resolution and is applicable when the in-situ and GNSS observations are not available.

Development of convolutional neural network (CNN) optimized for object detection, led to significant developments in ship detection. Although training data critically affect the performance of the CNN-based training model, previous studies focused mostly on enhancing the architecture of the training model. This study developed a sophisticated and automatic methodology to generate verified and robust training data by employing synthetic aperture radar (SAR) images and automatic identification system (AIS) data. The extraction of training data initiated from interpolating the discretely received AIS positions to the exact position of the ship at the time of image acquisition. The interpolation was conducted by applying a Kalman filter, followed by compensating the Doppler frequency shift. The bounding box for the ship was constructed tightly considering the installation of the AIS equipment and the exact size of the ship. From 18 Sentinel-1 SAR images using a completely automated procedure, 7489 training data were obtained, compared with a different set of training data from visual interpretation. The ship detection model trained using the automatic training data obtained 0.7713 of overall detection performance from 3 Sentinel-1 SAR images, which exceeded that of manual training data, evading the artificial structures of harbors and azimuth ambiguity ghost signals from detection.

Ice rumples are locally-grounded features of flowing ice shelves, elevated tens of meters above the surrounding surface. These features may significantly impact the dynamics of ice-shelf grounding lines, which are strongly related to shelf stability. In this study, we used TanDEM-X data to construct high-resolution DEMs of the Thwaites ice shelf in West Antarctica from 2011 to 2013. We also generated surface deformation maps which allowed us to detect and monitor the elevation changes of an ice rumple that appeared sometime between the observations of a grounding line of the Thwaites glacier using Double-Differential Interferometric SAR (DDInSAR) in 1996 and 2011. The observed degradation of the ice rumple during 2011–2013 may be related to a loss of contact with the underlying bathymetry caused by the thinning of the ice shelf. We subsequently used a viscoelastic deformation model with a finite spherical pressure source to reproduce the surface expression of the ice rumple. Global optimization allowed us to fit the model to the observed deformation map, producing reasonable estimates of the ice thickness at the center of the pressure source. Our conclusion is that combining the use of multiple high-resolution DEMs and the simple viscoelastic deformation model is feasible for observing and understanding the transient nature of small ice rumples, with implications for monitoring ice shelf stability.

Definite surface topography of ice provides fundamental information for most glaciologists to study climate change. However, the topography at the marginal region of ice sheets exhibits noticeable dynamical changes from fast flow velocity and large thinning rates; thus, it is difficult to determine instantaneous topography. In this study, the surface topography of the marginal region of Thwaites Glacier in the Amundsen Sector of West Antarctica, where ice melting and thinning are prevailing, is extracted using TanDEM-X interferometry in combination with data from the near-coincident CryoSat-2 radar altimeter. The absolute height offset, which has been a persistent problem in applying the interferometry technique for generating DEMs, is determined by linear least-squares fitting between the uncorrected TanDEM-X heights and reliable reference heights from CryoSat-2. The reliable heights are rigorously selected at locations of high normalized cross-correlation and low RMS heights between segments of data points. The generated digital elevation model with the resolved absolute height offset is assessed with airborne laser altimeter data from the Operation IceBridge that were acquired five months after TanDEM-X and show high correlation with biases of 3.19 m and −4.31 m at the grounding zone and over the ice sheet surface, respectively. For practical application of the generated DEM, grounding line estimation assuming hydrostatic equilibrium was carried out, and the feasibility was seen through comparison with the previous grounding line. Finally, it is expected that the combination of interferometry and altimetery with similar datasets can be applied at regions even with a lack of ground control points.