Dams, reservoirs, waterpower facilities, and other water-control structures can modify streamflow and propagate their influence downstream through connected river networks, complicating hydrologic modelling and aquatic ecosystem assessment. Large-domain assessments often rely on binary agency classifications at hydrometric gauges (e.g., “natural” and “regulated”) or on storage- and discharge-based metrics that are difficult to apply consistently where operational data are incomplete. We introduce the Streamflow Alteration Index (SAI), a parsimonious, network-based screening metric that estimates the percentage of upstream drainage area whose routed flow passes through mapped alteration signal sources. Under a first-order drainage-area scaling approximation, SAI provides a proxy for the fraction of streamflow potentially exposed to upstream alteration sources, rather than a measure of the exact magnitude of hydrologic alteration. The framework comprises two paired indices, SAI_I (0 %–100 %) and SAI_II (0 %–100 %), independently representing confirmed Level I and potential Level II sources; values are propagated downstream by drainage-area-weighted mixing and reset at new sources of the same level. We compiled 2528 infrastructure records, of which 2476 were retained after geospatial alignment and screening. Following source classification and consolidation of multiple records within the same subbasin, 643 unique subbasins were identified as Level I and/or Level II alteration signal locations. The resulting SAI v1.0 database provides province-wide subbasin and river-reach layers for screening potential instream alteration at gauged and ungauged locations. We evaluated SAI-derived natural, conditional, and altered classes against 148 Reference Hydrometric Basin Network (RHBN) gauges, 125 additional Water Survey of Canada (WSC) gauges labelled natural, and 215 WSC-regulated gauges. SAI classified 86 % of both natural reference groups as natural, with no RHBN gauge classified as altered. Among WSC-regulated gauges, 26 % were classified as altered, 40 % as conditional, and 34 % as natural, highlighting differences between binary agency labels and the SAI screening framework and identifying sites requiring follow-up assessment. SAI v1.0 therefore provides a practical, transparent, and reproducible screening baseline for large-domain hydrologic and ecological applications without requiring detailed operating rules and offers a scalable framework for broader network-based assessments of potential streamflow alteration. The SAI v1.0 dataset is available from Zenodo (Shen et al., 2026; 10.5281/zenodo.19547932).
Ministry of Natural ResourcesCollaborative Research Agreement titled: Meeting business needs of the Ministry of Northern Development, Mines, Natural Resources and Forestry with the Raven Hydrological Modelling FrameworkIntroduction
Dams and reservoirs are among the most significant infrastructural interventions in freshwater systems, constructed to serve a range of human needs, including water supply, flood and drought control, irrigation, hydropower generation, and recreation (Lehner et al., 2011; Maavara et al., 2020; Palmer and Ruhi, 2019; WCD, 2000). Global Dam Watch (GDW) v1.0 compiles 41 145 documented river barrier locations and 35 295 associated reservoir polygons, representing a cumulative storage capacity of approximately 7420 km3 and an artificial terrestrial surface water area of approximately 304 600 km2 (Lehner et al., 2024). GDW records are thought to be nearly complete for reservoirs between 10 and 1000 km2, but become increasingly incomplete below 10 km2, with almost no reservoirs smaller than 0.1 km2 represented (Lehner et al., 2024). Relative to the millions of natural lakes around the world, these mapped reservoirs have artificially expanded the estimated global lake storage volume by about 4 % and global lake surface area by about 11 % (Lehner et al., 2024).
Despite their societal benefits, the widespread construction and operation of dams and reservoirs have profoundly transformed natural flow regimes, altered sediment and nutrient fluxes, and generated substantial ecological and social trade-offs (Richter et al., 2010; Winemiller et al., 2016). Global assessments have documented widespread river fragmentation and flow alteration associated with dams, reservoirs, and water withdrawals (Döll et al., 2009; Grill et al., 2019; Nilsson et al., 2005). Dams and reservoirs can modify the magnitude, timing, frequency, duration, and rates of change of streamflow by storing, releasing, and diverting water. Such streamflow alteration is widely regarded as a first-order impact on fluvial systems, because flow regimes are fundamental drivers of riverine habitats and ecosystem processes (Burke et al., 2009; Jorde et al., 2007; Petts, 1984; Poff et al., 1997, 2010). Reservoir effects vary with storage capacity, operating practices, evaporation losses, hydroclimatic conditions, and network position, but commonly include attenuation of flood peaks and redistribution of flows through time; low flow responses may involve either augmentation or depletion (Brogan et al., 2022; Habets et al., 2018).
In this study, streamflow alteration denotes human-caused changes to natural streamflow patterns arising from instream structures and lake-outflow controls, primarily through the construction and operation of dams, reservoirs, and waterpower facilities. Hereafter, the term “alteration” is used throughout, except where “regulation” is retained as part of established terminology, such as degree of regulation and the natural and regulated classifications used in HYDAT (Water Survey of Canada, 2025), or when following the terminology of cited studies.
Streamflow alterations occur across temporal scales ranging from minutes to seasons, with particularly strong sub-daily variability downstream of hydropeaking waterpower facilities (WPFs). Hydropeaking refers to sub-daily operational changes in discharge used to match electricity generation to demand, which commonly produce higher releases during daytime, when electricity demand is greatest, and lower releases at night. This creates highly variable downstream hydraulic and habitat conditions at short-time scales (Bejarano et al., 2018; Kern et al., 2012). Because these fluctuations happen on sub-daily time scales, daily discharge records can mask key features of alteration such as ramping rates, the frequency of peaks, and the duration of high- and low-flow periods (Smokorowski, 2022). At longer (seasonal-to-annual) scales, reservoir operations commonly suppress flood peaks during wet periods and augment low flows during dry periods, which partially decouples discharge from natural hydroclimatic variability (Déry et al., 2016). The magnitude and spatial footprint of these alterations depend on interacting factors, including project purpose (e.g., flood control, waterpower, recreation), reservoir storage and dam operating rules, river-network configuration (e.g., reservoir size relative to inflow, cascades of dams and diversions, and channel hydraulics), and basin physiography and hydroclimate (Burke et al., 2009; Jorde et al., 2007).
A recent systematic review of 155 studies by Bipa et al. (2024) further showed that hydropeaking affects more than discharge alone: it can also modify water temperature, sediment processes, and dissolved gas conditions, with well-documented consequences for fish (e.g., movement, spawning, migration), benthic macroinvertebrates, and riparian vegetation. Global reliance on dams is expected to remain high or even increase in response to growing water stress, energy demand, and climate change, making robust assessment of dam impacts increasingly important (Steyaert et al., 2025; Tharme, 2003).
Streamflow alteration poses a persistent challenge for hydrologic modelling because many model applications assume natural conditions, whereas regulated flows reflect operational decisions in addition to hydroclimate and physiography. In process-based models, reservoir operations and other human-water processes are often simplified and poorly constrained, particularly at large scales, which can bias parameter estimation, reduce model transferability, and undermine predictions under changing conditions (Dang et al., 2020; Shrestha et al., 2024; Wada et al., 2017; Yassin et al., 2019; Zajac et al., 2017). Similar considerations apply to data-driven models: static catchment attributes can improve generalization across basins (Kratzert et al., 2019a, b), while regulation descriptors have been shown to improve prediction skill in human-altered catchments (Tursun et al., 2024). Flow alteration can also weaken assumptions of hydrologic similarity used in regionalization, complicating prediction in poorly gauged or ungauged basins (Blöschl et al., 2013; Dang et al., 2020; Du et al., 2022; Hrachowitz et al., 2013; Montanari et al., 2013). Recent data-driven studies further demonstrate that reference hydrology and hydrologic alteration can be reconstructed across large stream networks using gauged observations and geospatial predictors, enabling reach-scale estimates at ungauged locations (Lueders et al., 2026). These developments reinforce the need for scalable alteration assessments, consistent with regional environmental-flow frameworks that recognize the impracticality of evaluating rivers individually across large domains (Poff et al., 2010).
The impacts of alteration caused by dams and reservoirs on hydrology have been widely documented in recent decades (Déry et al., 2016; Grill et al., 2014; Habets et al., 2018; Lehner et al., 2011; Richter et al., 2010; Yang et al., 2022). In parallel, numerous indices have been developed to quantify regulation intensity and related impacts across river systems, ranging from storage-based metrics (e.g., degree of regulation) to fragmentation and connectivity metrics used in large-scale assessments (e.g., Nilsson et al., 2005; Grill et al., 2014), including reach-based, decadal dam impact metrics that propagate cumulative upstream storage effects through river networks (e.g., Wieczorek et al., 2021). Physically based indices have also been developed specifically to estimate flood peak attenuation by multiple reservoirs at the catchment scale. Cipollini et al. (2022) introduced an equivalent-reservoir index that accounts for reservoir number, relative location, storage characteristics, and climatic variability, while Cipollini et al. (2023) extended this framework to generic catchment geometries and arbitrary reservoir locations.
One of the most widely used storage-based measures is the degree of regulation (DOR), commonly defined as the cumulative upstream reservoir storage capacity relative to mean annual flow volume. Higher DOR values indicate greater potential for reservoirs to regulate downstream flow. DOR has been mapped along river networks and is included as an attribute in several global dam, reservoir, and catchment datasets (Lehner et al., 2011, 2024; Nilsson et al., 2005; Yin et al., 2024). However, like all metrics, DOR has several limitations. First, the use of total or gross reservoir capacity, including inactive or dead storage, may overestimate the storage volume that is operationally available to modify streamflow (Kumar et al., 2022; Lehner et al., 2011). Second, large-domain DOR estimates often depend on incomplete or empirically inferred reservoir capacities and modelled mean annual flow, introducing uncertainty into the resulting values (Lehner et al., 2011). Third, DOR is fundamentally a storage-based indicator of potential regulation and does not distinguish among reservoir purposes, operating regimes, seasonal management practices, or the timing and magnitude of releases. Consequently, similar DOR values may be associated with substantially different forms and magnitudes of hydrologic alteration. DOR is therefore useful for characterizing potential storage-based regulation, but it is not a direct measure of realized streamflow alteration.
Another commonly used concept for characterizing dam and reservoir impacts is the zone of influence (ZOI), which delineates the upstream and downstream river-network areas affected by a dam, reservoir, or WPF (SEI International, 2009; Stainton, 2012). The hydrologic ZOI provides a practical framing of the spatial footprint of flow regulation beyond the structure itself. The ZOI concept is also relevant in broader river management contexts, since flow alteration footprints can inform aquatic ecosystem assessment and impact evaluation, although the ecological consequences of regulation are not necessarily bounded by simple hydrologic distance- or area-based rules (Metcalfe et al., 2013).
