Forecasting of surface current velocities using ensemble machine learning algorithms for the Guangdong−Hong Kong−Macao Greater Bay area based on the high frequency radar data

Forecasting of ocean currents is critical for both marine meteorological research and ocean engineering and construction.Timely and accurate forecasting of coastal current velocities offers a scientific foundation and decision support for multiple practices such as search and rescue,disaster avoidance and remediation,and offshore construction.This research established a framework to generate short-term surface current forecasts based on ensemble machine learning trained on high frequency radar observation.Results indicate that an ensemble algorithm that used random forests to filter forecasting features by weighting them,and then used the AdaBoost method to forecast can significantly reduce the model training time,while ensuring the model forecasting effectiveness,with great economic benefits.Model accuracy is a function of surface current variability and the forecasting horizon.In order to improve the forecasting capability and accuracy of the model,the model structure of the ensemble algorithm was optimized,and the random forest algorithm was used to dynamically select model features.The results show that the error variation of the optimized surface current forecasting model has a more regular error variation,and the importance of the features varies with the forecasting time-step.At ten-step ahead forecasting horizon the model reported root mean square error,mean absolute error,and correlation coefficient by 2.84 cm/s,2.02 cm/s,and 0.96,respectively.The model error is affected by factors such as topography,boundaries,and geometric accuracy of the observation system.This paper demonstrates the potential of ensemble-based machine learning algorithm to improve forecasting of ocean currents.

Wave height estimation using the singular peaks in the sea echoes of high frequency radar

The popular methods to estimate wave height with high-frequency（HF） radar depend on the integration over the second-order spectral region and thus may come under from even not strong external interference. To improve the accuracy and increase the valid detection range of the wave height measurement, particularly by the smallaperture radar, it is turned to singular peaks which often exceed the power of other frequency components. The power of three kinds of singular peaks, i.e., those around ±1,±√2 and ±1√2 times the Bragg frequency, are retrieved from a one-month-long radar data set collected by an ocean state monitoring and analyzing radar,model S（OSMAR-S）, and in situ buoy records are used to make some comparisons. The power response to a wave height is found to be described with a new model quite well, by which obvious improvement on the wave height estimation is achieved. With the buoy measurements as reference, a correlation coefficient is increased to 0.90 and a root mean square error（RMSE） is decreased to 0.35 m at the range of 7.5 km compared with the results by the second-order method. The further analysis of the fitting performance across range suggests that the peak has the best fit and maintains a good performance as far as 40 km. The correlation coefficient is 0.78 and the RMSE is 0.62 m at 40 km. These results show the effectiveness of the new empirical method, which opens a new way for the wave height estimation with the HF radar.

Wave height measurement in the Taiwan Strait with a portable high frequency surface wave radar

As an important equipment for sea state remote sensing, high frequency surface wave radar （HFSWR） has received more and more attention. The conventional method for wave height inversion is based on the ratio of the integration of the second-order spectral continuum to that of the first-order region, where the strong external noise and the incorrect delineation of the first- and second-order Doppler spectral regions due to spectral aliasing are two major sources of errors in the wave height. To account for these factors, two more indices are introduced to the wave height estimation, i.e., the ratio of the maximum power of the second-or- der continuum to that of the Bragg spectral region （RSCB） and the ratio of the power of the second harmonic peak to that of the Bragg peak （RSHB）. Both indices also have a strong correlation with the underlying wave height. On the basis of all these indices an empirical model is proposed to estimate the wave height. This method has been used in a three-months long experiment of the ocean state measuring and analyzing ra- dar, type S （OSMAR-S）, which is a portable HFSWR with compact cross-loop/monopole receive antennas developed by Wuhan University since 2006. During the experiment in the Taiwan Strait, the significant wave height varied from 0 to 5 m. The significant wave heights estimated by the OSMAR-S correlate well with the data provided by the Oceanweather Inc. for comparison, with a correlation coefficient of 0.74 and a root mean square error （RMSE） of 0.77 m. The proposed method has made an effective improvement to the wave height estimation and thus a further step toward operational use of the OSMAR-S in the wave height extraction.

