| 摘要: | 全球衛星導航反射訊號接收儀（Global Navigation Satellite System-Reflectometry， GNSS-R）是以接收全球衛星定位系統在地球表面反射訊號形式，量測雷達散射截面積，以此獲取散射表面特徵之技術。自1998至2004年技術理論論證、雷達散射截面積校正演算法初步建立階段后，逐漸被廣泛應用於地球物理資訊如海表風速、海表熱通量、陸地水域面積、土壤濕度等之反演與量測，對於颱風預報、海水溢淹等災害防治有其重要意義。GNSS-R技術以L band受高風速下降水影響較小、衛星重訪率高、體積小、成本低等優勢成為微波遙測海表風速新的選擇。英國（2003， 2014）、美國（2017）、中國大陸（2019）相繼發射GNSS-R衛星用以進行風速等量測。 風速-波高-頻率作為高風速海況下互相聯動的變數（wind-wave triplet）已得到基於實測的論證，代表波浪發展階段之波齡在熱帶氣旋波浪學中亦被證實與風速及海表粗糙度為一組互相影響之變數，在Cyclone Global Navigation Satellite System （CYGNSS）團隊建立高風速Geophysical Model Function （GMF）時亦被提及是造成高風速反演不確定性的主要原因之一。有研究團隊已將波浪模式之示性波高納入風速反演之考量，但尚未有人報導使用波齡或波浪平均頻率與風速共同建立風速反演模組。2023年9月臺灣獵風者號（TRITON）GNSS-R衛星即將發射，為使衛星升空後有可信賴之海氣參數反演系統，本論文以2017年發射且已作業化公開校正及反演數據產品之美國CYGNSS量測數據為先導，為獵風者號星載平台開發海表粗糙度及風速反演模組前體。
本研究之目的為，首先建立測試使用Level 1b校正模組以獲取品質合格之雷達散射截面積。其次為將基於實測及理論模型的高風速下風與波浪交互作用之最新理論引入技術發展，開發出基於目前散射原理而物理上更貼近高風速海況之風速反演模組，以推進反演技術發展降低風速反演不確定性，本研究將首次測試以無因次化之波浪頻率--波齡作為第二變數，來反演風速，並評估其與傳統方法反演品質之差異， 並以相同參考風速與各國基於GNSS-R技術反演風速品質相比較。使其準確度及效能符合作業化需求，為大氣模式提供潛在可用於同化之雷達散射截面積、風速、海表粗糙度產品。
本研究貢獻在於，在各國已發展之GNSS-R衛星資料處理技術基礎上，自主建立Level 1b雷達散射截面積矯正模組，產生誤差在5%以內之Delay Doppler Map Average （DDMA）產品，並首次將代表波浪成長階段的波齡作為參數用以建立風速反演模組，將基於實測研究最新發展之高風速下波浪理論應用推廣於技術開發，並在風速品質評估階段發現，相比於直接將DDMA與風速產品相連接之方法反演的風速，反演風速之殘差降低近50%。 產生之DDMA與風速反演產品具提供大氣模式做同化處理之潛力。 產品品質具有國際競爭力。
;Global Navigation Satellite System-Reflectometry, or GNSS-R, is a method that involves receiving reflected signals from the Earth′s surface, transmitted by the GNSS. The aim is to measure the Normalized Bistatic Radar Cross Section (NBRCS), which enables the acquisition of geophysical information about the scattering surface. Following the technical theory demonstration from 1998 to 2004 and the initial establishment of the radar scattering cross-sectional area correction algorithm, the GNSS-R technique has been widely applied to the retrieval of geophysical information such as sea surface wind speed, sea surface heat flux, inland water extend detection, and soil moisture. These are significant implications for disaster prevention and mitigation, such as typhoon forecasting, inland flood inundation or seawater inundation. GNSS-R became a new candidate for microwave sea surface wind remote sensing for its low risk of being influenced by precipitation, high revisit period, low mass and low-cost properties. United Kingdom (2003, 2014), America (2017) and Mainland China (2019) have already launched their GNSS-R satellite, and they have all been applied to retrieve the sea surface wind speed, u_10. Wind speed-wave height-wave frequency has been proven to be a coupled variable in high wind conditions and is called a wind-wave triplet. Wave age which represents the wave development stage have been proved to vary with wind speed and ocean surface roughness, mean square slope. The wind speed retrieving algorithm for GNSS-R has been explored for years. Most studies directly link Delay Doppler Map observable to wind speed to generate a one-step geophysical model function (GMF). In Cyclone Global Navigation Satellite System (CYGNSS) mission, they have built a wind speed retrieving algorithm for high wind speed conditions (15~70 m/s) by linking DDM observable to hurricane penetrate-measured collocated data and mentioned that wave age and fetch length in the state of development of the long wave portion near or far from a hurricane would account for the high uncertainty in high wind speed GMF establishment. Significant wave height has been used to retrieve wind speed in NOAA′s algorithm, however, no study has reported the usage of wave age or wave frequency to retrieve the wind speed. The GNSS-R satellite TRITON will be launched in September, 2023. This study aims to build a pre-launch DDM calibration and retrieving system using the published operational products from CYGNSS system for TRITON.
There are two aims of this study. The first aim is to establish the Level 1b calibration module to obtain the qualified normalized bistatic radar cross section (NBRCS) from GNSS-R measured Delay Dopper Map (DDM). The second aim is to develop a wind speed retrieving module applying the latest theories based on theoretical models of wind and wave interactions under high wind speeds to get wind products that better represent the natural condition. For the first time, this research will test the use of dimensionless wave frequency -- wave age as a second variable for wind speed retrieval and evaluate its difference in quality from DDMA-u_10 GMF method. Furthermore, the results will be compared with the quality of wind speed retrieval based on GNSS-R technology from different countries using the same reference wind speed. The established system can provide NBRCS, wind speed and ocean surface roughness products for the data assimilation system.
The contribution of this study is, firstly, we have established a Level 1b calibration system base on the published literature and have produced Delay Doppler Map Average (DDMA) product in which the residual to the reference product is within 5%. The wind speed retrieved from the latest-built DDMA-Mean Square Slope-Wave age GMF improved the root-mean-square-difference to the degree of 50% compared to the traditional method, and the bias is within 0.1 m/s for wind speed within 0 to 12 m/s. The quality is competitive with the products from other teams. |