Stainton (2012) noted that the upstream ZOI is often straightforward to delineate (e.g., reservoir inundation), whereas the downstream ZOI is more difficult to define because it lacks a clear boundary and depends on operating regime, river hydraulics, and basin context. In this framing, the downstream ZOI extends from the tailrace to the point where operationally induced alterations become sufficiently attenuated or diluted that they are no longer discernible from background conditions. Stainton (2012) reviewed simple drainage area-based prediction rules, including a catchment-ratio threshold adapted from Petts (1978, 1979, 1980) and the British Columbia “area of impact” multiplier (Lewis et al., 2004), but showed that these heuristics can substantially under- or over-estimate influence when tested in Ontario basins. These studies highlight both the usefulness and the limitations of simple drainage area-based heuristics for delineating downstream alteration footprints and point to the need for approaches that more explicitly represent cumulative network effects and are better suited to large-sample hydrologic applications.
The decadal dam impact (DI) metric of Wieczorek et al. (2021) characterizes the degree of dam regulation for individual river reaches over time based on cumulative upstream reservoir storage, the relative drainage area represented by upstream dams, and the volume of precipitation within the reach drainage area. Unlike reservoir-level indices, DI is defined at the river-reach scale and explicitly represents cumulative upstream effects through the river network. This reach-based, temporally structured formulation is a major strength, as it moves beyond single-reservoir metrics and enables large-scale assessment of potential cumulative dam influence. However, DI remains fundamentally dependent on reservoir storage information and thus inherits limitations common to storage-based approaches, including uncertainty in reported or inferred storage capacity and the fact that storage alone cannot fully represent reservoir operations or actual streamflow alteration. Although DI is available for millions of river reaches across the conterminous United States (US), its transferability to Canada and other regions is constrained by incomplete reservoir storage information and by the current US-only coverage of the published dataset.
Ontario, Canada, provides a useful domain for developing and evaluating a new and improved streamflow alteration index. The province combines large and hydrologically diverse lake and river networks with extensive hydrometric monitoring and numerous dams, waterpower facilities, and other water-control structures, creating a practical need to identify locations potentially affected by upstream alteration across a large domain. Such identification is key for hydrologic modelling and other water resources applications. The Ontario Lake and River Routing Product Version 2 (OLRRP v2) provides a high-resolution, topologically connected hydrographic network suitable for spatial propagation of alteration signals, while detailed facility-specific storage and operating information remains incomplete for many locations (Shen et al., 2024). This combination of practical need, detailed network information, and incomplete operational data makes Ontario well suited for developing a parsimonious, network-based approach to estimating streamflow alteration.
To address this need, we introduce the Streamflow Alteration Index (SAI), a relatively simple network-based metric for screening and mapping potential streamflow alteration. SAI represents the potential influence of mapped instream alteration sources at any subbasin outlet along the river network. Specifically, SAI is defined as the percentage of upstream drainage area whose routed flow passes through one or more mapped alteration sources and, under a first-order drainage-area scaling approximation, provides a proxy for the fraction of streamflow potentially exposed to those sources. By propagating alteration signals downstream through the river network, SAI addresses a key limitation of many traditional metrics that focus on individual dams or reservoirs and do not explicitly represent the downstream propagation of their potential influence.
Moreover, SAI is designed to be practical in data-limited settings. Storage- and discharge-based indices such as DOR and DI require quantities such as reservoir capacity and mean flow, while physically based flood-attenuation indices require additional reservoir and catchment response parameters, such as storage coefficients, travel-time characteristics, and climatic inputs (Cipollini et al., 2022, 2023). By contrast, SAI requires two primary types of information: a topologically connected lake–river–subbasin network with drainage-area attributes and a georeferenced inventory of candidate flow-control structures with metadata sufficient to classify their likely alteration influence. In the Ontario application, candidate sources are classified using documented source attributes including facility operation type, reservoir purpose, and dam ownership. Once the sources are classified, SAI is calculated through a static, deterministic traversal of the river network, with alteration signals propagated downstream using drainage-area-weighted mixing and then reset when a new source of the same level is encountered. This parsimonious formulation avoids requiring consistently available quantitative information such as reservoir storage capacity, mean discharge, detailed operating rules, release time series, or reservoir-response parameters, thereby enabling consistent application across large domains where such information is incomplete or unavailable. We developed the SAI v1.0 database for Ontario using OLRRP v2, which discretizes the province into nearly 250 000 subbasins with an average area of 4.3 km2 (Shen et al., 2024). The database provides subbasin-scale SAI maps and associated GIS layers that support the classification of subbasin outlets and hydrometric gauge locations as natural, conditional, or altered, together with a harmonized table describing alteration signal sources.
It should be noted that the present SAI framework focuses on flow alteration arising from instream structures and lake outflow control structures, including dams, reservoirs, waterpower facilities, and lockstations, that modify water levels, storage, and releases. Accordingly, it does not explicitly account for other anthropogenic influences on streamflow, such as wastewater treatment plant effluent, industrial discharges, irrigation withdrawals and return flows, managed aquifer storage and recovery, landscape change (e.g., urbanization), or inter-basin transfers and canal or diversion systems. Their relative importance varies among basins and may become substantial during low-flow periods. These mechanisms are not incorporated here because consistent province-wide datasets are not available for all source types. Nevertheless, the SAI framework could be extended to include them in future applications where suitable databases and network attribution become available.
The remainder of this paper is structured as follows. Section 2 details the data sources and methodology for the development of the SAI v1.0 database for Ontario, as well as the approach to evaluating the database. Section 3 presents the SAI v1.0 database and the technical evaluation results. Section 4 discusses the intended use, limitations, and broader applicability of the dataset. Section 7 summarizes the main conclusions.
Materials and MethodsData Sources
Six main data-source groups were used to develop and evaluate the SAI database for Ontario:
Ontario Lake and River Routing Product Version 2 (OLRRP v2). The OLRRP v2 is a high-resolution, vector-based lake and river routing GIS product over the Province of Ontario (Shen et al., 2024) and is an update to OLRRP v1 introduced by Han et al. (2023). It supplies network topology and various subbasin, channel, and lake attributes required for hydrological modelling. Shen et al. (2024) defined OLRRP v2 subbasin outlets at locations reaching a flow accumulation threshold of 5000 30 m × 30 m grid cells, at lake outlets, or at points of interest (POIs) included in OLRRP v2. These criteria result in an average subbasin area of approximately 4.3 km2, ensure that each represented lake is fully contained within a single subbasin, and allow a complete upstream contributing catchment to be delineated for each POI. The OLRRP v2 connects 245 576 subbasins and 82 928 lakes (surface areas ranging from 0.1 km2 to 4506 km2) and assigns 4033 points of interest (POIs) to these subbasins and lakes. Lake polygons and attributes were obtained from the HydroLAKES database (Messager et al., 2016). Each POI was treated as a pour point during the delineation process, ensuring accurate placement at subbasin outlets. These POIs were derived from multiple point-based datasets, including hydrometric monitoring gauges from the Water Survey of Canada (WSC). The OLRRP v2 demonstrated high spatial accuracy for POI locations, with 74 % of WSC gauges showing less than 2 % difference between their delineated and documented drainage areas. Only 4 % of WSC gauges exhibit absolute drainage area differences greater than 10 % (Shen et al., 2024). Overall, this product provides a high-resolution lake and river routing network for SAI development.
Ontario dams. Two provincial dam datasets provided by the Ontario Ministry of Natural Resources (MNR) were used: (1) the Ontario Dam Inventory (ODI), a point-based inventory of medium and large dams throughout Ontario that provides dam names and ownership information (Ontario Ministry of Natural Resources, 2014). We used the version accessed in January 2024. The ODI contains 1839 records representing dams owned and managed by a variety of entities, including government agencies at federal, provincial, and municipal levels, conservation authorities, Indigenous communities, private owners, and Ontario Power Generation; and (2) a supplementary dataset of 393 Ontario MNR-operated dams that contains two additional attributes: dam operation status (operational or decommissioned) and dam purpose (e.g., flood control, low-flow augmentation, navigation, and recreation).
Waterpower facilities (WPFs). We used the Waterpower Generation Stations dataset maintained by the Ontario MNR (we used the version published in December 2024) to characterize potential streamflow alteration associated with waterpower infrastructure. For all 223 retrieved stations, the dataset provides an operation type field (OPERATION_TYPE), which distinguishes run-of-river facilities from plants designed for load-following (e.g., peak-load, intermediate) or pumped-storage operation. This field provides a general indication of how a facility is typically operated and of the management objectives that may influence water levels and downstream releases. The source also provides information on the purpose of the associated dam or reservoir (RESERVOIR_PURPOSE), where available, including whether waterpower is an identified purpose. In addition, we compiled the Water Management Plan (WMP) type associated with each river system by parsing the plan name, river name, and plan type reported in the Maintaining Water Management Plans: Technical Bulletin (Ministry of Natural Resources and Forestry, 2016). The bulletin classifies WMPs as simple or complex, providing a system-level indicator of operational context and the extent of coordinated water-level and flow management.
Lockstations. Lockstations were incorporated to represent flow regulation associated with navigational water-level management in Ontario. We digitized 73 lockstations located on the Trent-Severn Waterway, the Rideau Canal, and the Welland Canal, which collectively include the lock systems in the Trent River-Rice Lake-Kawartha Lakes-Lake Simcoe corridor, along the Rideau Canal, and along the Welland Canal between Lake Ontario and Lake Erie. Lockstation locations and supporting site information were obtained from the official Parks Canada lockstation directories and associated lockstation pages for the Trent-Severn Waterway and Rideau Canal, and from official Great Lakes St. Lawrence Seaway sources for the Welland Canal locks. These lockstations were treated as point-based infrastructure that can alter downstream streamflow through operational control structures (e.g., lock gates and related water-control works) and were integrated into the routing network as part of the SAI v1.0 alteration signal inventory.
Hydrometric Gauges. We compiled 1015 hydrometric gauges from the Water Survey of Canada (WSC) that are embedded in the OLRRP v2 POI layer. Many of these gauges carry a WSC regulated/natural flag (Water Survey of Canada, 2025). However, the criteria and rationale used to assign these classifications are not fully documented, which can introduce uncertainty in analyses that distinguish regulated from natural conditions using WSC flags alone. Despite this limitation, the WSC regulation flag remains widely used in Canadian hydrological studies since it is currently the only publicly available, nationwide indicator of regulation status (Arsenault et al., 2020; de Rham et al., 2020). Here, we compare the WSC binary flag with SAI-derived conditions at gauge locations to assess their consistency and illustrate the additional spatial information provided by a network-propagated streamflow alteration metric. Of these 1015 gauges, 488 met the gauge-status, record-length, and diversion-screening criteria described in Sect. 2.5 and were retained for evaluating SAI v1.0.
