River Bathymetry Retrieval From Landsat-9 Images Based on Neural Networks and Comparison to SuperDove and Sentinel-2

The Landsat mission has kept an eye on our planet, including water bodies, for 50 years. With the launch of Landsat-9 and its onboard Operational Land Imager 2 (OLI-2) in September 2021, more subtle variations brightness (14-bit dynamic range) can be captured than previous sensors series (e.g., 12-bit Landsat-8). enhanced radiometric resolution OLI-2 appeals to aquatic remote sensing community because instrument might capable resolving smaller differences water-leaving radiance. This study evaluates potential map river bathymetry from imagery. We employ a neural network (NN)-based regression model retrieval compare results with optimal band ratio analysis (OBRA). effect pan-sharpening depth is also examined. In addition, we perform intersensor comparison Sentinel-2 newly available 8-band SuperDoves PlanetScope constellation. Depth Colorado Potomac Rivers imply that provided accurate across range depths up 20 m, particularly when pan-sharpened. Downsampling SuperDove data improved due signal-to-noise ratio, most notably deep waters (maximum detectable increased ∼15 ∼20 m). Similarly, spectral relative 4-band Doves. NN-based outperformed OBRA by incorporating information.