| 摘要: | 高頻雷達可應用於用於與海洋打撈、溢油響應、近岸區域管理，理解上層海洋動力等方面，目前全世界已有數百座高頻 (HF) 岸基雷達系統正在運作。目前，台灣海岸線安裝了二十五個高頻海岸雷達站，包括交叉/環路單極子和相控陣列雷達系統，主要用於觀測洋流。下一階段，二十一座相控陣列高頻（HF）和甚高頻 (VHF)雷達站將於2022年底建成，將被用於台灣近岸海域海浪之長期監測。為滿足高頻雷達應用軟體開發需求，本研究將聚焦於兩個高頻雷達應用議題，其一是評估不同海況下海表波浪參數反演結果之不確定性，其二是開發近岸船舶偵測算法。
首先都普勒頻譜模擬結果將被用於理解雷達截面積（RCS）與海況參數之理論關係。基於巴里克理論（Barrick’ s theory），高頻雷達截面積之都普勒頻譜可以從指定的方向波譜中模擬出來。方向波譜可由海浪譜計算得來，穩定風速（steady wind）可採用風波譜模型計算，季風或熱帶氣旋條件下可採用第三代海浪波譜模型計算。目前的初步結果為，已完成在不同風速、風向、雷達運行頻率、波譜分散度參數等狀況下都普勒頻譜模擬之敏感性測試。測試結果與文獻中譜型結果吻合良好，這表明我們的模擬軟體運行良好。
為從高頻雷達海洋回波估算波浪參數，如即示性波高、平均週期和海浪譜，研究採用已有方法來建立測試和評估方法表現之估算系統。最初，數值模擬被應用於評估不同天氣狀況下波浪參數之偏差。結果顯示，估算自都普勒頻譜之波浪參數，其不確定性在颱風條件下要低於季風條件。另外，為獲得來自實際運作之高頻雷達之波浪參數偏差，本研究採用頻率為27.75MHz接收後向散射訊號之高頻雷達系統（LERA MKIII）。該系統於2018年11月建於台灣臺中港北方，擁有4根發射天線及16根接收天線，是用於觀測海浪參數之高頻陣列雷達系統。
估算結果與位於雷達覆蓋區之聲波波流剖面儀（AWAC）實地觀測數據進行比對，比對結果顯示雷達系統運作良好，估算系統表現良好。此外，研究發現波浪關聯參數（connection coefficient of wave parameter）是存在的，此關聯參數是雷達徑向-波向夾角、微小參數（smallness parameter）以及譜寬參數（spectral width parameters）之函數。研究提出修正算法來估算比例因子（scaling factor），比例因子將被用於校驗（calibrate）雷達推算之波浪參數。結果顯示波高比例因子具有與雷達觀測方向-波向夾角之相依性，而波浪週期比例因子則主要受微小參數（smallness parameter）之影響。
本研究應用兩種方法從高頻雷達後向散射訊號偵測近岸海域船舶位置，即都普勒-距離（DR）方法與方位角-距離（AR）頻譜方法。船舶位置估算訊息已經與船舶自動偵測系統（AIS）提供之訊息比對驗證，結果顯示兩種方法均有良好估算效果。AR方法可以偵測船隻軌跡，RD方法可以提供這些目標物之雷達徑向速度。此外，我們也發現船舶長度、方向及航向等特征將影響目標物數量之偵測。整體上，本研究討論了將高頻雷達技術應用於海浪監測及船舶偵測之優點與限制。
;Hundreds of high-frequency (HF) coastal radar systems are operated in the world for purposes related to marine salvage, oil spill response, coastal zone management, and understanding of upper ocean layer dynamics. At present, 25 HF coastal radar stations consisting of both cross/loop monopole and phased array systems were installed along Taiwan’s coastline and are primarily operated modes for observing ocean currents. In the next period, 21 phased array HF and very high-frequency (VHF) radar stations are on-going to install at the end of 2022 to monitor the evolution of ocean surface waves around the Taiwan island’s coastal area in the long term. In order to fulfill the demands for high-frequency radar application exploitation, this study focuses on two HF radar applications’ essential topics, which are assessing the uncertainty of ocean surface wave parameters under various sea-states and developing algorithms for coastal vessel monitoring.
First, the simulation of radar Doppler spectra was implemented to understand the theoretical relationship between radar cross-section (RCS) and sea-state parameters. Based on Barrick′s theory, the Doppler spectra of HF radar cross-section are simulated from the given directional wave spectrum, which can be generated by applying the wind-wave spectra model for steady wind conditions or using the 3rd generation wave spectra model for monsoon and typhoon conditions. As preliminary results, various sensitivity tests of the simulated Doppler spectra are implemented under different wind speeds, wind directions, operating radar frequencies, and spreading parameters. The result showed that the simulated Doppler spectra agree well with those in the literature. This indicated that our simulation toolbox works well.
To estimate the wave parameters. i.e., significant wave height, mean period, and wave spectrum from HF radar sea-echoes, existing methods are implemented to establish estimators for testing and evaluating the method’s performance. In the beginning, the numerical simulation is used to assess the bias estimation of wave parameters under various weather conditions, such as the steady homogenous wind, monsoons, and typhoons. The results showed that the uncertainty of wave parameters estimated from simulated Doppler spectra under typhoon conditions is lower than those in monsoon conditions. In addition, to assess wave parameters’ bias from the actual HF radar data, the backscattered signal of the 27.75 MHz HF radar (LERA MKIII) system is used. This system consists of 16 Rx antennas in a linear array installed at the northern of the Taichung harbor, Taichung City, Taiwan, in late November 2018 for wave monitoring in the long term. Estimation results are compared with those of in-situ data observed by an acoustic wave and current profiles (AWAC) deployed in the radar’s footprint. The comparison results indicated that the radar system and estimators perform very well. Furthermore, it is found that the connection coefficients of wave parameters (which might be the function of radar-to-wave angle, smallness parameters, and spectral width parameters) have existed. Correction algorithms are proposed to estimate the scaling factor for calibrating radar-deduced wave parameters. The results demonstrated the dependence of wave height scaling factor on the radar-to-wave angle, while wave period scaling factors are mainly influenced by smallness parameters.
On the other hand, this study implements two approach methods to identify coastal vessel locations using HF radar backscattered signals: the rang-Doppler (RD) (or Doppler-Range, D-R) spectra and range-Angle brightness distribution methods. The estimated position of coastal vessels is compared with ship’s information from Automatic Identification System (AIS) data for assessing the performance of the radar system and the efficiency of detection methods. The results showed that both approaches work well. Furthermore, while the RA method can monitor ships’ trajectories, the RD approach can provide the radial speed of those targets. Besides, we also found the influence of the ship’s characteristics, including the length, the direction, and the heading, on the detection number of targets. Overall, this study illustrated the advantages and the limits of the HF radar technique for wave monitoring and ship detection. |