Reference Hydrometric Basin Network (RHBN). The RHBN is a subset of Canada's national hydrometric monitoring stations, maintained by the WSC. It features high-quality, long-term data from catchments with minimal human impacts, which are essential for understanding natural hydrological variability and serve as a benchmark for climate-related studies (Brimley et al., 1999; Ouarda et al., 1999). The RHBN, updated in 2020, includes 1027 stations across Canada, of which 154 stations are within Ontario (Pellerin, 2019). Of these, 148 met the gauge-status, record-length, and diversion-screening criteria described in Sect. 2.5 and were retained as the primary natural-reference group for evaluating SAI v1.0.
The six main data-source groups used in this study differ in scope and completeness, because they were developed for different hydrographic, infrastructure, monitoring, and reference network purposes. The OLRRP v2 provides province-wide river network coverage, but its HydroLAKES-derived lake layer only includes waterbodies with surface areas of 0.1 km2 or greater; smaller ponds, including waterbodies smaller than 0.01 km2 as considered by Schmadel et al. (2019), are therefore omitted. The Ontario Dam Inventory (ODI) focuses on medium-to-large dams, whereas the complementary Ontario MNR dataset includes only MNR-operated dams. Some small water-control structures, including culverts and small dams, may be underrepresented in the ODI and MNR dam datasets. The waterpower facility dataset provides broad coverage of known generation stations, although the availability and completeness of operational attributes vary among facilities. The compiled lockstation dataset covers the Trent-Severn Waterway, Rideau Canal, and Welland Canal, which include the major lock systems represented in this study. The WSC gauge network provides extensive federal hydrometric coverage but excludes gauges operated exclusively by other agencies, and its natural/regulated metadata are not fully documented for all stations.
Despite these limitations, these sources represent, to our knowledge, the best available and most spatially comprehensive province-wide datasets that can be consistently integrated for SAI development. Missing or incompletely documented water-control structures may lead to underrepresentation of alteration sources; the implications of these omissions for SAI are discussed further in Sect. 4.
Future versions of SAI in Ontario could reduce these uncertainties by incorporating higher-resolution hydrographic data, updated provincial and local inventories of dams and water-control structures, and more complete facility-specific operational metadata as these datasets become available. For applications beyond Ontario, implementation of the basic SAI propagation framework requires, at minimum, a topologically connected lake–river–subbasin network with drainage area attributes and a georeferenced inventory of dams, reservoirs, waterpower facilities, and other flow control structures that can be assigned to the network. Reproducing the SAI framework would additionally benefit from facility-specific information on operational status, purpose, regulation type, reservoir storage, and other indicators of expected alteration magnitude. Independent hydrometric classifications and streamflow observations would further support evaluation of the resulting SAI product.
Geospatial Alignment and Screening of Candidate Records
The Streamflow Alteration Index (SAI) is calculated at the OLRRP subbasin scale, with subbasins roughly 4.3 km2 in area on average. Accurate assignment of alteration signal sources, including dams, reservoirs, waterpower facilities and lockstations, to the correct OLRRP subbasins is therefore essential. We refer to this geospatial alignment process as snapping.
Alteration signal sources and hydrometric gauges were initially represented as point features. Since source datasets contain positional uncertainty, raw point locations do not always coincide with the correct rivers, lakes, or subbasin outlets in OLRRP. We aligned these points to the OLRRP network using a two-stage workflow: (1) automated snapping to the most plausible topologically consistent location, followed by (2) detailed manual review. Snapping was constrained by OLRRP lake and subbasin topology, feature names (string matching), and agency-documented drainage areas where available. Manual verification compared the snapped locations against satellite imagery and online basemaps (Google Maps and Esri Basemaps; Google, 2024; Esri, 2024).
Because the OLRRP network is pre-delineated and fixed, some points, particularly small run-of-river structures, may remain located within a subbasin after snapping rather than exactly at its outlet. Given the high spatial resolution of OLRRP, such within-subbasin placement is expected to introduce limited spatial error in the SAI calculation.
Points located on the main stems of the Ottawa and St. Lawrence rivers were excluded where their upstream contributing areas extended beyond the OLRRP domain and were therefore not completely delineated within the network. Records were also excluded where substantial spatial discrepancies could not be resolved through manual review or supporting information. A total of 2528 candidate infrastructure records were compiled before geospatial screening. Following geospatial alignment and screening, 211 of 223 waterpower facility records, 1803 of 1839 Ontario Dam Inventory records, 389 of 393 MNR-operated dam records, and all 73 lockstation records were retained. In total, 2476 records were retained and 52 were excluded, corresponding to an overall inclusion rate of approximately 98 %.
The mean and median displacements between the original source coordinates and the final network-aligned locations were 77 and 15 m, respectively, for ODI records; 54 and 33 m for WPF records; and 32 and 13 m for MNR-operated dam records. Lockstations were digitized directly at their corresponding subbasin outlets and therefore required no additional spatial adjustment.
At least one retained ODI record and one retained MNR-operated dam record co-occurred in 347 OLRRP subbasins. Because a subbasin may contain multiple dam records within either dataset or across the two datasets, and the available information does not support reliable one-to-one matching among individual records, these 347 subbasins should not be interpreted as representing 347 duplicate dams. Accordingly, no deduplicated dam count was derived from this subbasin-level overlap. The final retained record counts are summarized in Table 1. Unless otherwise noted, all subsequent counts and statistics are based on the 2476 retained records.
Data sources, retained record counts after geospatial screening, and criteria used to classify candidate records as confirmed (Level I), potential (Level II), or negligible (Level III) alteration signals. Note: Alteration levels are assigned using dataset-specific evidence for streamflow alteration rather than facility type or reported purpose alone. A waterpower-related purpose is considered a Level I indicator only when reported in the dedicated WPF dataset, where more detailed operational information is available.
DatasetsDescriptionsNumber of RecordsLevel I CriteriaLevel II CriteriaLevel III CriteriaWaterpower Facilities (WPFs)Waterpower generation stations obtained from the Ontario MNR. Generation stations contain attributes such as the operation type (e.g., run-of-river, peaking, etc.), associated reservoir purposes (e.g., flood control, navigation, etc.), and water management plan type (complex vs. simple) (Ministry of Natural Resources and Forestry, 2016)211Associated with a complex Water Management Plan; classified as a pumped-storage, peak-load, or intermediate plant; or the purpose of the associated reservoir is reported as waterpower.All remaining generation stations not included in Level INot applicableOntario Dam Inventory (ODI)The ODI provides medium- to large-sized dams along with attributes including their ownership and a name string across Ontario.1803Not applicableDams containing keyword “control dam” in their dam names; or dams owned by conservation authorities or Ontario Power Generation (OPG)All remaining ODI records not included in Level IIMNR DamsThe Ontario MNR-operated dam dataset contains additional information not available in the ODI, such as dam operation status and dam purpose.389Not applicableOperational dams whose stated purpose includes flood control and/or low-flow augmentation; decommissioned and non-operational dams are not assigned to Level II.All dams not included in Level IILockstationsLockstations from the Trent-Severn Waterway, the Rideau Canal, and the Welland Canal are included as point-based signals representing navigational flow control structures.73All lockstationsNot applicableNot applicableIdentification of Alteration Signal Sources
Alteration signal sources refer to point-based locations where streamflow is first modified by instream infrastructure, represented here by the outlet location of a dam, reservoir, waterpower facility, or lockstation. It should be noted that not all structures captured in available inventories meaningfully alter downstream streamflow. For example, small barriers (e.g., weirs) constructed primarily for local water-level augmentation have limited influence on discharge compared with actively managed reservoirs (e.g., facilities operated for flood control or waterpower generation). Consequently, a key step in developing the SAI database is to filter out sources that are likely to have minor impacts and retain those expected to produce substantial downstream alteration.
At the provincial scale, however, detailed operating rules and regulatory planning information are not consistently available for the thousands of structures managed by diverse entities (e.g., federal and provincial agencies, conservation authorities, the Ontario Ministry of Natural Resources, and private owners). Our assessment of minor versus major alteration therefore relies on information that is broadly available and consistently documented across datasets. We classified each candidate alteration signal source into one of three levels: confirmed (Level I), potential (Level II), and negligible (Level III). Level I sources are identified with high confidence as substantially modifying downstream streamflow. Level II sources are likely to affect streamflow but with greater uncertainty in the presence and magnitude of alteration, and a smaller expected influence than Level I. Level III sources are assumed unlikely to meaningfully alter streamflow and are treated as negligible or insignificant alteration sources for SAI mapping. These categories are summarized in Table 1, and Level I and Level II alteration signal sources are described further below. The classification is evidence-based and dataset-specific, reflecting the strength and consistency of the information available from each source dataset rather than facility type or reported purpose alone. Consequently, similar facility attributes reported in different inventories do not necessarily provide the same level of evidence for streamflow alteration.
Following geospatial alignment and screening, the 2476 retained candidate records were evaluated to determine whether they represented confirmed, potential, or negligible streamflow alteration signals. For SAI initialization (Sect. 2.4), multiple records within the same subbasin were consolidated by source level, while subbasins containing both Level I and Level II records retained both classifications. This resulted in 643 unique subbasins containing at least one Level I and/or Level II alteration signal source. The spatial distributions of the compiled infrastructure records and the resulting Level I–III alteration signal classifications are shown in Figs. A1 and A2 in the Appendix, respectively.
Confirmed alteration signal sources (Level I). Level I signals represent locations where streamflow alteration is expected to be substantial and is supported by strong evidence. For waterpower facilities (WPFs), we classify a facility as a Level I alteration source based on three operational indicators: (1) association with a complex Water Management Plan (WMP), which is typically developed for river systems with multiple WPFs and/or other control structures where coordinated operations exert substantial control over water levels and releases (Ministry of Natural Resources and Forestry, 2016); (2) an operation type reported as pumped-storage, peak-load, or intermediate; or (3) the purpose of the associated reservoir is reported as waterpower in the WPF source dataset. Moreover, all lockstations are treated as Level I signals, as they are operated as water-control structures for navigation and water level management. Accordingly, a waterpower-related purpose reported in the broader ODI or MNR dam inventories is not, by itself, considered sufficient evidence for a Level I classification because these inventories do not provide the same consistently documented operational information available for WPF records.