Surface tidal currents in the open sea area to the east of the Zhoushan Islands measured with high frequency surface wave radar

Based on the quasi-harmonic analysis of 11 d vector ocean currents obtained from two high frequency sur- face wave radars located at Zhujiajian Island and Shengshan Island, the spatial distribution characteristics of surface tidal currents in the open sea area to the east of the Zhoushan Islands of Zhejiang Province, China are studied. The following conclusions are drawn from the analysis： the tidal current pattern in the open sea area to the east of Zhoushan Islands is primarily regular semidiurnal, which is significantly affected by the shallow water constituents. The directions of the major axes of tidal current ellipses of M2 lie approx- imately in the NW-SE direction. With the increasing of distance away from the coast, the directions of the tidal current ellipses gradually shift toward the E-W direction. The tidal currents are mainly reversing cur- rents. The spatial distribution of probable maximum current velocities decreases gradually from northeast to southwest which is basically in accordance with the spatial distribution of measured maximum current velocities. The residual currents near the coast are larger than those far away from the coast. The directions of the residual currents are basically north by east, and the angle to the due north increases gradually with the increasing distance away from the coast. The topography shows a certain impact on the spatial distri- bution of shallow water constituents, the rotation of tidal currents, the probable maximum currents and the residual currents.

Comparison and error analysis of remotely measured waveheight by high frequency ground wave radar

High frequency ground wave radar （HFGWR） has unique advantage in the survey of dynamical factors, such as sea surface current, sea wave, and sea surface wind in marine conditions in coastal sea area. Compared to marine satellite remote sensing, it involves lower cost, has higher measuring accuracy and spatial resolution and sampling frequency. High frequency ground wave radar is a new land based remote sensing instrument with superior vision and greater application potentials. This paper reviews the development history and application status of high frequency wave radar, introduces its remote-sensing principle and method to inverse offshore fluid, and wave and wind field. Based on the author＇s ＂863 Project＂, this paper recounts comparison and verification of radar remote-sensing value, the physical calibration of radar-measured data and methods to control the quality of radar-sensing data. The authors discuss the precision of radar-sensing data＇s inversing on offshore fluid field and application of the assimilated data on assimilation.

Contrast validation test for retrieval method of high frequency ground wave radar

In this paper, on the basis of the working principles of high frequency ground wave radar for retrieval of ocean wave and sea wind elements were used to systematically study the data obtained from contrast validation test in Zhoushan sea area of Zhejiang Province on Oct. 2000, to validate the accuracy of OSMAR2000 for wave and wind parameters, and to analyze the possible error caused when using OSMAR2000 to retrieve ocean parameters.

Target height and multipath attenuation joint estimation with complex scenarios for very high frequency radar

Low-angle estimation for very high frequency(VHF)radar is a difficult problem due to the multipath effect in the radar field,especially in complex scenarios where the reflection condition is unknown.To deal with this problem,we propose an algorithm of target height and multipath attenuation joint estimation.The amplitude of the surface reflection coefficient is estimated by the characteristic of the data itself,and it is assumed that there is no reflected signal when the amplitude is very small.The phase of the surface reflection coefficient and the phase difference between the direct and reflected signals are searched as the same part,and this represents the multipath phase attenuation.The Cramer-Rao bound of the proposed algorithm is also derived.Finally,computer simulations and real data processing results show that the proposed algorithm has good estimation performance under complex scenarios and works well with only one snapshot.

New method to implement digital down converter in radar system

Digital down converter （DDC） is the main part of the next generation high frequency （HF） radar. In order to realize the single chip integrations of digital receiver hardware in the next generation HF Radar, a new design for DDC by using FPGA is presented. Some important and practical applications are given in this paper, and the result can prove the validity. Because we can adjust the parameters freely according to our need, the DDC system can be adapted to the next generation HF radar system.

