Improving River Slope Estimation in Noisy SWOT Water Surface Elevation Data Using Quantile-Based Filtering
Year: 2025
Type: preprint
Abstract: The Surface Water and Ocean Topography (SWOT) satellite mission offers novel global water surface elevation (WSE) and discharge estimates, providing new opportunities for river measurements in ungauged and data-limited regions. However, accurately estimating river slopes from WSEs remains a challenge, especially in steep and complex terrains. As river slopes are non-linear, an approach that integrates this complexity is needed. Traditional methods, such as ordinary least squares (OLS) regression, are sensitive to outliers, which can lead to biased slope estimates and inaccurate discharge predictions. This field-based study and accompanying analyses introduce a novel quantile regression filtering approach to address these challenges by systematically identifying and removing slope-influencing outliers. Using in situ Global Navigation Satellite System (GNSS) measurements along the Rio Petrohue, Chile (-42.3 degrees, -72.4 degrees), a river with highly variable slopes, we validate the performance of SWOT-derived slopes and assess the stability of WSE trends. Our approach highlights the limitations of current regression techniques and demonstrates how slope-based quantile filtering enhances slope estimation accuracy between GNSS and SWOT measurements. This filtering method establishes a framework for future automated algorithms, with positive implications for hydrological modeling, flood prediction, and water resource management. Additionally, we outline a novel method for smoothing the SWOT PIXC point cloud, rather than aggregating it, and applying quantile regression-based segmentation to detect slope transitions, suggesting a future avenue to preserve spatial structure while enabling automated hydraulic analysis.
Authors: Trevor Wilkerson, Eric Allan Sproles, Kathryn Plymesser, Pablo Iribarren-Anacona, J. Toby Minear
Language: English
Subfield: Water Science and Technology
Field: Environmental Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4413885329