Remote Sensing of Ocean Surface Wind Speed Using Super-Resolution Delay-Doppler Maps and Reflected BPSK/BOC Signals

Abstract

Over the past few years Global Navigation Satellite System reflectometry (GNSS-R) has become a new and feasible remote sensing method for geophysical information monitoring on the Earth’s surface For instance the Cyclone GNSS (CYGNSS) hosted by NASA is dedicated to improving the trajectory prediction and structural analysis for hurricanes In the future the TRITON satellite constructed by the National Space Organization (NSPO) Taiwan will also use the GNSS-R method to provide ocean wind observation data The basic principle of this method is to receive and process reflected GNSS signals scattered from the Earth’s surface extract valid observations from them and retrieve the target parameters through the developed geophysical model function (GMF) Since the power of the reflected signal is very weak it is necessary to restore the signal strength through a long integration process For these space-borne missions the GNSS-R receiver involves the trade-off between computation complexity/communication cost and data retrieval A GNSS-R system will produce a set of delay-Doppler map (DDM) measurements that reveal surface characteristics Even though the technique to retrieve different geophysical parameters have been developed there is relatively little research on the analysis of DDM resolution on data retrieval quality A fully-resolved DDM reveals more information but consumes more resources in terms of onboard processing and downlinking As a result existing tasks typically use compression in the retrieval process while the fully-resolved DDM is reserved for specific purposes such as calibration In this dissertation a deep learning super-resolution algorithm is developed to reconstruct a high-resolution DDM also known as a super-resolution DDM based on a low-resolution DDM The proposed method is applied to the DDM product disseminated by the CYGNSS to verify the feasibility of the proposed method The experimental results show that using the super-resolution DDM leads to almost identical performance to that obtained using the fully-resolved DDM in terms of wind speed retrieval The statistical analysis shows that the proposed method may save 94% of the DDM data generation volume and 15% of the data transmission volume and the performance degradation in terms of wind speed retrieval is negligible These findings provide a potential strategy for future spaceborne GNSS-R missions With the advent of the new GNSS era multi-frequency multi-modulation signals are expected to enhance not only positioning performance but also remote sensing applications It is known that for some constellations navigation satellites broadcast signals employing both binary phase-shift keying (BPSK) modulation and binary offset carrier (BOC) modulation in the same frequency band This dissertation proposes a new GNSS-R measurement called a composite delay-Doppler map (cDDM) that utilizes the received reflected GNSS signals with different modulation techniques for the purpose of retrieving wind speed It is assumed that the GNSS-R receiver can receive BPSK and BOC signals simultaneously in the same frequency band (e g GPS III L1 C/A and L1C or QZSS L1 C/A and L1C) and can process the signals to generate the proposed GNSS-R measurements An exploration of the observable features extracted from the composite DDM and the wind speed retrieval algorithm is also provided The simulation verifies the proposed method under a configuration that is specified for the orbital and instrument specification of the upcoming TRITON mission

Date of Award	2020
Original language	English
Supervisor	Jyh-Chin Juang (Supervisor)