Signal decomposition of HF radar maneuvering targets by using S^2-method with clutter rejection

Multiple maneuvedng targets signal processing in high frequency radar is challenging due to the following difficulties： the interference between signals is severe because of significant spread of the target Doppler spectrum, the low signal to clutter ratio （SCR） environment degrades the performance of signal process- ing algorithms. This paper addresses this challenging problem by using an S2-method and an adaptive clutter rejection scheme. The proposed S2-method improves the S-method by eliminating inter- ference between signals, and thus it enables multi-target signals to be reconstructed individually. The proposed adaptive clutter rejec- tion scheme is based on an adaptive notch filter, which is designed according to the envelop of the clutter spectrum. Experiments with simulated targets added into radar sea clutter echo and real air target data illustrate the effectiveness of the proposed method.

Adaptive modified hough transform track initiator forHFSWR tracking of fast and small targets

High frequency surface wave radar (HFSWR) is well proved to have over the horizon (OTH) detection capability to weak aerial targets, such as concealed airplanes or cruise missiles. The most important problem of detection of fast and small targets using HFSWR is earlier warning, i.e. enlargement of detection range oftargets. Therefore, the detection threshold should be decreased as low as possible, but numerous false alarms are brought about at the same time. On this condition, conventional track initiation techniques, which normally require the probability of false alarm to be at the level of 10-6, will initiate enormous false tracks and lead to abnormal operation of tracking system. An adaptive modified hough transform (AMHT) track initiator is proposed accordingly and the relation of detection range to the performance of track initiator is analyzed in this paper. Simulations are performed to confirm the capability of track initiation to fast and small targets in dense clutter by AMHT track initiator. The tolerable probability of false alarm of detector can reach the level of 10 -3 . And it performs better than track initiator based on modified hough transform (MHT).

Application of the Three-Dimensional Shallow Sea and Continental Shelf Mode for Inversion of Undercurrent

Abstract. Due to the incapability of high frequency surface wave radar （HFSWR） to detect undercurrent parameters, a new algorithm is proposed to apply a three-dimensional （3D） nonlinear barotropic shallow sea and continental shelf model in coordinate system to the inversion of undercurrent. The calculation domain of this model is the area detected by HFSWR. Considering the benthal topography of the detected area and the ocean dynamic parameters, such as surface current, wind and wave detected by HFSWR, the relation between surface current and undercurrent is established in this model, Accordingly, the undercurrent parameters of corresponding area are obtained. The inversion results agree with the law of ocean dynamics and reproduce the movement of undercurrent.

Effective Sea Clutter Spectrum Extraction Method for HFSWR in Adverse Conditions

An effective approach in solving the sea clutter spectrum extraction problem is studied in the paper.Different from the conventional signal to noise ratio（SNR）method based on Doppler frequency or range domain information,a method is developed to characterize the differences between the sea echo and those interferences are by signal to interference plus noise ratio（SINR）which jointly utilizing the range,Doppler frequency and azimuth domain information.Furthermore,these differences can be adaptable to adverse conditions by forming the necessary boundaries and constraints in searching of the maximum SINR,which greatly promotes the extraction of sea clutter spectrum.The real high frequency surface wave radar（HFSWR）data demonstrate that the proposed method is less influenced by those interferences and can effectively extract the sea clutter spectrum even under the adverse conditions.Furthermore,it has been shown as an effective method for ship detection and sea state remote sensing of HFSWR.

First order sea clutter cross section for bistatic shipborne HFSWR

This paper studies the development on the first order sea clutter cross section for bistatic high frequency surface wave radar (HFSWR). Based on the received first order electric field expression, a closed-form of cross sections is derived to account for the case of receiving antenna array being mounted on the shipborne platform. The uniform linear motion and sway motion components are assumed to be responsible for the observed differences in comparison with the bistatic fixed antenna case. Correspondingly, simulations are conducted to study the sea clutter spectral characteristics for these two cases versus different system parameters and sea state conditions. It is shown numerically that the forward motion component will spread the Bragg lines severely and the influence triggered by the sway motion can be explained as the Bessel function modulation of the ordinary sea clutter spectra. The obtained results have important implications in the application of shipborne HFSWR technology to ocean remote sensing and target detection.

