Remote sensing for geomorphic and hydrodynamic modeling and process understanding

Abstract

River discharge and geomorphic processes controlling floodplain-river interactions are fundamental measures of fluvial geomorphology and important inputs and considerations for hydrodynamic modeling. Due to our anthropogenic dependency and interconnection with rivers, humans have developed technologies and methods to measure and estimate river discharge and fluvial geomorphic change through field techniques, hydrodynamic modeling, and remote sensing. Due to the novelty of remote sensing data, there are still unknowns that need investigation to better understand how these data can be properly applied to our current models and empirical understanding. This dissertation contains three studies that examine how remote sensing, in conjunction with other data sources, can be used to accurately represent terrain at various resolutions in a hydrodynamic model (Chapter 2), characterize and understand dunefields along the Colorado River in the Grand Canyon in relation to geomorphic variables (Chapter 3), and lastly estimate river discharge from radar satellite measurements at river gage locations (Chapter 4).

Chapter 2 of this dissertation determines how hydrodynamic model results (water depths, water velocities, and inundation extent) are affected by drone lidar DEM resolution and computational mesh resolution over 1.5 km of Stroubles Creek in Blacksburg, Virginia, USA. This was done by running several simulations of a hydrodynamic model with various resolutions of computational mesh (1 m and 2 m) and DEM (0.1 m, 0.25 m, 0.5 m, 1 m, and 2 m). We found that for water depths, the largest differences between simulations occurred laterally throughout the floodplain, whereas for water velocities differences were concentrated along the channel-floodplain boundary. The water depth and velocity differences were not correlated with lidar ground point density. We also found that the inundation extent was dependent on DEM resolution. Lastly, changing DEM resolution was found to be equivalent to altering floodplain roughness by up to 12% for water depths and 44% for water velocities. These findings show that the modeler's choice of terrain resolution and computational mesh resolution do matter and should be considered in the context of features that are expected to affect flow in the floodplain.

Chapter 3 consists of a study that focuses on characterizing and developing geomorphic understanding of aeolian dunefields along the Colorado River in the Grand Canyon, USA. These 58 dunefields were formed by pre-dam floods and are now disconnected from the river and represent the sediment matrix in the Grand Canyon that are currently used for modern recreation, biota habitat, and protect hundreds of archaeological sites of historical and indigenous importance. They are now maintained only by wind-blown sand from modern sandbars along the Colorado River. Motivated by their importance and threatened sediment supply from hydropower dam operations, this study determines how dunefield area is influenced by aeolian-fluvial interactions, climate, river hydrology and geomorphology, and canyon physiography variables derived from remote sensing data, hydrodynamic modeling results, and field observations. We found that dunefield area is significantly influenced by pre-dam river width and the distance from dunefield center to river center, thus demonstrating that having enough accommodation space to grow, while still being near their primary sediment source allows dunefields to be maintained. Additionally, from the 196 dunes that were measured from remotely sensed digital terrain data, migration rates varied from 0.01 m/year to 3.0 m/year and dune heights varied from 2.0 m to 8.2 m, with climbing/parabolic dunes being the most common morphological type (n = 88). We end our study by proposing an analog study between Grand Canyon dunefields and dunefields found in the Valles Marineris canyon system on Mars to better understand and constrain planetary sediment budgets and primary wind patterns to dictate sediment transport in these unique geomorphic settings.

Lastly, Chapter 4 of this dissertation consists of assessing and characterizing performance of empirical equations used to estimate river discharge from data collected by the Surface Water and Ocean Topography (SWOT) satellite with comparison back to river discharge measured at gage locations. The SWOT satellite was launched in December 2022 and is the first satellite ever to explicitly include river measurements and discharge estimation in its science goals. This study was motivated by previous findings that demonstrated how empirical equation agreement with SWOT data correlates with the success of Mass-Conserved Flow Law Inversion algorithms, which are used to estimate roughness and cross-sectional area, both parameters that cannot be detected by SWOT but are necessary to estimate river discharge. Due to variable SWOT measurement quality, we heavily filtered and cleaned the data, allowing for 68 river gages to be used in our study. We then evaluated five empirical discharge equations, two variations of the Manning-Gaukler-Strickler equation and three rating curve equations, and compared these modeled discharge data back to field data collected at river gages. We found that the stage rating curve performed the best in comparison to field data since SWOT was successful in consistently measuring water surface elevations and its variability in many rivers. We also determined that SWOT river width potentially contains high amounts of uncertainty, thus any empirical equation that contained SWOT width had poor performance. Our study shows that there are advantages to allowing consideration of multiple empirical discharge equations due to variation in SWOT measurement quality.