Potential alteration signal sources (Level II). Since complete and consistently documented information on the operational significance of individual control structures is rarely available at large scales (e.g., generally unavailable in Canada), we define Level II sources as features that are likely to influence streamflow but for which the magnitude and persistence of alteration are more uncertain and are generally expected to be smaller than those associated with Level I sources. Level II sources are identified using three complementary criteria: (1) All remaining WPFs in the provincial dataset that are not classified as Level I (i.e., facilities lacking strong indicators of substantial alteration but still plausibly capable of affecting flows); (2) Structures in the ODI whose metadata suggest control functions, including dams with names containing the keyword “control dam” and dams owned by conservation authorities or Ontario Power Generation (OPG). We include OPG-owned dams as potential alteration sources because OPG operates a large portfolio of waterpower infrastructure in Ontario, and ownership by a major waterpower operator is a practical proxy for the presence of active water-level and release management even when facility-specific operational details are not available; and (3) MNR dams that are operational (excluding decommissioned or non-operational structures) and whose stated purpose includes flood control and/or low-flow augmentation. Together, these criteria provide a pragmatic, evidence-based way to capture plausible alteration signals where detailed operational data are unavailable.
Negligible alteration sources (Level III). Level III sources represent inventoried control structures that are unlikely to meaningfully alter streamflow and are therefore treated as negligible alteration in this study. After identifying Level I and Level II alteration signal sources, all remaining structures in the ODI and the MNR dam dataset are assigned to Level III. These features may include small or pass-through structures, localized controls, or infrastructure with limited storage or operational influence, for which available metadata do not indicate active alteration (e.g., flood control, low-flow augmentation, or waterpower-related management). Level III records are retained in the archived source table to document the screening decisions, but they are not used to initialize or propagate SAI values.
Streamflow Alteration Index (SAI)
In many hydrologic studies, including modelling applications, whether a location (e.g., a hydrometric gauge) is affected by upstream alteration is often represented using a binary agency flag (natural vs. regulated), while detailed reconstruction of dam operating rules is uncommon in large-scale applications. Although explicit reservoir operating rules can improve model performance in regulated basins (Craig et al., 2020; Hughes et al., 2021; Salwey et al., 2024; Yassin et al., 2019), these data are rarely available consistently at large scales. Such practical constraints motivate spatially explicit, network-aware indicators of alteration that can be derived from limited but widely available information.
Our Streamflow Alteration Index (SAI) quantifies potential upstream alteration influence at locations along a river network, represented in this study by subbasin outlets. SAI is an area-based metric that propagates alteration signals through the drainage network, summarizing the percentage of upstream drainage area whose routed flow passes through locations containing alteration signal sources. The drainage-area weighting assumes that long-term streamflow is approximately proportional to contributing drainage area.
Drainage-area scaling has long been used as a parsimonious approach for transferring streamflow information from gauged to ungauged locations, including in regional water-balance applications (Fry et al., 2013). Empirical studies have also identified simple scaling relationships between annual mean streamflow and drainage area across Canada (Yue and Gan, 2004), while more recent analysis across minimally disturbed catchments in the United States and Canada found approximately linear discharge–drainage area scaling for most hydrologic regions, with regional departures associated with runoff efficiency and hydroclimatic conditions (Erikson et al., 2024). To assess the applicability of this approximation within Ontario, we additionally examined the relationship between long-term mean annual discharge and drainage area at 327 WSC discharge gauges with at least 10 valid years of observations during 1990–2024. Because gauges were included irrespective of regulation status, this analysis characterizes the observed discharge–drainage area relationship across the broader Ontario hydrometric network rather than natural runoff scaling alone. The fitted power-law exponents (β) were close to unity for Ontario as a whole and for three broad regional groupings, indicating strong, near-linear scaling at the spatial and temporal scales relevant to the present application (β= 0.936–1.013; R2= 0.967–0.991; Appendix B).
Accordingly, SAI uses proportional drainage-area scaling as a first-order mixing approximation rather than as a physical model of runoff generation. Under this approximation, a SAI value of 30 % at a subbasin outlet on the river network is interpreted to mean that approximately 30 % of the streamflow has passed at least one mapped upstream alteration source. We compute SAI at the subbasin scale (at the subbasin outlet points only) to support spatially complete mapping of potential instream alteration across large domains, including ungauged basins.
Conceptual illustration of the streamflow alteration index (SAI) propagation along a routing network. (a) A simplified subbasin-river schematic (lakes omitted for brevity) showing the main stem (A → B → C → D → E → F → G → H → I → J) with two tributaries (K → L and M → N) joining at subbasins E and J. Four alteration signal sources are located in subbasins B, E, G and M (filled triangle: Level I; open triangle: Level II). Subbasins are overlaid with equal-sized dashed grid cells to facilitate visual comparison of contributing areas, and blue numbers in the lower-right corners denote drainage area at the subbasin outlet (expressed as the number of dashed grid cells). (b) Example longitudinal profiles of SAI_I (red) and SAI_II (orange) along the main stem, illustrating (1) downstream attenuation (dilution) of alteration signals with increasing contributing area and tributary inflows, and (2) SAI resetting to 100 % when the flow passes a new same-level signal source (e.g., SAI_I resets to 100 % at the outlet of subbasin G). Note that the lines connecting successive subbasin outlets are shown only to visualize changes along the main stem; the SAI calculation itself is performed only at discrete subbasin outlets. The paired points at subbasins E and J show SAI values immediately before and after tributary confluence. The distance shown on the x-axis in (b) is approximate, since the river segments in (a) are schematic and not drawn to scale.
To reflect differences in confidence and expected influence among alteration sources, we compute two complementary indices: SAI_I, representing influence associated with confirmed (Level I) sources, and SAI_II, representing influence associated with potential (Level II) sources. The two indices are calculated independently: SAI_I is derived from Level I sources only and ranges from 0 % to 100 %, whereas SAI_II is derived from Level II sources only and ranges from 0 % to 100 %. Figure 1 illustrates the SAI calculations using a hypothetical subbasin-river network. The calculation procedure consists of three steps.
Step 1: Identify alteration signal sources. Identify all Level I and Level II alteration signal sources for the basin (subbasins B, E, G and M in Fig. 1a). Signal source classification is described in Sect. 2.3.
Step 2: Initialize SAI_I and SAI_II. For each subbasin, initialize the two indices based on the presence of alteration sources within the subbasin. If a subbasin contains one or more Level I sources, set SAI_I = 100 %; otherwise set SAI_I = 0 %. If a subbasin contains one or more Level II sources, set SAI_II = 100 %; otherwise set SAI_II = 0 %. Since the indices are independent, a subbasin containing both Level I and Level II sources is initialized as SAI_I = 100 % and SAI_II = 100 %.
Step 3: Propagate indices through the river network. For a given static set of alteration signal sources and their classifications, deterministically traverse the routing network from headwaters to the outlet. For each downstream subbasin k, compute SAI_I and SAI_II as drainage-area-weighted averages of the indices of its directly connected upstream subbasins, as given by Eqs. (1) and (2), respectively. During the network traversal, if subbasin k contains a Level I alteration signal source, SAI_Ik is set to 100 % regardless of upstream values (e.g., SAI_I is reset to 100 % at the outlet of subbasin G in Fig. 1); similarly, if subbasin k contains a Level II signal source, SAI_IIk is set to 100 %. This same-level “reset” reflects the assumption that all flow at the outlet of a source-containing subbasin has passed through at least one mapped source of the corresponding level; the index is therefore set to 100 % before further downstream mixing.1SAI_Ik=100%,SubbasinkcontainsLevelIsignalsource∑j=1nSAI_Ij×DjDkSubbasinkcontainsnoLevelIsignalsource,2SAI_IIk=100%,SubbasinkcontainsLevelIIsignalsource∑j=1nSAI_IIj×DjDkSubbasinkcontainsnoLevelIIsignalsource,where k is the index of a given subbasin within the routing network; n is the total number of subbasins immediately upstream of subbasin k (i.e., those directly draining into subbasin k), indexed by j; j denotes the jth directly connected upstream subbasin to subbasin k; Dk and Dj represent the total drainage areas of subbasins k and j, respectively, including each subbasin's own area and the areas of its upstream subbasins; SAI_Ij and SAI_IIj represent the SAI values calculated for each directly connected upstream subbasin j. The resulting SAI values SAI_Ik and SAI_IIk are expressed as percentages.
SAI values quantify upstream spatial exposure to mapped alteration sources and, under the drainage-area scaling approximation described above, provide a first-order proxy for the fraction of flow passing at least one mapped upstream alteration source at each subbasin outlet along the routing network (Fig. 1b). SAI_I ranges from 0 % to 100 % and reflects the degree of influence associated with confirmed (Level I) sources, whereas SAI_II ranges from 0 % to 100 % and reflects the degree of influence associated with potential (Level II) sources. Downstream propagation follows a drainage-area-weighted mixing assumption: SAI may decrease as unaltered incremental area or lower-SAI tributary inflows dilute an upstream signal or increase at confluences receiving higher-SAI tributary inflows. This behaviour is illustrated in Fig. 1b, where SAI decreases between confluences as contributing area increases, and step changes (i.e., resetting) occur at subbasins containing new sources. When a subbasin contains a new alteration source, the corresponding same-level index is reset to 100 % prior to further downstream propagation. Unlike indices based on the summed storage of all upstream reservoirs, this resetting rule is designed to track the spatial imprint of alteration signals along the network, so that SAI reflects downstream propagation and dilution rather than simple cumulative accumulation of upstream infrastructure. Furthermore, because SAI_I and SAI_II are propagated independently, this formulation allows an upstream Level I signal to remain traceable downstream even after the network passes a Level II source, thereby avoiding artificial truncation by a weaker downstream Level II signal.
Overall, the SAI framework provides a parsimonious, network-based characterization of potential upstream alteration influence at the subbasin scale that is readily interpretable for applications such as delineating downstream influence zones, screening hydrometric gauges for dam impacts, and supporting aquatic and ecological analyses, without requiring detailed dam or reservoir operating information. For subsequent applications, SAI_I and SAI_II values can be translated into categorical classes (e.g., natural, conditional, altered) using predefined thresholds (Sect. 2.5).
Evaluation of SAI at Hydrometric Gauges
The accuracy and reliability of SAI depend on the completeness and quality of the datasets used to identify alteration signal sources. We therefore prioritized provincial inventories over global datasets such as the GeoDAR (Wang et al., 2022) and Global Dam Watch (Lehner et al., 2024) when developing SAI v1.0 for Ontario. Although global products provide broad coverage and have undergone quality control, our province-scale data source inspection showed that global sources can miss important dams, reservoirs, and major waterpower facilities in Ontario, which would lead to spatial under-identification of alteration signals and underestimation of alteration extent.
