Marine X-band radar has been suggested to be capable of monitoring significant wave heights in both offshore and open sea areas. In contrast to studies on offshore radar, significant wave height estimations from coastal radar images, which exhibit complicated radar backscattering features, have received little attention. This study proposes a method for retrieving the significant wave height from coastal areas that are often influenced by nononshore winds. The square root of the signal-to-noise ratio in radar images has been widely applied to estimate the significant wave height. However, nononshore wind cases show a poor correlation between the square root of the signal-to-noise ratio and the in situ significant wave height. In addition, the spectral shapes from radar images in nononshore wind cases are very different from those in onshore wind cases. To improve the significant wave height estimations from coastal radar images, we implement an artificial neural network algorithm. After training and testing the algorithm, we confirm that the estimated significant wave heights are more reliable for both onshore and nononshore wind cases if the square root of the signal-to-noise ratio, power from nearshore radar subimages, and in situ wind components are included in the input layer of the neural network.
6 March 2017 in IEEE Geoscience and Remote Sensing Letters
This letter presents an image processing technique based on the theory of the Hilbert transform to determine the coastal bathymetry from marine radar image sequences. Use of the Hilbert transform enables the difficulties and complications of inhomogeneous image analysis to be avoided. In addition, a number of steps and complex computations can be avoided using the numerical algorithm of the Hilbert transform. Because the Hilbert transform is suitable for only monocomponent signals, we first applied the image procedure to decompose the irregular wave patterns into different single wave period images. Analysis of the simulated wave field and radar image sequences demonstrated that there was a high correlation between the estimated depths and reference depths. The causes for errors in the bathymetry estimations are also discussed.
This study aims to establish a comprehensive workflow for developing emergency response plans for both actual and scenario oil spill incidents in the Taiwan waters while addressing the resource allocation for oil spill containment as well. This workflow comprises two vital components. The first component involves the integration of numerical tools and observational data, which includes the incorporation of wind data from sources such as the National Centers for Environmental Prediction (NCEP) or meteorological stations. Additionally, it incorporates ocean current data simulated by the semi-implicit cross-scale hydroscience integrated system model (SCHISM) into the general NOAA operational modeling environment (GNOME) model, which is a new approach for this purpose. In order to assess the efficacy of this component, two distinct case studies were conducted. The first case study focused on an incident in a northern coastal area of Taiwan under open sea conditions, whereas the second case study examined an incident within a major commercial harbor in central Taiwan. The second component of this workflow involves creating oil risk maps by integrating the results from the first component with specific geographical factors into Google Earth. These oil risk maps serve multiple purposes. They offer real-time information to emergency response commanders regarding oil spill hazard prediction, and they also enable the effective development of emergency response strategies and disposal plans for potential oil spill incidents. This is achieved by generating risk maps for various scenarios using the approach outlined in the first component. Additionally, these maps assist in the assessment and planning of resource allocation for oil containment.
Remotely sensed images of the ocean are a useful tool for revealing wave direction information in the spatial domain. Considering wave inhomogeneities in a coastal area, this paper addresses issues concerning the analysis of local wave directions using a simultaneous time series of X-band radar echo intensities. The method employed to analyze the directional wave information in our paper is based on a cross-spectral analysis of radar echo intensity time series from several neighboring locations constituting a spatial array within a small area. Compared with simultaneous sea surface elevation records, which have demonstrated their efficacy for estimating directional wave information, the spatial wave features present in radar image sequences are more sensitive to environmental factors and image noise. To confirm the practicality of this approach, we examined the relationship between the wave directions estimated from the X-band radar and moored buoy data. Most mean wave directions and dominant wave directions estimated from the radar images are consistent with the buoy results. In most cases, the differences between the radar and buoy wave directions are within ±30°. We also present a radar image case with an obvious refracted wave pattern. The results estimated from this image case demonstrate the feasibility of determining the spatial pattern of wave directions from a highly inhomogeneous wave field.
8 August 2024 in IEEE Transactions on Geoscience and Remote Sensing
The potential of marine X-band radar images for estimating sea surface currents in open sea areas was confirmed. However, when spatial current estimation was performed using radar images of the semi-enclosed bay areas, the challenge of dealing with inhomogeneity became apparent. Our study employed the method of the local phase gradient to estimate the spatial currents in bay areas. The feasibility of using the phase gradient method for current estimation has been demonstrated in various studies; however, for application of this method in highly inhomogeneous bay waters, we conducted a review of the various parameters that needed to be determined via the algorithm. Through comparison with actual measurement data, we confirmed that the local phase gradient method was more suitable than the traditional method for estimating sea currents within bay areas. Spatial flood and ebb currents obtained from radar images revealed the presence of eddies within the bay. We also verified the significant relationship between spatial current inhomogeneity and tidal phases in the study area.
A phase gradient (PG)-based algorithm is proposed in this study to determine coastal bathymetry from X-band radar images. Although local wavenumbers with the same spatial resolution of the wave field can be obtained from the wave field using the PG method, only a single wavenumber result can be extracted from each location theoretically. Due to the influence of unavoidable noise on the wave field image, single wavenumber estimation often shows high uncertainty. This study combines a bandpass filter and directional pass filter to produce different nearly monocomponent wave fields from X-band radar images and then estimates more wavenumbers from these wave fields using the PG method. However, the distributions of wavenumbers in higher-frequency bins still show high variance because the strength of wave signals is weak. We confirmed that the uncertain wavenumber–frequency pairs can be improved using the Kalman filter and are more consistent with the dispersion relation curve. To decrease the influence of inaccurate wavenumbers, we also use the strength of the wave signals as the weights for the least-squares fit. Although the depth errors from shallow-water areas are still unavoidable, we can remove the inaccurate depth estimation from shallow-water areas according to the coefficients of determination of the fitting. In summary, the algorithm proposed in this study can obtain a bathymetry map with high spatial resolution. In contrast to the depth result estimated using a single wavenumber of each frequency bin, we confirm that more wavenumbers from each of the frequency bins are helpful in fitting the dispersion relation curve and obtaining a more reliable depth result.