Abstract
Synthetic Aperture Radar (SAR) signal processing from airborne and satellite platforms, capable of resolutions of the order of a metre, is a novel sensing technology. It
can generate high-resolution images in all weather and light conditions and the final product is comparable with and complementary to, that obtained with optical sensors.
The field of SAR applications is very wide and it has grown enormously in recent years, such as the number of SAR sensors in orbit. As a consequence, there is a necessity to provide improved accuracy and reliability of Earth Observation (EO) data and, above all, a method to compare these data, if generated by different sensors.
Moreover, looking at the recommendation made by the Committee Earth Observation Satellites (CEOS) in 2001, it is also important to ensure that all measurements and associated instrumentation used for any quantitative purpose in remote sensing be fully traceable to International System of Units (SI) as part of the Quality Assurance process.
To realise a traceable and trustworthy measurement, robust assessments of the uncertainties associated with SAR products are required.
A new methodology for the analysis of uncertainties in Synthetic Aperture Radar (SAR) data, based on the Guide to the expression of uncertainty in measurement (GUM) [1]
published by the Joint Committee for Guides in Metrology, is here presented.
The main idea is to analyze how some input parameters uncertainties, related to instrument specifications during the data acquisition, propagate through the SAR signal
processing chain to the parameters that are used to describe the quality of SAR performance.
Through an Uncertainty Analysis (UA), these quality parameters of SAR performance are provided to the end-users not only in terms of their measured values as usual but also with associated uncertainty intervals evaluated for a stipulated confidence level.
Moreover, a Sensitivity Analysis is performed permitting to assign to each input parameter its role in the uncertainty associated with the output quantities. This analysis aims to give better knowledge about the quality of SAR data enabling the comparison of data derived from the same or different types of sensors, by their traceability to international reference standards.
In this thesis, the GUM methodology is presented, describing the different approaches to implement it. Moreover, a description of the SAR processing chain is shown, defining all components steps, which generate the Single Look Complex (SLC) product.
The several SAR levels along with the processing chain have been simulated and UA
and SA have been performed, implementing a Matlab code.
The application of the methodology to a simulated scene is performed and the first results are discussed, representing a novelty in the SAR community.