We retained only active Water Survey of Canada (WSC) streamflow and water level gauges for evaluation that had at least three years of observations during 1990 to 2024. This screening removes discontinued gauges and helps ensure that the selected gauges reflect alteration conditions over recent decades, which are more consistent with the present-day influence of dams and reservoirs. Although water diversions occur in Ontario and are represented in the OLRRP network through enhanced hydrologically conditioned flow directions, diversions reflect basin-scale redistribution rather than the instream dam and reservoir alteration targeted by SAI. Gauges primarily influenced by diversions were therefore excluded, including those associated with regulated diversion works and watershed diversions recognized under Ontario water management frameworks (International Joint Commission, 2020; Ontario Ministry of Natural Resources and Forestry, 2017).
After the screening, 488 WSC gauges remained for evaluation. These gauges are classified in the HYDAT database as either “Regulated” or “Natural” (Water Survey of Canada, 2025). Since the RHBN gauges are a subset of the WSC gauge network, 148 RHBN gauges were included within the 488-gauge WSC set. The evaluation dataset therefore comprised three gauge groups: (1) 148 RHBN natural gauges, (2) 125 WSC-natural gauges outside the RHBN set, and (3) 215 WSC-regulated gauges.
RHBN gauges have been previously screened to represent virtually natural streamflow conditions with high confidence (Brimley et al., 1999; Ouarda et al., 1999; Pellerin, 2019). By contrast, according to the HYDAT metadata (see Data Availability), WSC-regulated gauges are defined as gauging stations with an upstream regulating structure, such as a dam or weir. This classification is useful, but it is generally less certain than RHBN, because WSC regulation metadata are not as complete or consistently documented. Accordingly, comparison with the WSC regulated/natural classification should be interpreted only as a broad consistency check rather than as a rigorous validation of SAI.
We evaluated SAI v1.0 by comparing SAI-derived conditions at hydrometric gauge locations against independent agency classifications from the WSC and the Reference Hydrometric Basin Network (RHBN). For each gauge, we extracted the corresponding SAI_I and SAI_II values and assigned one of three categories: a location is natural if SAI_I ≤ 20 % and SAI_II ≤ 40 %; it is altered if SAI_I > 20 %; and it is conditional otherwise (i.e., SAI_I ≤ 20 % but SAI_II > 40 %). This three-class scheme prioritizes SAI_I for identifying altered conditions, because Level I sources represent confirmed and typically stronger alteration signals, while SAI_II is used to flag cases where alteration is plausible but more uncertain. Thus, the conditional class indicates limited confirmed alteration but elevated potential alteration influence.
These classification thresholds were selected to provide a conservative separation between SAI-based screening classes. We adopted an SAI_I threshold of 20 % because Level I signals represent higher-confidence and typically stronger evidence of alteration, which should remain limited in basins classified as natural. This threshold is also consistent with agency instream flow assessment practice, which commonly applies pragmatic decision rules to delineate downstream impact limits for waterpower projects, including watershed-area scaling criteria such as the “5 times drainage area” rule described by Lewis et al. (2004). We adopted an SAI_II threshold of 40 % because Level II signals indicate potential rather than confirmed alteration, and thus a somewhat more permissive threshold is appropriate. Both thresholds are pragmatic screening criteria rather than calibrated or universal thresholds of hydrologic alteration; they are used only for classification in the evaluation and related SAI-based applications and do not affect the underlying continuous SAI calculations.
Data Records and Technical Evaluation
This section describes the Streamflow Alteration Index (SAI) v1.0 database, including the GIS layers provided and their key attributes (Sect. 3.1), followed by the technical evaluation of the database (Sect. 3.2).
SAI v1.0 Database
The SAI v1.0 database for Ontario is publicly available under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence at 10.5281/zenodo.19547932 (Shen et al., 2026). The release contains three primary data products:
Subbasin polygon layer (ontario_subbasin_sai_v1-0.shp): an Esri Shapefile containing SAI values for all subbasins in the Ontario Lake and River Routing Product (OLRRP) v2 network.
River polyline layer (ontario_river_v2-0-1.shp): an Esri Shapefile containing SAI values for all river reaches in the Ontario Lake and River Routing Product (OLRRP) v2 network, which is paired with the subbasin layer. Each river-reach feature carries the SAI_I and SAI_II values of its associated subbasin outlet; SAI is not recalculated continuously along individual reaches.
Alteration signal source table (ontario_alteration_signals.csv): an ASCII (tabular) dataset describing all Level I, Level II, and Level III alteration signal sources in Ontario.
The Esri Shapefiles are in the NAD 1983 Ontario MNR Lambert coordinate reference system (Lambert Conformal Conic). The datasets and their attributes are described below.
Ontario Alteration Signal Sources
This CSV table contains all 2476 retained candidate infrastructure records evaluated in the SAI workflow, including their assigned Level I, Level II, or Level III classifications. Since sources were compiled from multiple agencies and inventories, original metadata and attribute structures were not uniform. We therefore retained only the core fields required for signal identification and SAI processing and harmonized key identifiers and location fields to a consistent schema.
Attributes in the alteration signal sources table of the Streamflow Alteration Index (SAI) v1.0 database for Ontario.
AttributesDescriptionsRegionOLRRP v2 drainage region used as a delineation subdomain, including SW, NE, SE, NW, NC, FNE, FNC1, FNC2, and FNW.SRC_obsSource dataset associated with each infrastructure record. ODI: dam points from the Ontario Dam Inventory; MNR: dam points from the Ontario MNR; WPF: waterpower facility points from the Ontario MNR; LOCKSTATION: Lockstation records from the Trent-Severn Waterway, the Rideau Canal, and the Welland Canal.Obs_NMSource-record identifier reported by the originating agency; a unique identifier was generated where no valid identifier was available.Gauge_nmDescriptive source name retained from the original metadata, where available.SubIdUnique identifier of subbasins in OLRRP v2.Lat/LonNAD 1983 geographic coordinates.Level_IBinary variable to indicate if an alteration signal source is Level I.Level_IIBinary variable to indicate if an alteration signal source is Level II.Level_IIIBinary variable to indicate if an alteration signal source is Level III.
As summarized in Table 2, fields include: the OLRRP drainage region label (Region); the source dataset indicator (SRC_obs), which records the provenance of each record (e.g., provincial dam inventories, waterpower facility datasets, and lockstations); a source identifier (Obs_NM) and, where available, a descriptive name (Gauge_nm); the associated OLRRP subbasin identifier (SubId); geographic coordinates (Lat, Lon) referenced to NAD 1983; and binary indicators (Level_I, Level_II, Level_III) that flag whether the source is treated as a confirmed, potential, or negligible alteration signal for SAI mapping.
Name-related fields are not available for all source types. Records in the signal-source table can be linked to the subbasin polygon layer via SubId (recommended) to retrieve the corresponding SAI_I/SAI_II values and additional OLRRP v2 attributes for the hosting subbasin.
Ontario Subbasin-Scale SAI
The attribute fields of the subbasin polygon layer are documented in Table 3 and include: (1) the OLRRP v2 drainage region identifier (Region); (2) the unique subbasin identifier (SubId) and its immediate downstream identifier (DowSubId), which together define the directed river network topology; and (3) subbasin-scale SAI values (SAI_I and SAI_II).
Attributes shared by the subbasin and river-reach layers of the Streamflow Alteration Index (SAI) v1.0 database for Ontario.
AttributesDescriptionsRegionOLRRP v2 drainage region used as a delineation subdomain, including SW, NE, SE, NW, NC, FNE, FNC1, FNC2, and FNW.SubIdUnique identifier of subbasins in OLRRP v2.DowSubIdID of downstream subbasin.SAI_IPercentage of upstream drainage area whose routed flow passes through at least one Level I alteration signal source at the outlet of the subbasin; used as a first-order proxy for potential upstream flow exposure to Level I sources.SAI_IIPercentage of upstream drainage area whose routed flow passes through at least one Level II alteration signal source at the outlet of the subbasin; used as a first-order proxy for potential upstream flow exposure to Level II sources.
Since SubId and DowSubId follow the OLRRP v2 schema, users can link the SAI layer to the full OLRRP v2 attribute tables to retrieve additional subbasin- and reach-level descriptors (e.g., drainage area, river length, river slope, and other routing parameters) using table joins on SubId. A companion river-reach polyline layer is provided with a consistent identifier schema; however, the number of reach features is smaller than the number of subbasins, because some subbasins (e.g., headwaters) do not contain a mapped river reach.
The Streamflow Alteration Index (SAI) v1.0 database provides a province-wide, subbasin-scale representation of upstream spatial exposure to mapped instream alteration sources by propagating point-based alteration signals downstream through the connected OLRRP v2 network. The resulting spatial patterns of SAI_I (0 %–100 %) and SAI_II (0 %–100 %) are mapped across Ontario in Fig. 2, revealing pronounced heterogeneity in potential alteration influence within and among the nine drainage regions. To illustrate these propagation patterns more clearly at the regional scale, Fig. 3 presents higher-resolution maps for the Southeast (SE) region, where the spatial imprint of individual signal sources and their downstream influence can be seen more clearly along connected river networks. Figures 2 and 3 are intended to provide province-wide and regional overviews of SAI spatial patterns, respectively. For more detailed exploration, subbasin-level SAI values, alteration signal sources, OLRRP v2 hydrographic layers, and their associated attributes can be viewed through the interactive online map described in Sect. 5, “Data availability”.
Spatial distribution of subbasin-scale (a) SAI_I and (b) SAI_II across Ontario. Subbasin boundaries (gray outlines; fill colors indicate SAI magnitude), rivers (blue lines; thicker lines indicate higher Strahler stream order), and lakes (blue outlined polygons; only lakes with area > 100 km2 are shown) are from the Ontario Lake and River Routing Product Version 2 (OLRRP v2). Orange points are selected cities. Ontario is partitioned into nine independent drainage regions: Southwest (SW), Southeast (SE), Northeast (NE), North Central (NC), Northwest (NW), Far Northeast (FNE), Far North Central 1 (FNC1), Far North Central 2 (FNC2), and Far Northwest (FNW).
Higher-resolution maps of subbasin-scale (a) SAI_I and (b) SAI_II for the Southeast (SE) region of Ontario. Black points indicate Level I and Level II signal sources for SAI_I and SAI_II, respectively. Subbasin boundaries (gray outlines; fill colors indicate SAI magnitude), rivers (blue lines), and lakes (blue outlined polygons; all lakes are shown) are from the Ontario Lake and River Routing Product Version 2 (OLRRP v2).
Technical Evaluation of SAI v1.0 Database
This evaluation is intended to assess the usefulness of SAI as a screening dataset for instream alteration, rather than to verify the exact magnitude of the estimated hydrologic alteration at individual sites. For this evaluation, we first extracted SAI_I and SAI_II values for the OLRRP v2 subbasins containing hydrometric gauges, then used these values to assign each hydrometric gauge to an SAI-based class (natural, conditional, altered) based on the thresholds defined in Sect. 2.5. We compared the resulting SAI classifications against two external reference-classification sources: Reference Hydrometric Basin Network (RHBN) gauges and Water Survey of Canada (WSC) gauge classifications (Sect. 2.5). Evaluation gauges were organized into three reference groups: (1) 148 RHBN natural gauges, (2) 125 WSC-natural gauges outside the RHBN subset, and (3) 215 WSC-regulated gauges. Results are summarized in Fig. 4.
