Leveraging Remote Sensing for Global River Monitoring

Abstract

The availability and distribution of fresh water resources has always been an item of great global interest, not only because it is a critical resource for human consumption, agriculture, and industry, but also because the hazards posed by hydrologic extremes (both excesses and shortages) can have lasting global impacts. With respect to these hazards, observations of surface waters can aid in monitoring reservoir conditions, providing early warning for possible flooding conditions, or predicting areas which may become susceptible to hydrologic drought. Traditionally, these observations were done through discharge gauging stations; however, the global availability of these has declined in recent years. As such, there is a need for alternative methods and data sources to supplement these observations. The goal of this dissertation is to leverage some of the remote sensing observations available to derive new methods for monitoring global water resources. In Chapter 2, an algorithm is created to derive continuous estimates of discharge from limited in-situ gauges. This spatiotemporal interpolation method is tested in a set of synthetic experiments illustrating the potential for basin wide discharge reconstruction from limited observations. Chapter 3 builds upon this work, applying the spatiotemporal interpolation method in the context of the upcoming Surface Water and Ocean Topography mission that will provide increased spatial coverage compared to current in-situ gauge networks but will have significant temporal gaps in observations. The SWOT mission is further explored in Chapter 4, where the proposed mission orbit is examined in relation to the amount of information it might be able to provide about global river basins. As a result of the orbital pattern and the uniqueness of individual river basins, careful consideration will be required to maximize the utility of SWOT. Finally, in Chapter 5 an alternative source of observations is utilized to provide rapid predictions of inundated areas as flooding occurs. Using a machine learning algorithm and passive microwave brightness temperature observations, high resolution estimates of surface water extents are generated. In the context of global hydrology, the work presented in this dissertation provides new pathways for monitoring the availability and distribution of fresh water resources.