National

Aquatic Barrier Inventory & Prioritization Tool

Aquatic Barrier Tool

Network Analysis Methods

Aquatic barriers were analyzed using the National Hydrography Dataset - High Resolution Plus (NHDPlusHR) dataset from the U.S. Geological Survey. These data are defined at a 1:24,000 scale or better.

The following methods represent our best attempt to correctly snap and analyze dams and road-related barriers. If you see errors in the dam, road-related barrier, or waterfall locations used in the analysis, or would like additional help interpreting the results of the analysis, please contact us.

Aquatic Network Preparation

  1. All coastline reaches were first removed from the NHDPlusHR flowlines dataset. "Loops" were retained for purposes of snapping waterfalls and barriers below, but barriers that snapped to these loops were excluded from the network connectivity analysis.
  2. All isolated pipelines or pipelines >250 meters were removed from the flowlines dataset. Short pipelines through dams were retained, in order to ensure that these dams could be evaluated for network connectivity.
  3. All waterbodies were extracted from the NHDPlusHR waterbody dataset and intersected with the flowlines to determine flowlines that are within waterbodies. Flowlines outside these waterbodies were considered free-flowing for the network connectivity analysis.
  4. In some regions, network reaches were manually removed from the analysis (e.g., Chesapeake Bay).

Methods for Waterfalls

Waterfalls are used to define natural breaks within the aquatic network. All waterfalls included in this analysis were considered "hard" breaks to the aquatic network, which prevent upstream or downstream movement of aquatic species.

  1. Waterfalls were obtained from U.S. Geological Survey (data publication in prep.).
  2. Waterfalls were automatically snapped to the nearest flowline within a 100 meter tolerance.
  3. Waterfalls were de-duplicated within a 10 meter tolerance: those that occurred closer together than 10 meters were considered duplicates, and only the first instance of each set of duplicates was retained.
  4. Waterfalls that were successfully snapped to the aquatic network were used to cut the aquatic network. These networks formed the basis of aquatic networks that were further subdivided by dams and road-related barriers below.

Methods for Dams

The location of dams were compiled by SARP from a variety of data providers, including the National Inventory of Dams, National Anthropogenic Barrier Database, and state agencies. Due to variations in the vintage and accuracy of these datasets, not all dam points are located correctly, and some dams are duplicated between these data providers. Some of these dams were supplemented with input from partners with on the ground knowledge of specific structures.

  1. Dams that are known to have been removed for conservation purposes or that were determined to not otherwise exist were removed from the analysis.
  2. Invasive species barriers or dams that were determined not to be significant barriers based on field reconnaissance or manual review of aerial imagery were excluded from the analysis, but are displayed on the map for reference.
  3. The location of dams and dam-related features (e.g., spillways) were extracted from NHDPlusHR dataset. These were buffered by 5-10 meters and adjacent / overlapping features were dissolved in to a single feature. The intersection points between the flowlines and these dam features were used as reference points for snapping dams in the SARP inventory.
  4. For each waterbody extracted above, the downstream-most point(s) on any intersecting flowlines were extracted and used as reference points for snapping dams. These are called "drain points" below.
  5. Estimated dams based on a previous analysis by SARP using smaller waterbodies and flowlines that were not otherwise accounted for in the barrier inventory were snapped to the nearest drain point of the waterbody they were contained within, up to 2 kilometers away.
  6. Dams were snapped to the above dam-related features if they fell within 50 meters. They were snapped to the nearest intersection point between the dam-related feature and the flowlines. For very large dams, this could be up to 1km or more away from the unsnapped location.
  7. Remaining dams were snapped to the above dam-related features if they fell within 150 meters of an intersection point between that feature and the flowlines. This distance was reduced to 50 meters for dams determined to be likely off-network based on manual review.
  8. Remaining dams that occur within the waterbodies extracted above were snapped to the nearest drain point of that waterbody if they were within 150 meters (50 for those likely off-network). Any that were closer to the drain point of another waterbody were snapped to that other waterbody's drain point instead (these occur where there are chains of waterbodies), otherwise those that were within 250 meters of the drain point of the waterbody that contains them were snapped to that point.
  9. Remaining dams not in waterbodies were snapped to the nearest waterbody drain point within a tolerance of 150 meters (50 for those likely off-network).
  10. Remaining dams were snapped to the nearest flowline within a tolerance of 150 meters (50 for those likely off-network).
  11. Remaining dams that were within 50 meters of a large waterbody (>0.25 square kilometers and >1 km flowlines within waterbody) were snapped to the nearest drain point for that waterbody within a tolerance of 250 meters.
  12. Dams were de-duplicated within a tolerance of 10 meters (50 meters for those likely to be duplicates based on manual review). The dam with the most information available in the inventory was retained from each set of duplicates.
  13. Only dams that were successfully snapped to the aquatic network and were not otherwise excluded from the analysis were analyzed for their impacts to network connectivity.
  14. Aquatic networks were cut at each dam. The network topology was used to determine the upstream functional network.
  15. Metrics for network length, network complexity, percent unaltered, and natural landcover were calculated based on these functional networks.

Note: not all dams could be snapped properly. Dams that were closer to loops in the aquatic network or the intersection points between loops and NHD dam-related features or waterbodies were snapped to those loops instead of the primary aquatic network. This was done to prevent snapping dams incorrectly. In many cases where these were manually reviewed, these dams are in the correct location, but limitations of the network analysis methods prevent including loops within the analysis. Thus, these dams were not included in the analysis.

Methods for Road-Related Barriers

Only road-related barriers that have been formally assessed for impacts to aquatic organisms using a defined protocol and were determined to be a likely barrier to those organisms were included in the network connectivity analysis. These represent a small subset of the road/stream crossings (potential barriers) within the region. These barriers were analyzed within the aquatic networks already subdivided by dams and waterfalls above.

  1. Barriers were snapped to the nearest flowline within a tolerance of 50 meters.
  2. Barriers were de-duplicated within a tolerance of 10 meters. The barrier with the greatest impact to aquatic organisms was retained from each set of duplicates.
  3. Barriers within 10 meters of dams were dropped as being likely duplicates of those dams.
  4. Only barriers that were successfully snapped to the aquatic network and were not otherwise excluded from the analysis were analyzed for their impacts to network connectivity.
  5. Aquatic networks were cut at each barrier. The network topology was used to determine the upstream functional network.
  6. Metrics for network length, network complexity, percent unaltered, and natural landcover were calculated based on these functional networks.

Note: not all barriers could be snapped properly. The snapping methods above do not include the locations of road / stream crossings, which means that the snapped location of the barrier may not be precisely located on the nearest road.


For more detailed information about the analysis methods, please see the source code repository for this project.