Uncertainty Assessment of Wave Remote Sensing

  • 董 東璟

Student thesis: Doctoral Thesis

Abstract

Ocean wave, among the natural phenomena, is one of the most complex factors prevalent in coastal and ocean engineering. Its features are extremely random, affected by meteorological factors, topological conditions and currents, which cannot be fully understood by numerical or physical models alone. Field measurement must be performed to increase the knowledge of waves.

 The most common approaches to wave measurement can be categorized as direct and indirect methods represented by In-Situ measurement and Remote Sensing technique respectively. The data measured from the direct in-situ method is viewed as ground truth data. The remote sensing technique provides extensive sets of ocean surface data. It points to a developing trend of coastal and ocean monitoring methods. Among all the sensors, RADAR (Radio Detection And Ranging) is potentially the key instrument for wave measurement purposes due to by its properties of high resolution and fewer environmental limitations. However, although the remote sensing technique is such a promising tool, with it exist some problems such as confidence in the correctness and accuracy of remotely sensed data.

 Any measurement is subject to imperfections and potential errors. This uncertainty is defined as the measure of measurement error and serves as the quantitative indication of the quality of measurement data. Assessment of uncertainty increases the confidence of data users and provides information for technical improvements in radar wave observation systems. The uncertainty of wave remote sensing is multiplied by variables such as data acquisition, image processing, correction and so on. Due to the remote sensing technique being an in-direct observation method, significant parameters need to be calibrated with in-situ data which demonstrates that the calibration process is the vital procedure in setting a radar wave observation system. The purpose of this study is to assess the uncertainty of remotely sensed wave heights produced by the calibration processes.

 To achieve this, several tasks need to be performed beginning with the establishment of the calibration model. Using the idea that the radar wave height is correlated with the image SNR (Signal to Noise Ratio), a linear regression model can be identified. Second comes the estimation of uncertainties, including both uncertainty from the parameters within the model and uncertainty from the model itself. In the final part of this study, the uncertainty reduction approaches are presented and discussed.

 In order to establish the calibration model, the spectrum has to be extracted from radar images. The non-homogeneity property is identified in the nearshore area in chapter 4 leading to the development of a non-homogeneous image spectrum analysis method used to extract the spectrum from radar images in chapter 5. This algorithm is derived and developed by this study and validated as a proper tool for non-homogeneous image spectrum analysis. In order to provide high quality in-situ data for calibration and assess the parameter uncertainty, a data quality-check program is developed in chapter 3. This data quality-check program contains both time series raw data checks and statistical parameters checks. Variations in data after applying the quality-check program are regarded as uncertainties of in-situ data and are used to assess the parameter uncertainty in the calibration process. In chapter 6, the uncertainties of wave remote sensing are evaluated by the uncertainty propagation method.

 This study will show that the dominant uncertainty in remotely sensed wave heights results from the quality of in-situ data and that the minor uncertainty comes from the image spectrum analysis method. The formation of the calibration model produces yet less uncertainty. Thus, in order to improve the accuracy of remotely sensed data, the work of collecting high quality representative in-situ data for calibration must be emphasized alongside the choice of calibration model.
Date of Award2003
Original languageEnglish
SupervisorZsu-Hsin Chuang (Supervisor)

Cite this

'