Interbasin water transfers in the United States and Canada
Interbasin water transfers have been defined many ways within the literature12,13,14 and by government agencies. For this study, we define an IBT as a human-mediated movement of surface water or groundwater from one sub-drainage area or subregion (HUC4) to another sub-drainage area or subregion through man-made or artificial pathways (e.g., canals, pipelines, aqueducts). Subregion15 and sub-drainage16 boundaries come from the United States Geological Survey (USGS) and Natural Resources Canada, respectively. We further narrow our IBT definition to exclude the transfer of treated water and wastewater due to the lack of data describing complex municipal water and wastewater distribution systems across Canada and the US. The movement of untreated (or “raw”) water between the intake location of a water distribution system and the water treatment facility is deemed an IBT if it traverses a basin boundary (i.e., sub-drainage or subregion boundary; e.g., Fig. 3a); however, if water within the distribution system crosses a basin boundary after treatment, we do not include this instance within our IBT datasets (Fig. 3b). We have also removed inconsequential drainage ditches that drain less than 0.5 square kilometers. Such drainage ditches constituted a significant fraction of previous US IBT datasets7, even though they have a negligible hydrologic, ecological, or societal impact.Fig. 3Examples of potential interbasin transfers of raw water (a) and treated water (b). Raw water transfers are represented by yellow lines, while a treated water transfer is represented by a magenta line. If raw water crosses a subregion boundary (blue lines), it is included in our dataset, as is the case for the Schoharie and Delaware Aqueducts that bring raw water for New York City public water supply (a). If only treated water crosses a subregion boundary, as is the case for Gwinnett County’s public water supply system in Georgia (b), then it is not included within our IBT datasets.Full size imageThe creation of our IBT data products involved four steps: i) data collection, ii) data standardization, iii) data visualization, and iv) data validation. The first three steps are described in this section (Methods), while data validation is described within the ‘Technical Validation’ section.Data collectionTo create a national IBT dataset, we started with potential IBTs identified by Dickson and Dzombak7. Dickson and Dzombak extracted all artificial flow paths that crossed subregion boundaries from the USGS National Hydrography Dataset (NHD). These IBTs were not verified and lacked descriptive details, such as water use purpose or transfer volume. Furthermore, the number of IBTs reported by Dickson and Dzombak is artificially large since it counts each instance a conveyance structure crosses a basin boundary as an individual IBT, even if it is part of one larger IBT project (e.g., Central Arizona Project). These records were paired with older IBT datasets produced by USGS8,9. Together, these datasets represent the most complete US IBT datasets to date. We filtered out records from the combined datasets that did not meet our IBT definition, were duplicates, or were verified as being either decommissioned or erroneous. We also connected flowlines that are part of the same IBT project.Next, we searched state and federal reports, data repositories, and websites for data describing the location, properties, and flow volumes of IBTs. Findings from these searches allowed us to remove erroneous records within our current dataset, as well as add new IBTs that were not captured by previous datasets. Mostly, though, our review of government records allowed us to confirm IBT records and to provide more complete documentation of already identified IBTs. Websites for federal agencies that have a role in building, administering, or maintaining records on IBTs, such as the USGS, US Bureau of Reclamation (USBR), US Army Corps of Engineers (USACE), and the Environmental Protection Agency (EPA), were searched for relevant records. Approval by USACE is required when building across a navigable waterway, which is sometimes required for IBTs. Much of the major federal water supply infrastructure in the Western US, including IBTs, were built and are currently operated by USBR. The EPA has records related to water distribution systems17, including water intake and treatment locations, which were used to identify IBT locations. The USGS gauge network reports time-series records for 79 IBTs. Relevant state websites for IBT data collection were identified through the survey of state-level water data platforms developed by Josset et al.18.After reviewing the scientific literature and publicly available government reports, data repositories, and websites, we contacted federal, state, and local representatives for additional data records and to verify our existing records. Federal employees at USGS and USBR reviewed and provided additional records for our initial IBT dataset. The USGS Water Use Science Project regularly collects water use and water infrastructure data from states. The USGS Water Use team helped us identify the state agency and contact person that would most likely maintain IBT data for each state.We sent IBT data requests to each state via email and phone calls. In cases where these attempts were unsuccessful, we filed an Open Records Act or Freedom of Information Act (FOIA) request to collect any remaining data we were missing. In cases where neither federal or state agencies maintained the data we sought, we contacted IBT operators directly. Direct contact with IBT operators was primarily done when collecting time-series flow data for irrigation districts and municipal water suppliers.Canadian IBT data were collected from an Access to Information Act records request. The Environment and Climate Change Canada (formerly, Environment Canada) had maintained records of IBTs throughout Canada until 2011. Several reports published by Environment Canada researchers10,11 document Canadian IBTs and their properties. These reports highlight select IBTs but do not provide complete IBT records. Our Access to Information Act request provided us an unpublished report and associated data from 2004 that described the full collection of IBTs in Canada.Data were collected between August 2019 and June 2022. Our data products reflect the most up-to-date data held by primary data collectors on the date of our request. The date each IBT entry was collected is reported in the IBT Inventory Dataset. We collected all time-series flow data available for each IBT, with some records going back as far as 1901.Data standardizationThe data we collected were in a variety of file formats and data types. We created a data standard, which we named the Interbasin Transfer Database Standard Version 1.0. (IBTDS 1.0), to provide a consistent way of representing and defining data for all IBTs. The standardized IBT Inventory Dataset follows a node-link structure. Nodes represent places of water diversion, water use, or change in flow (e.g., reservoir, channel junction). Links represent