Across gauges identified as natural by the reference datasets, SAI showed strong agreement. For RHBN natural gauges (n=148), 86 % were classified as natural by SAI and 14 % as conditional, with no gauges classified as altered (Fig. 4a). The 21 RHBN gauges assigned to the conditional class indicate that there may be some inconsistency with this high-confidence natural benchmark, which highlights that the present comparison should be interpreted as a screening-level consistency check rather than an exact validation. For WSC-natural gauges outside the RHBN subset (n=125), 86 % were classified as natural, 12 % as conditional, and 2 % as altered (Fig. 4b). Agreement was lower, as expected, for WSC-regulated gauges (n=215), for which SAI classified 26 % as altered, 40 % as conditional, and 34 % as natural (Fig. 4c). At minimum, this result suggests that the WSC regulation flag is more inclusive than the narrower instream alteration definition represented by SAI and may include basins where the type or magnitude of alteration represented by the current SAI framework is limited or uncertain. Overall, the pattern of disagreement in Fig. 4 is consistent with expectations: the least disagreement occurs for the highest-confidence reference dataset (RHBN), whereas the greatest disagreement occurs for the lower-confidence WSC-regulated class.
Summary of the Streamflow Alteration Index (SAI) evaluation results for three reference gauge groups: (a) RHBN natural gauges, (b) WSC-natural (non-RHBN) gauges, and (c) WSC-regulated gauges. Donut charts show the proportion of gauges assigned to each SAI class (natural, conditional, altered) based on the classification rules in Sect. 2.5. Natural is defined as SAI_I ≤ 20 % and SAI_II ≤ 40 %; altered is defined as SAI_I > 20 %; all remaining cases are classified as conditional. The number of gauges in each reference group (n) is shown at the center of each chart, and percentages indicate the share of gauges in each SAI-based class.
Disagreements between SAI-based classes and agency labels are expected for several reasons: (1) The reference classifications are not designed to isolate instream development impacts alone: RHBN screening may reflect broader watershed pressures beyond instream infrastructure, and the basis for WSC “natural” and “regulated” labels is not consistently reported at a level that enables one-to-one mapping to SAI; (2) SAI explicitly represents uncertainty through SAI_I and SAI_II by incorporating both confirmed and potential alteration signal sources, which can yield conditional classifications where evidence is incomplete; and (3) The alteration signal source inventory may be incomplete for some locations, which can lead to underestimation of alteration where regulation exists but is not represented in available datasets. More broadly, the mismatches in Fig. 4 should be interpreted in light of uncertainty in both the reference labels and the SAI-based classes, rather than as definitive classification errors in either dataset. These considerations motivate interpreting SAI as a practical screening tool that provides consistent, spatially comprehensive classifications, while allowing users to incorporate additional local knowledge (e.g., hydrograph inspection or consultation of detailed upstream dam and reservoir information) when agency labels and SAI classes disagree.
Despite these limitations, Fig. 4 indicates that SAI v1.0 provides a practical provincial-scale screening tool. It preserves a large majority of gauges labelled natural by RHBN and WSC, while partitioning WSC-regulated gauges into altered, conditional, and natural subsets that can be prioritized for follow-up assessment.
Discussion and Limitations
The Streamflow Alteration Index (SAI) is not intended to quantify the precise magnitude of streamflow alteration at a specific location – for example, the volume of water stored in reservoirs, operating releases, or detailed changes in hydrograph magnitude and timing relative to natural conditions. Instead, SAI is designed as a network-wide screening metric that quantifies upstream spatial exposure to mapped alteration sources using the percentage of upstream drainage area whose routed flow passes through those sources.
SAI v1.0 provides a static screening baseline across the represented river network for characterizing potential streamflow alteration. This baseline can support broader-scale comparisons among subbasins and provide a reproducible benchmark from which updated or dynamic representations of SAI may be developed. Such extensions could take three complementary forms:
The alteration-source inventory can be updated as new information becomes available. Updates may include newly identified existing structures, improved local information on the operational significance of currently mapped facilities, or newly constructed, modified, or decommissioned dams, reservoirs, waterpower facilities, and other water-control structures. For regional or basin-specific applications, the province-wide inventory archived with SAI v1.0 can be supplemented with more detailed local information, the Level I and Level II classifications can be revised where appropriate, and SAI_I and SAI_II can then be recalculated. The publicly available SAI calculation code supports reproducible recalculation as the source inventory and associated metadata are improved.
The SAI framework could be extended to represent temporally varying operations. Some alteration sources may exert stronger influence only during particular seasons, hydrologic conditions, or operating periods. Seasonal or period-specific SAI values could therefore be calculated using different sets or classifications of active alteration sources for different time intervals. Where information on operating schedules, facility status, seasonal management practices, or reservoir releases is available, it could be used to define time-dependent alteration signals and generate corresponding time-varying SAI maps. Such an extension would preserve the existing network-propagation framework while allowing the represented alteration influence to vary through time.
The SAI framework could support prospective scenario analysis by introducing proposed or synthetic alteration sources into the routing network and recalculating their potential downstream influence. This capability could be used to compare alternative infrastructure locations, combinations of proposed developments, or infrastructure-removal scenarios and could support regional water-resources planning and cumulative-impact assessment. These applications would provide screening-level representations of the potential spatial extent of alteration rather than predictions of the exact magnitude or timing of streamflow change.
In practice, SAI helps delineate the potential downstream footprint of instream alteration and identify hydrometric gauges that may be influenced by upstream dams and reservoirs. This capability reduces uncertainty when screening hydrometric gauges, particularly for large monitoring networks where “natural” versus “regulated” labels are not always accompanied by transparent, consistent criteria. As demonstrated in Sect. 3.2, SAI aligns strongly with RHBN natural gauges and provides a consistent, network-wide complement to agency classifications when selecting calibration and validation data for large-domain hydrologic modelling.
A key limitation is that SAI v1.0 targets instream alteration signals and does not explicitly represent landscape-driven alterations (e.g., land-use change, drainage modifications, agriculture, or urbanization). As a result, gauges classified as natural by SAI may still be affected by non-structural watershed pressures that are outside the scope of the current signal inventory. Hydrologic modellers should therefore treat SAI as an initial filter and incorporate additional information, such as land-use datasets, water withdrawals, or hydrograph-based diagnostics, when evaluating gauge suitability for specific applications. Even so, SAI can reduce the initial workload by identifying gauges with clear evidence of upstream instream alteration.
Another limitation is that downstream alteration impact in SAI is represented using drainage area alone. This assumes that long-term streamflow is approximately proportional to contributing drainage area. Erikson et al. (2024) found approximately linear discharge–drainage area scaling for most hydrologic regions across the United States and Canada, but also identified systematic regional departures associated with differences in runoff efficiency and hydroclimatic conditions. Our Ontario-specific analysis similarly showed strong, near-linear scaling between long-term mean annual discharge and drainage area across the province and three broad regional groupings, while also indicating modest departures from exact proportionality (Appendix B). Streamflow is therefore not determined by drainage area alone: spatial variability in precipitation, evapotranspiration, snow accumulation and melt, soils, geology, groundwater contributions, lakes, wetlands, and other physiographic controls can produce departures from proportional scaling. In addition, subbasins with similar drainage area may differ in their effective alteration impact attenuation capacity because of lakes, wetlands, channel storage, and routing characteristics; subbasins containing lakes, for example, may exert stronger flow-buffering and hydrograph-smoothing effects than lake-free subbasins of comparable size (Hudson et al., 2021).
A related spatial simplification arises because SAI is evaluated at subbasin outlets: a source located within a subbasin is treated as influencing the full flow at that outlet. This approximation does not resolve within-subbasin routing and may locally overestimate exposure where a structure affects only part of the flow reaching the subbasin outlet, although the relatively fine OLRRP v2 discretization limits the spatial scale of this effect.
Although SAI_I and SAI_II distinguish sources according to the confidence and expected influence of alteration, all sources within a given level are assigned the same initial SAI value of 100 %. The current formulation therefore does not explicitly account for differences in waterbody area, volume, depth, shape, storage relative to inflow and outlet configuration. These characteristics may influence storage, residence time, mixing, routing, and the magnitude and downstream persistence of alteration (Habets et al., 2018; Hudson et al., 2021). Schmadel et al. (2018), for example, showed that the local and cumulative effects of lakes and reservoirs on nitrogen removal depend on their relative size, shape, connectivity, and position within the river network. In particular, ponded-water roundness and alignment with the river network affected mixing and nitrogen-removal behaviour. Although their study focused on nitrogen processing rather than streamflow alteration, it illustrates the broader importance of distinguishing among waterbodies in network-based assessments. Future versions of SAI could therefore replace the uniform source values with source-specific alteration weights or attenuation functions informed by waterbody geometry, storage characteristics, operating information, hydrologic connectivity, and network position. Such refinements would require more complete and spatially consistent source attributes and additional validation, whereas SAI v1.0 intentionally retains a parsimonious formulation suitable for large-domain screening.
A further methodological simplification is that Level I and Level II signals are propagated independently. As a result, a downstream Level II source does not attenuate or replace an upstream Level I signal. This behaviour was intentional, because Level I signals dominate the identification of altered sites in the current screening framework, and preserving their traceability helps avoid artificial suppression by more uncertain downstream Level II sources. Therefore, this simplification is expected to have limited effect on the final natural–conditional–altered classification, although it may not fully represent more complex interactions among sequential source types along the same river path.
Moreover, the compiled infrastructure inventory should not be considered exhaustive. Although multiple known provincial and sector-specific datasets were harmonized to provide broad spatial and facility-type coverage, their completeness cannot be independently verified for every region or facility type, and small, perhaps locally managed and privately owned structures may be missing. In addition, the HydroLAKES-derived OLRRP v2 lake layer excludes waterbodies smaller than 0.1 km2, including many small ponds and impoundments. Although individually small, these waterbodies can collectively alter annual discharge, low and peak flows, and hydrograph timing, while also exerting substantial cumulative effects on nutrient and sediment retention (Brogan et al., 2022; Habets et al., 2018; Schmadel et al., 2019). For SAI specifically, omissions of control structures are expected to be less consequential when they involve structures with primarily localized water-level effects, which would generally be treated as negligible (Level III) alteration sources and excluded from SAI initialization. However, omission of facilities that substantially regulate streamflow would lead to underestimation of alteration along their downstream river networks. We therefore recommend using SAI v1.0 as a network-wide screening reference, supplemented by targeted basin-specific review and additional local infrastructure information where available.
