Analysis of Water Resource Networks Using Parallel Coordinates

摘要

The size and complexity of water resources networks typically require a large number of computationally intensive simulations to test effects of changes in network structure or management. Current tools can only visualize the effects of a few changes. Here, we introduce a new method and tool that uses parallel coordinates to simultaneously visualize large water resources networks and identify nodes that are vulnerable (their connectivity depends on the existence of particular nodes), topologically significant (when removed or added to the network, they cause other nodes to be vulnerable), and redundant. We apply the tool to the 56-node lower Bear River water system that stretches from southern Idaho to the Great Salt Lake, Utah. Nodes that are connected to only one other node are the most vulnerable, including Great Salt Lake, Malad River, and Evaporation from Hyrum Resevoir. The three most topologically significant nodes are Cutler and the two junctions connecting the South Cache Valley and the Weber branches to the rest of the network. There are five highly redundant node pairs with over 96% of the same connections including the Cache Valley Irrigation and Cache Valley New Municipal and Industrial service areas. This redundancy suggests that Cache Valley Irrigation is a promising source to transfer water from agriculture to urban use. The New Box Elder County Irrigation and South Cache Irrigation service areas have very low topological significance ranks and suggest that these irrigation areas may also be promising sources of water transfers. Our tool can also suggest candidate locations for dam removal (reservoirs with low topological significance or high redundancy), areas most benefiting from conservation measures (vulnerable nodes), and locations for monitoring (vulnerable nodes). Future work with this tool should incorporate flow direction and magnitude. The results of our tool can direct detailed simulation studies and sensitivity analysis. The tool scales to very large networks and identifies the most promising nodes to subsequently focus computationally-intensive simulation and sensitivity analysis efforts.