Figure 2 | Scientific Reports

Figure 2

From: Automatic delineation of glacier grounding lines in differential interferometric synthetic-aperture radar data using deep learning

Figure 2

Different steps of the pipeline for (a) a sample interferogram with hand-drawn grounding line in white; (b) output of the neural network in raster format with a variable width that is a measure of uncertainty of the output. The gray mask in the background represents training (white) and testing (gray) sites; (c) Vectorized output and uncertainty contours; (d) zoomed-in region highlighted by the maroon box in panel (c) showing both manual and ML results and uncertainty bars; (e) Comparison of the manual and ML performances in the area outlined in blue in panel (c). The interferograms are obtained from Sentinel-1a/b data in 201813,14, and are plotted with Python 3.7.417 using Rasterio version 1.0.2118 and Matplotlib version 3.3.219.

Back to article page