"For the river network, nodes are defined when two or more reaches come together, or if two or more rivers drain into a marsh, and come out as a single reach downstream. The hydrology at both scales was digitized into polyline shapefiles and then combined by attaching vector lines of both scales, using the lower resolution data where the higher resolution data was not available . I then simplified the river geometry into single polylines breaking only at river confluences, eliminating any trace of the vectorization process. These confluences would become one kind of node in the river network."

"Adding nodes to the river polylines was accomplished in two steps using ET Geowizards, first by splitting the polyline into 10 km segments and then by creating a point shapefile comprised of the beginning and ending vertices of the polyline segments using ‘Renode Polyline’ . I then obtained the geographic coordinates (easting and northing in meters, from a UTM projection) and elevation (masl, from a SRTM raster) of each point in the shapefile, which serve as attributes for the nodes in network software (4).

In the process of creating a node shapefile, the ‘Renode Polyline’ Wizard application in ET Geowizards, also adds two columns to the original polyline shapefile, the ‘ET_FNode’ and ‘ET_TNode’ columns. These values refer to the nodes in the created point shapefile, essentially a list of edges for an adjacency matrix (that is, two columns specifying the connected nodes). This was exported as a text file for manipulation in UCINET.

The next step is the creation of a distance matrix for a vector of XY (easting and northing) pairs. This was accomplished by taking the locational coordinates of each river node from the ArcGIS output and producing a matrix using the stock dist() function in R. The output is a matrix in meters between each node.

The remaining operations were carried out mostly in Ucinet and NetDraw, simple network software available for matrix analysis and display (5). The edges of the river network were imported to Ucinet and then exported as a 2D array for manipulation in Microsoft Excel. In this matrix, river nodes are either connected (1) or they are not (0). To achieve a matrix with connections identifiable as river or land, however, a multiplier matrix was created in Excel. In this ‘Inverse’ matrix, produced using the ‘Find and Replace’ function in Excel, river connection cells are zero and all other cells are one. Using the ‘Command line / matrix algebra’ tool in Ucinet, the ‘Inverse’ matrix is multiplied by the distance matrix to cancel out the distance values where there are river connections. The original river matrix is then added to the distance matrix using the ‘Command line / matrix algebra’ tool. In this network, all nodes are connected to each other by ‘1’ or higher values representing the Euclidean distance between nodes in meters."