conveyance infrastructure or natural waterways that connect two or more nodes within an IBT project. Unique link identifiers (Link ID) connect two or more unique node identifiers (Node ID). One or more links constitute an IBT project. The owner/operator of each IBT project, as well as the year the IBT project was commissioned and decommissioned (if applicable), is reported within the IBT Inventory Dataset.Geospatial details are reported for each IBT project in the IBT Inventory Dataset and the IBT Geospatial Dataset. We obtained the precise latitude and longitude of each node using the various data sources noted previously, as well as visual inspection of high-resolution aerial imagery from Google Earth and Esri’s World Imagery layer. Precise geospatial information is reflected in the IBT Geospatial Dataset. The IBT Inventory Dataset lists the hydrologic and geopolitical boundaries that contain each node. For the US, the state and county name and the Federal Information Processing System (FIPS) Code is also provided for each node. Likewise, the province and Census Geographic Unit is given for each node in Canada. The IBT project name (e.g., Heron Bayou Drainage Ditch, Hennepin Canal) associated with each node and link segment is also reported.As is often the case with irrigation and drainage IBT projects, there are sometimes several relatively small, adjacent diversions/ditches along an IBT project. We focus on capturing the main components of the IBT, instead of representing dozens or even hundreds of connected small ditches that divert or collect water along the IBT project. Nonetheless, when the collective impact of these small water diversions or inputs may noticeably change IBT flows, we depict these small ditches together as a representative two-node pair connected by a link segment. One of the nodes represents approximately the middle of where these small ditches intersect with the main IBT channel. The other node is the approximate centroid of water users served by or areas drained by these small ditches. If one of the secondary channels is large relative to the main channel (i.e., ability to divert more than ~25% of the main channel flow based on channel top width or flow records), it is recorded with its own Node ID and Link ID (Fig. 4). Likewise, if a secondary channel has an official name granted by a government agency or its owner/operator, we also record this segment with its own Node IDs and Link ID(s).Fig. 4An example of an interbasin water transfer project in Arizona with major (yellow) and minor (orange) project components. The thick yellow lines represent primary components of the project that are recorded in our dataset and assigned a Link ID (white text label). The thin orange lines represent secondary or tertiary canals or ditches that are small relative to the main (yellow) project segments and are therefore not represented in our dataset. The blue lines represent HUC4 subregion boundaries.Full size imageWe record the primary, secondary, and tertiary purpose of each IBT project and these purposes are the same for all links within the IBT project. One of “water supply – public supply”, “water supply – irrigation”, “flood control”, “navigation”, “waste discharge”, “environmental flows”, “energy – hydroelectric”, “energy – thermoelectric”, “energy – mining”, “other”, or “unknown” is assigned to each IBT project based on online records, design documents, reports, and/or personal correspondence with local, state, or federal officials. Link infrastructural properties, such as whether the link is a lined canal, unlined canal, pipe/tunnel, or other structure, are recorded for each link segment.The average water transfer rate (m3/d) is reported for each link segment where this information is known. The average water transfer rate only represents flows for the identified link segment, not necessarily the entire IBT project since upstream/downstream diversions and inputs may mean flow rates are different in different portions of the project. The average water transfer rate is converted from the units provided to us but is otherwise left unchanged. The primary data records are often unclear or do not specify the time period used to estimate average water transfer rates. The IBT Inventory Dataset reports whether time-series data is available for each Link ID in the IBT Time-Series Flow Dataset.The IBTDS 1.0 data standard was also applied to the IBT Time-Series Flow Dataset. The unique Link ID identifying the location where the transferred flow rate was measured is recorded for each time-series entry, relating the time-series data records to the IBT Inventory Dataset. The recorded flow rate only represents water transfer rates for the given link segment where the measurement was made, not necessarily the entire IBT project. Time-series data describing IBT flow rates were recorded at various temporal resolutions, ranging from instantaneous gauge readings every 15 minutes to average annual records. The standardized time-series dataset converted all reported water transfers to a common measurement unit (m3) and temporal resolution (day). When available, a web link to the original data source is published with the standardized data. The original timestep which the data was collected is also reported for each entry.In a few instances, there is more than one flow measurement for a link segment. Measurements are typically reported by different agencies and the measurements do not always align perfectly, either in their quantity or frequency of their reporting. Unless one of the records is known to be erroneous or of inferior quality, both sets of records are standardized and reported. For example, USBR reports monthly water transfer volumes along the Central Arizona Project (Link ID: CAP.AZ.01), while USGS reports daily water transfer volumes for the same link segment.Data visualizationWe provide an online visualization of the IBT Geospatial Dataset using ArcGIS Online (https://virginiatech.maps.arcgis.com/apps/mapviewer/index.html?webmap=b2cfac9b70ea44e4938734da0b1a7c8e), which is also summarized in Fig. 1. Every IBT node and link segment in the IBT Inventory Dataset is included. An arrowhead at the end of a link segment depicts the flow direction of transferred water. Link segments imported into ArcGIS Online were initially represented as a straight path between connected nodes. When the IBT flowpath was visible from aerial imagery or the flowpath was available from existing sources (e.g., NHD or detailed engineering drawings), the exact path of transferred water was mapped; otherwise, the flowpath remained a straight line between connected nodes.State and federal agencies restricted some of the data we are able to share publicly. Specifically, we are not permitted to reveal the exact water intake and treatment locations of some public water suppliers. Instead of mapping the precise latitude and longitude of points of diversion, points of flow change, and points of use like with other IBTs, IBTs whose primary purpose is public water supply are depicted as a straight line connecting the centroids of subwatersheds (HUC12) where the IBT node is located. More