Spatial Distribution Characteristics of Surface Tidal Currents in the Southwest of Taiwan Strait

This study was conducted on the spatial distribution characteristics of surface tidal currents in the southwestern Taiwan Strait based on the quasi-harmonic analysis of current data obtained by two high frequency surface wave radar(HFSWR) systems. The analysis shows that the tidal current pattern in the southwestern Taiwan Strait is primarily semi-diurnal and influenced significantly by shallow water constituents. The spatial distribution of tidal current ellipses of M2 is probably affected by the interaction between two different systems of tide wave, one from the northern mouth of Taiwan Strait and the other from the Bashi Channel. The directions of the major axes of M2 tidal current ellipses coincide roughly with the axis of the Taiwan Strait. The spatial distribution of the magnitudes of the probable maximum current velocity(PMCS) shows gradual increase of the velocity from northeast to southwest, which is in accordance with the spatial distribution of the measured maximum current velocity(MMCS). The directions of the residual currents are in accordance with the direction of the prevailing monsoon wind at the Taiwan Strait and the direction of the Taiwan warm current during summer. The bathymetry also shows a significant effect on the spatial distribution characteristics of tidal currents.

Iterative physical optics model for electromagnetic scattering and Doppler analysis

An iterative physical optics（IPO） model is proposed to solve extra large scale electric electromagnetic（EM） scattering from randomly rough surfaces. In order to accelerate the convergence of the IPO model, the forward-backward methodology and its modification with underrelaxation iteration are developed to simulate the rough surface scattering; the local iteration methodology and the fast far field approximation（Fa FFA） in the matrix-vector product are proposed to reduce greatly the computational complexity. These techniques make Monte Carlo simulations possible. Thus, the average Doppler spectra of backscattered signals obtained from the simulations are compared for different incident angles and sea states. In particular, the simulations show a broadening of the Doppler spectra for a more complicated sea state at a low grazing angle（LGA）.

Drifter Trajectory Tracking Experiment and Analysis with Multi-Frequency HFSWR in the East China Sea

We conducted the drifter trajectory tracking experi- ment for two drifters in the East China Sea, in which the currents were detected by two multi-frequency HFSWRs using the Lagrangetracking method. The tracking drifter trajectory shows good agree- ment with the GPS records by qualitative and quantitative analysis that indicates that drifter tracking by HFSWR is valid. In the 12- hour tracking for drifters P1 and P2, the average errors are 1.84 kin, 1.73 kin, and the maximum errors are 3.52 km and 3.12 km, respec- tively. The current measurement is evaluated by an acoustic Doppler current profilers （ADCP） in-situ observation, and it is the first multi-frequency HFSWR ocean surface monitoring experiment in China. In addition, the main aspects （HFSWR current measurement error, wind, and wave） influencing the tracking accuracy are dis- cussed, and especially the wind factor＇s impact is analyzed through the wind filed data.

Experimental Determination of Significant Waveheight by OSMAR071: Comparison with Results from Buoy

OSMAR071 is the latest product of the OSMAR （ocean state monitor and analysis radar） series of high frequency surface wave radar （HFSWR）, which was developed by the Radiowave Propagation Laboratory of Wuhan University. It adopts a modified Barrick waveheight inversion model. The modifications are introduced to improve the model＇s performances under the effect of noises and interferences and in the case of broad beam radar detection. The two unknown coefficients in the modified model are figured out by fitting the HFSWR significant waveheight results to those output from a wave buoy located in the radiating coverage of the radar site. The model is applied to inverse the waveheights from radar data for the duration from Dec. 1st, 2008 to Feb. 25th, 2009, and then the radar waveheights are compared with the buoy measurements. Results show that the rms difference between radar-derived significant waveheights and those from the buoy is 0.38 m and the correlation coefficient between the two series is 0.66. This study describes OSMAR071 observation of significant waveheight with relatively satisfactory accuracy during about three months.