A curve-fitting approach to river bathymetry estimation using SWOT data
Abstract
Accurate river discharge estimation requires reliable bathymetric information. However, bathymetry data are costly to measure in situ and are not easily retrievable from remote sensing data. Hence, aside from a few locations where bathymetric surveys have been conducted, they are essentially unavailable even for major rivers. Prediction of channel bottom depth using measurements of water surface elevation (WSE) and river width (w) - which SWOT is expected to produce - has been suggested, but its full potential has yet to be explored. We report on an exploratory study of the Upper Mississippi River using a 5 m resolution bathymetry DEM interpolated from USGS sonar measurements covering nearly half the length of the 2000 km river from eastern Minnesota to Cairo, Illinois. We developed WSE-w relationships similar to what we expect to be produced by SWOT starting with the high resolution sonar data after corrupting the synthetic observations with errors, and fit curves to the inferred cross-sections at different levels of channel exposure. We then used the fitted WSE-w relationships to predict the unobserved channel bed elevations (below the lowest WSE that we expect to be "seen" by SWOT during its 3-year mission life). We tested linear and nonlinear models with one or more structural breaks for fitting the synthetic WSE-w data points over a range of river depths, from bankfull to low-flow. Our nonlinear and piecewise models extend previous work which has used only linear models; we show that using nonlinear models can predict bottom depth while requiring less channel exposure than linear models. They were able to predict bottom depth with a bias of less than 0.5 m for about 45-70% channel exposure, outperforming the piecewise linear model. However, at higher levels of channel exposure, the piecewise linear model is superior.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFMOS53C1334S
- Keywords:
-
- 1845 Limnology;
- HYDROLOGYDE: 1856 River channels;
- HYDROLOGYDE: 4520 Eddies and mesoscale processes;
- OCEANOGRAPHY: PHYSICALDE: 4544 Internal and inertial waves;
- OCEANOGRAPHY: PHYSICAL