Hundreds of years of man-made alterations to the rivers and streams of the Southeast U.S. has resulted in heavily fragmented aquatic habitat. The Southeast also has some of the highest aquatic biodiversity, which means that there are now many endangered, threatened, or other at-risk aquatic species throughout the region.
These aquatic barriers include a wide range of dams, from tiny mill-pond dams to massive hydroelectric or water management dam systems, as well as culverts and other types of road crossings over rivers and streams. While some of these barriers may include fish ladders or other systems that assist migrating fish, most do not. In some cases, these barriers have fallen into disrepair, and are now a hazard in the aquatic landscape. There is a strong need to identify what barriers even exist across the region, their condition, and which ones can be removed or otherwise mitigated to improve habitat connectivity and reduce hazards.
The Southeast Aquatic Resources Partnership (SARP) have been leading the development of a region-wide inventory of aquatic barriers for several years. They have been instrumental in forming aquatic connectivity teams in states throughout the region that focus on using this inventory to identify aquatic barriers that can be removed or mitigated. As part of this, they need to find the barriers provide a reasonable balance between increasing aquatic connectivity while remaining feasible both in terms of financial cost as well as overall socioeconomic support. For example, removing a large active hydropower dam could greatly increase the amount of connected rivers and streams in an area, but may come at a very steep economic and social cost. In contrast, an aging unused dam created several decades ago may be both feasible to remove, and contribute a significant amount of re-connected habitat.
In order to help SARP's partners better identify and prioritize aquatic barriers, we created and have continued to improve the Southeast Aquatic Barrier Prioritization Tool. I developed the initial version of this tool while at the Conservation Biology Institute. I partnered directly with SARP under this Phase II to further improve the capabilities of this tool and underlying data analysis framework.
This tool allows users to:
The amount of data that we process to make this tool possible is staggering:
Under Phase I, I worked with SARP to migrate the network analysis framework from ArcGIS to Python. In ArcGIS, a subset of the Southeast would take over a week to process, and often failed. It couldn't even process the largest subregion used in the analysis. Clearly, that wasn't a sustainable solution.
By the end of Phase I, I had created a full analysis suite implemented in Python using open source libraries, which allowed us to snap the barriers to the aquatic network, subdivide the network between the barriers, calculate various network statistics, and use that information to prioritize aquatic barriers based on metrics that characterize how much connected habitat they would contribute. While it could complete the entire Southeast over the course of several hours, running it was still a bit more manual than ideal.
Under Phase II, our goals were to improve on this framework to make it easier to process new inventory data as they become available. We also found that we needed to greatly improve our snapping and deduplication of data in order to ensure better results. The inventory is compiled from many individual datasets managed by different organizations with varying data quality, which made it somewhat common to find several nearby points that all indicated the same dam. Furthermore, for larger dams, representing a linear feature that could be more than a mile wide with single point meant that we often missed these major barriers in our analysis. They were simply too far away from the aquatic network to snap correctly, even though they have a major impact on that network.
In addition to major improvements in our snapping and deduplication functionality, I also focused on performance improvements. One of the slowest parts of working with data like these is reading and writing to traditional GIS formats. To get around that issue, I developed geofeather, which uses the excellent feather format to provide a column-oriented, very fast read / write file format. By using this for all internal geospatial files, this greatly improved the speed of the overall analysis. We can now complete the analysis across the entire region in just a couple hours.
Phase II also involved:
SARP and partners have been actively engaging with this tool to identify and prioritize barriers. They have already identified several removal projects that are in various phases of planning and have completed at least one removal project based on this tool.