SAI v1.0 should not be interpreted as a direct water-quality indicator. Nevertheless, dams, reservoirs, and other ponded waterbodies can influence nutrient retention, sediment transport, temperature, dissolved oxygen, and other water-quality processes through changes in storage, residence time, mixing, and biological activity (Maavara et al., 2020; Schmadel et al., 2018). SAI's network-based formulation could therefore provide a screening measure of the proportion of upstream flow potentially exposed to processing within such waterbodies. In this sense, SAI may serve as an explanatory attribute in water-quality assessments or as the structural basis for a complementary water-quality influence index.
The evaluation against independent reference classifications and the demonstrated utility of SAI v1.0 in Ontario suggest strong potential for extending the methodology to broader spatial domains. Ongoing advances in large-sample dam and reservoir inventories, together with the increasing availability of high-resolution hydrographic routing frameworks at national, continental, and global scales, make it feasible to implement the SAI framework beyond Ontario and to develop network-based screening products that support more consistent assessment of instream alteration for hydrologic and ecological applications at large scales.
Data availability
The Streamflow Alteration Index (SAI) v1.0 database developed in this study, including the subbasin polygon layer, river polyline layer, alteration signal source table, and accompanying documentation, is archived on Zenodo under the Creative Commons Attribution 4.0 International (CC BY 4.0) licence: 10.5281/zenodo.19547932 (Shen et al., 2026).
A province-wide interactive map of SAI v1.0 is available at: https://www.arcgis.com/apps/mapviewer/index.html?webmap=71a9a1859b13409bb85d9d1a8fc31bb3 (last access: 25 October 2026).
The map includes OLRRP v2 subbasin, river, and lake layers; SAI_I and SAI_II; and Level I, Level II, and Level III alteration signal sources. Users can interactively explore the map at different spatial scales and inspect the attributes associated with each layer. The corresponding GIS shapefiles and alteration signal source table are also available from Zenodo (10.5281/zenodo.19547932, Shen et al., 2026) for download and local analysis.
Most input datasets used in this study are publicly available from their original sources. The supplementary dataset of Ontario MNR-operated dams is not available for public download, as described below. The input data sources and access information are as follows:
Ontario Lake and River Routing Product Version 2 (OLRRP v2): available through Shen et al. (2024). OLRRP website at: https://uwaterloo-olrrp.shinyapps.io/OLRRP-V2/ (last access: 25 October 2026).
Ontario Dam Inventory (ODI): available from the Ontario Ministry of Natural Resources (MNR) through Ontario GeoHub: https://geohub.lio.gov.on.ca/maps/mnrf::ontario-dam-inventory/about (Ontario Ministry of Natural Resources, 2014).
Waterpower Generation Stations: available from the Ontario Ministry of Natural Resources through Ontario GeoHub: https://geohub.lio.gov.on.ca/documents/waterpower-generation-stations/about (last access: 25 October 2026).
Water Management Plan (WMP) classifications: the simple and complex Water Management Plan classifications used in this study were parsed and compiled from the report Maintaining Water Management Plans: Technical Bulletin published by the Ministry of Natural Resources and Forestry (2016): https://files.ontario.ca/technical-bulletin-maintaining-wmps.pdf.
Hydrometric gauge data: Water Survey of Canada hydrometric data are available from the HYDAT database: https://collaboration.cmc.ec.gc.ca/cmc/hydrometrics/www/ (Water Survey of Canada, 2025).
HYDAT metadata documentation is available at: https://collaboration.cmc.ec.gc.ca/cmc/hydrometrics/www/ECDataExplorer_EN.pdf (last access: 25 October 2026).
Reference Hydrometric Basin Network: RHBN documentation and data access are available from Environment and Climate Change Canada: https://www.canada.ca/en/environment-climate-change/services/water-overview/quantity/monitoring/survey/data-products-services/reference-hydrometric-basin-network.html (last access: 25 October 2026).
Lockstation locations and information: official sources were used for the Trent-Severn Waterway, Rideau Canal, and Welland Canal: https://parks.canada.ca/lhn-nhs/on/trentsevern/visit/posteeclusage-lockstation (last access: 25 October 2026); https://parks.canada.ca/lhn-nhs/on/rideau/visit/posteeclusage-lockstation (last access: 25 October 2026); https://greatlakes-seaway.com/en/the-seaway/our-locks-and-channels/ (last access: 25 October 2026).
Ontario MNR-operated dams: the supplementary dataset of 393 MNR-operated dams was provided by the Ontario Ministry of Natural Resources and is not available for public download. The harmonized locations, identifiers, and Level I–III classification fields derived from this dataset and used in the SAI workflow are included in the archived alteration signal source table.
Code availability
The scripts used to calculate and propagate SAI values through the OLRRP v2 routing network are available at https://github.com/hongren-shen/streamflow-alteration-index (last access: 25 October 2026) and archived as version v1.0.0 at 10.5281/zenodo.23602080 (Shen, 2026). These scripts reproduce the core SAI calculation workflow used to generate the SAI v1.0 database.
Conclusions
We developed the Streamflow Alteration Index (SAI) v1.0 database for Ontario to provide a spatially complete, subbasin-scale screening product that maps how exposure to upstream instream alteration signal sources propagates downstream across a connected hydrographic network. The database includes paired indices (SAI_I and SAI_II) that represent two different levels of alteration influence based on confirmed and potential sources, enabling consistent classification of subbasin outlets and hydrometric gauge locations as natural, conditional, or altered. Our SAI v1.0 database and province-wide interactive map enable water resources researchers and practitioners to quickly assess upstream spatial exposure to mapped alteration sources at subbasin outlets and corresponding river reaches and, under the drainage-area scaling approximation, use this information as a first-order proxy for the proportion of streamflow potentially exposed to those sources, while accounting for differences in source confidence. Evaluation against independent reference classifications indicates that SAI classified 86 % of RHBN natural gauges and 86 % of WSC-natural gauges as natural. Among WSC-regulated gauges, 26 % were classified as altered, 40 % as conditional, and 34 % as natural.
Overall, SAI v1.0 offers a practical, network-based tool for large-domain hydrologic and ecological applications, including gauge screening, model calibration/validation support, and basin-scale prioritization of locations potentially affected by upstream alteration. Future work will focus on improving alteration source inventories, extending the framework to larger scales (e.g., national applications across Canada), and integrating complementary information to better represent non-structural watershed alterations.
Spatial distribution of infrastructure records and alteration signal sources
Figures A1 and A2 provide province-wide and detailed views of the spatial distributions of the 2476 infrastructure records retained after geospatial alignment and screening (Sect. 2.2) and their resulting alteration signal classifications (Sect. 2.3), respectively. Figure A1 shows the compiled records from the four infrastructure datasets, whereas Fig. A2 shows the resulting Level I, Level II, and Level III alteration signal sources. Following subbasin-level consolidation, Level I and Level II sources occurred in 643 unique subbasins and were retained for SAI initialization, whereas 1288 records were classified as Level III and treated as negligible.
Spatial distribution of the 2476 infrastructure records retained for identifying potential streamflow alteration sources across Ontario. (a) Province-wide distribution of records from the Ontario Dam Inventory (ODI), Ontario Ministry of Natural Resources (MNR), Waterpower Facilities (WPFs), and lockstation datasets. (b) Zoomed-in view of infrastructure-dense areas. The dashed box in (a) indicates the spatial extent shown in (b). Rivers (blue lines; thicker lines indicate higher Strahler stream order) and lakes (blue polygons; only lakes with areas > 100 km2 are shown) are from the Ontario Lake and River Routing Product Version 2 (OLRRP v2).
Spatial distribution of the classified alteration signal sources across Ontario. (a) Province-wide distribution of Level I (confirmed), Level II (potential), and Level III (negligible) alteration signal sources. Following subbasin-level consolidation, Level I and/or Level II sources occurred in 643 unique subbasins, whereas 1288 candidate records were classified as Level III. Level I and Level II sources are retained for Streamflow Alteration Index (SAI) initialization, whereas Level III sources are treated as negligible and excluded from SAI initialization and propagation. (b) Zoomed-in view of alteration signal-dense areas. The dashed box in (a) indicates the spatial extent shown in (b). Rivers (blue lines; thicker lines indicate higher Strahler stream order) and lakes (blue polygons; only lakes with areas > 100 km2 are shown) are from the Ontario Lake and River Routing Product Version 2 (OLRRP v2).
Empirical assessment of discharge–drainage area scaling in Ontario
To provide an Ontario-specific empirical assessment of the drainage-area scaling approximation used in SAI, we examined the relationship between long-term mean annual discharge and drainage area using Water Survey of Canada (WSC) discharge observations from the HYDAT database (Water Survey of Canada, 2025). The analysis covered complete calendar years from 1990 through 2024. We retained 327 discharge gauges with at least 10 valid years of observations, where a valid year was required to have at least 95 % daily discharge coverage. Gauges were included irrespective of their WSC natural or regulated classification; the resulting sample therefore characterizes observed long-term discharge scaling across the broader Ontario hydrometric network rather than natural runoff scaling alone. For each gauge, long-term mean annual discharge was calculated from the available valid years.
Following the general power-law framework used by Erikson et al. (2024) to investigate drainage area and discharge relationship linearity, the discharge–drainage area relationship was represented as
Q=cAβ
where Q is long-term mean annual discharge, A is OLRRP-derived delineated drainage area, c is a fitted coefficient, and β is the scaling exponent. Ordinary least-squares regression was performed in log–log space for Ontario as a whole and separately for three broad regional groupings representing major geographic and hydroclimatic differences across the province: South (SW and SE), North (NE, NC, and NW), and Far North (FNE, FNC1, FNC2, and FNW). Under this formulation, β=1 corresponds to exact proportional scaling between long-term mean discharge and drainage area. The 95 % confidence interval for β was calculated from the standard error of the fitted slope in log–log space using the Student's t distribution with n-2 degrees of freedom.
Strong discharge–drainage area relationships were observed for Ontario and all three regional groupings (Fig. B1). The fitted scaling exponent β was 0.967 for Ontario, 1.013 for South, 0.936 for North, and 0.967 for Far North, with coefficient of determination (R2) values ranging from 0.967 to 0.991. These results indicate strong, near-linear scaling between observed long-term mean annual discharge and drainage area across the Ontario hydrometric network, while also showing modest regional departures from exact proportionality. Because the analysis includes both natural and regulated gauges, it should not be interpreted as an assessment of natural runoff scaling alone. Rather, together with the approximately linear relationships reported by Erikson et al. (2024) for minimally disturbed catchments, the Ontario analysis provides empirical support for using proportional drainage-area weighting as a parsimonious first-order approximation in SAI. The results do not imply that drainage area alone determines streamflow contributions at individual locations, as spatial variability in hydroclimate, physiography, water storage, and human regulation can produce departures from proportional scaling.
