Accurate maps of regional surface water features are integral for advancing ecologic, atmospheric and land development studies. The only comprehensive surface water feature map of Alaska is the National Hydrography Dataset (NHD). NHD features are often digitized representations of historic topographic map blue lines and may be outdated. Here we test deep learning methods to automatically extract surface water features from airborne interferometric synthetic aperture radar (IfSAR) data to update and validate Alaska hydrographic databases. U-net artificial neural networks (ANN) and high-performance computing (HPC) are used for supervised hydrographic feature extraction within a study area comprised of 50 contiguous watersheds in Alaska. Surface water features derived from elevation through automated flow-routing and manual editing are used as training data. Model extensibility is tested with a series of 16 U-net models trained with increasing percentages of the study area, from about 3 to 35 percent. Hydrography is predicted by each of the models for all watersheds not used in training. Input raster layers are derived from digital terrain models, digital surface models, and intensity images from the IfSAR data. Results indicate about 15 percent of the study area is required to optimally train the ANN to extract hydrography when F1-scores for tested watersheds average between 66 and 68. Little benefit is gained by training beyond 15 percent of the study area. Fully connected hydrographic networks are generated for the U-net predictions using a novel approach that constrains a D-8 flow-routing approach to follow U-net predictions. This work demonstrates the ability of deep learning to derive surface water feature maps from complex terrain over a broad area.

This paper reports on a new Area Preserving Segment Collapse (APSC) algorithm for simplifying polygonal boundaries while preserving the polygonal area at simplified target scales and minimizing areal displacement. A general segment collapse algorithm is defined by iteratively collapsing segments to Steiner points in priority order, guided by placement and displacement functions. The algorithm is specified by defining functions that minimize areal displacement under the constraint that the areas of adjoining polygons are preserved exactly. Self-intersections can be avoided by testing for intersections with two new line segments associated with each segment collapse operation. The paper demonstrates simplification results for a sample of 10 lakes formed from alpine, Karst, glacial and arid desert processes as well as artificial dams. APSC results are compared with three other simplification routines and evaluated for area preservation, linear and areal displacement, complexity and introduction of boundary self-intersections. Results confirm that the APSC algorithm preserves area exactly and indicate that it outperforms the other tested algorithms for minimizing areal displacement while producing reasonably low measures of linear displacement. Self-intersections can occur more commonly with APSC than other algorithms but are avoided with the proposed topology check. The APSC algorithm additionally preserves polygon complexity better than other tested algorithms.

This paper describes a workflow for automating the extraction of elevation-derived stream lines using open source tools with parallel computing support and testing the effectiveness of procedures in various terrain conditions within the conterminous United States. Drainage networks are extracted from the US Geological Survey 1/3 arc-second 3D Elevation Program elevation data having a nominal cell size of 10 m. This research demonstrates the utility of open source tools with parallel computing support for extracting connected drainage network patterns and handling depressions in 30 subbasins distributed across humid, dry, and transitional climate regions and in terrain conditions exhibiting a range of slopes. Special attention is given to low-slope terrain, where network connectivity is preserved by generating synthetic stream channels through lake and waterbody polygons. Conflation analysis compares the extracted streams with a 1:24,000-scale National Hydrography Dataset flowline network and shows that similarities are greatest for second- and higher-order tributaries.