GeoNet

Welcome to GeoNet

GeoNet is a computational tool for the automatic extraction of channel networks and channel heads from high resolution topography. The most recent version is GeoNet 2.2. The tool is now also available in Python (see Download page).

GeoNet combines nonlinear filtering for data preprocessing and cost minimization principles for feature extraction. The use of nonlinear filtering achieves noise removal in low gradient areas and edge enhancement in high gradient areas, i.e., near feature boundaries. After preprocessing, GeoNet extracts channels as geodesics—lines that minimize a cost function based on fundamental geomorphic characteristics of channels such as flow accumulation and curvature. Please see our Publications for details on the method.

GeoNet can be downloaded through the Download page. The package contains example data sets and scripts to run them. Information on how to run GeoNet on other data sets can be found in the How-to page.

GeoNet extraction of the Rio Col Duro river basin in the Eastern Italian Alps and comparison to the surveyed network (results obtained with GeoNet 1.0.1) [Passalacqua, Tarolli and Foufoula-Georgiou, WRR, 2010].

The latest version of GeoNet has 1 user-defined parameter:

See Sangireddy et al. (2016) for recent analyses of GeoNet's outputs and the reduction from 3 parameters to 1 (Parameters.flowThresholdForSkeleton).

- Contributing area threshold (Parameters.flowThresholdForSkeleton): The contributing area threshold is simply used to “clean up” or “thin” the curvature skeleton by eliminating pixels with very small contributing areas for the purpose of subsequent accurate channel centerline extraction of the whole network. This threshold area is NOT used to impose a channel initiation area and has to be set to a value smaller than the minimum channel initiation area. Some indication of a reasonable range can be obtained from the slope-area plot or from visual inspection of the lidar data and a crude calculation of some of the smallest channel initiation areas.