Log-log relationships between drainage area and long-term mean annual discharge for Ontario and three broad regional groupings: South (SW and SE), North (NE, NC, and NW), and Far North (FNE, FNC1, FNC2, and FNW). Each point represents a discharge gauge with at least 10 valid years of observations during 1990–2024, and n denotes the number of gauges included in each panel. The solid black line shows the fitted power-law relationship, Q=cAβ, where Q is long-term mean annual discharge, A is drainage area, and β is the scaling exponent; β=1 corresponds to exact proportional scaling between discharge and drainage area. The fitted equation and coefficient of determination (R2) are shown in each panel. The 95 % confidence interval (CI) for β was calculated from the standard error of the fitted slope in log–log space using the Student's t distribution with n-2 degrees of freedom.
Author contributions
Hongren Shen led the study, including data compilation and processing, methodological development, SAI calculation, analysis, figure preparation and visualization, and manuscript writing and revision. Bryan A. Tolson and James R. Craig contributed to methodological design, interpretation of results, figure visualization, and manuscript revision. Robert A. Metcalfe and James J. Luce contributed data resources, interpretation of results, and manuscript revision. Jonathan Romero-Cuellar contributed to interpretation of results and manuscript revision. All authors reviewed and approved the final manuscript.
Competing interests
The contact author has declared that none of the authors has any competing interests.
Disclaimer
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. The authors bear the ultimate responsibility for providing appropriate place names. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.
Acknowledgements
The authors thank the Ontario Ministry of Natural Resources, the Water Survey of Canada, Parks Canada, and other organizations that maintain and provide access to the datasets used in this study. We also thank the three anonymous referees for their constructive comments and suggestions, which helped improve the manuscript.
Financial support
This research was supported by a 2022–2026 Collaborative Research Agreement between the Ontario Ministry of Natural Resources and the University of Waterloo, led by Principal Investigator Bryan A. Tolson. The agreement supported Hongren Shen's research appointment.
Review statement
This paper was edited by Yue Qin and reviewed by three anonymous referees.
ReferencesArsenault, R., Brissette, F., Martel, J.-L., Troin, M., Lévesque, G., Davidson-Chaput, J., Gonzalez, M. C., Ameli, A., and Poulin, A.: A comprehensive, multisource database for hydrometeorological modeling of 14,425 North American watersheds, Sci. Data, 7, 243, 10.1038/s41597-020-00583-2, 2020.Bejarano, M. D., Jansson, R., and Nilsson, C.: The effects of hydropeaking on riverine plants: a review, Biol. Rev., 93, 658–673, 10.1111/brv.12362, 2018.Bipa, N. J., Stradiotti, G., Righetti, M., and Pisaturo, G. R.: Impacts of hydropeaking: A systematic review, Sci. Total Environ., 912, 169251, 10.1016/j.scitotenv.2023.169251, 2024.Blöschl, G., Sivapalan, M., Wagener, T., Viglione, A., and Savenije, H. H. G. (Eds.): Runoff Prediction in Ungauged Basins: Synthesis Across Processes, Places and Scales, Cambridge University Press, 10.1017/CBO9781139235761, 2013.Brimley, B., Cantin, J. F., Harvey, D., Kowalchuk, M., Marsh, P., Ouarda, T. B. M. J., Phinney, B., Pilon, P., Renouf, M., Tassone, B., Wedel, R., and Yuzyk, T.: Establishment of the reference hydrometric basin network (RHBN) for Canada, Environment Canada, Ottawa, Catalogue No. En4-701/1999E-PDF, 41 pp., https://publications.gc.ca/pub?id=9.936545&sl=0 (last access: 25 October 2026), 1999.Brogan, C., Burgholzer, R., Keys, T., Kleiner, J., Shortridge, J., and Scott, D.: The Cumulative Role of Impoundments in Streamflow Alteration, JAWRA J. Am. Water Resour. As., 58, 119–133, 10.1111/1752-1688.12979, 2022.Burke, M., Jorde, K., and Buffington, J. M.: Application of a hierarchical framework for assessing environmental impacts of dam operation: Changes in streamflow, bed mobility and recruitment of riparian trees in a western North American river, J. Environ. Manage., 90, S224–S236, 10.1016/j.jenvman.2008.07.022, 2009.Cipollini, S., Fiori, A., and Volpi, E.: A New Physically Based Index to Quantify the Impact of Multiple Reservoirs on Flood Frequency at the Catchment Scale Based on the Concept of Equivalent Reservoir, Water Resour. Res., 58, e2021WR031470, 10.1029/2021WR031470, 2022.Cipollini, S., Volpi, E., and Fiori, A.: Toward a Catchment-Scale Assessment of Flood Peak Attenuation by Multiple Reservoirs, Water Resour. Res., 59, e2022WR032821, 10.1029/2022WR032821, 2023.Craig, J. R., Brown, G., Chlumsky, R., Jenkinson, R. W., Jost, G., Lee, K., Mai, J., Serrer, M., Sgro, N., Shafii, M., Snowdon, A. P., and Tolson, B. A.: Flexible watershed simulation with the Raven hydrological modelling framework, Environ. Model. Softw., 129, 104728, 10.1016/j.envsoft.2020.104728, 2020.Dang, T. D., Chowdhury, A. F. M. K., and Galelli, S.: On the representation of water reservoir storage and operations in large-scale hydrological models: implications on model parameterization and climate change impact assessments, Hydrol. Earth Syst. Sci., 24, 397–416, 10.5194/hess-24-397-2020, 2020.de Rham, L., Dibike, Y., Beltaos, S., Peters, D., Bonsal, B., and Prowse, T.: A Canadian River Ice Database from the National Hydrometric Program Archives, Earth Syst. Sci. Data, 12, 1835–1860, 10.5194/essd-12-1835-2020, 2020.Déry, S. J., Stadnyk, T. A., MacDonald, M. K., and Gauli-Sharma, B.: Recent trends and variability in river discharge across northern Canada, Hydrol. Earth Syst. Sci., 20, 4801–4818, 10.5194/hess-20-4801-2016, 2016.Döll, P., Fiedler, K., and Zhang, J.: Global-scale analysis of river flow alterations due to water withdrawals and reservoirs, Hydrol. Earth Syst. Sci., 13, 2413–2432, 10.5194/hess-13-2413-2009, 2009.Du, T. L. T., Lee, H., Bui, D. D., Graham, L. P., Darby, S. D., Pechlivanidis, I. G., Leyland, J., Biswas, N. K., Choi, G., Batelaan, O., Bui, T. T. P., Do, S. K., Tran, T. V., Nguyen, H. T., and Hwang, E.: Streamflow Prediction in Highly Regulated, Transboundary Watersheds Using Multi-Basin Modeling and Remote Sensing Imagery, Water Resour. Res., 58, e2021WR031191, 10.1029/2021WR031191, 2022.Erikson, C. M., Renshaw, C. E., and Magilligan, F. J.: Spatial variation in drainage area – Runoff relationships and implications for bankfull geometry scaling, Geomorphology, 446, 108998, 10.1016/j.geomorph.2023.108998, 2024.Esri: World Topographic Map/World Imagery Basemap, Esri, https://www.arcgis.com/home/item.html?id=30e5fe3149c34df1ba922e6f5bbf808f and https://www.arcgis.com/home/item.html?id=10df2279f9684e4a9f6a7f08febac2a9 (last access: 25 October 2026), 2024.Fry, L. M., Hunter, T. S., Phanikumar, M. S., Fortin, V., and Gronewold, A. D.: Identifying streamgage networks for maximizing the effectiveness of regional water balance modeling, Water Resour. Res., 49, 2689–2700, 10.1002/wrcr.20233, 2013.Google: Google Maps base layer, Google, https://www.google.com/maps (last access: 25 October 2026), 2024.Grill, G., Ouellet Dallaire, C., Fluet Chouinard, E., Sindorf, N., and Lehner, B.: Development of new indicators to evaluate river fragmentation and flow regulation at large scales: A case study for the Mekong River Basin, Ecol. Indic., 45, 148–159, 10.1016/j.ecolind.2014.03.026, 2014.Grill, G., Lehner, B., Thieme, M., Geenen, B., Tickner, D., Antonelli, F., Babu, S., Borrelli, P., Cheng, L., Crochetiere, H., Ehalt Macedo, H., Filgueiras, R., Goichot, M., Higgins, J., Hogan, Z., Lip, B., McClain, M. E., Meng, J., Mulligan, M., Nilsson, C., Olden, J. D., Opperman, J. J., Petry, P., Reidy Liermann, C., Sáenz, L., Salinas-Rodríguez, S., Schelle, P., Schmitt, R. J. P., Snider, J., Tan, F., Tockner, K., Valdujo, P. H., van Soesbergen, A., and Zarfl, C.: Mapping the world's free-flowing rivers, Nature, 569, 215–221, 10.1038/s41586-019-1111-9, 2019.Habets, F., Molénat, J., Carluer, N., Douez, O., and Leenhardt, D.: The cumulative impacts of small reservoirs on hydrology: A review, Sci. Total Environ., 643, 850–867, 10.1016/j.scitotenv.2018.06.188, 2018.Han, M., Shen, H., Tolson, B. A., Craig, J. R., Mai, J., Lin, S. G. M., Basu, N. B., and Awol, F. S.: BasinMaker 3.0: A GIS toolbox for distributed watershed delineation of complex lake-river routing networks, Environ. Model. Softw., 164, 105688, 10.1016/j.envsoft.2023.105688, 2023.Hrachowitz, M., Savenije, H. H. G., Blöschl, G., McDonnell, J. J., Sivapalan, M., Pomeroy, J. W., Arheimer, B., Blume, T., Clark, M. P., Ehret, U., Fenicia, F., Freer, J. E., Gelfan, A., Gupta, H. V., Hughes, D. A., Hut, R. W., Montanari, A., Pande, S., Tetzlaff, D., Troch, P. A., Uhlenbrook, S., Wagener, T., Winsemius, H. C., Woods, R. A., Zehe, E., and Cudennec, C.: A decade of Predictions in Ungauged Basins (PUB) – a review, Hydrolog. Sci. J., 58, 1198–1255, 10.1080/02626667.2013.803183, 2013.Hudson, D. T., Leach, J. A., Webster, K., and Houle, D.: Streamflow regime of a lake-stream system based on long-term data from a high-density hydrometric network, Hydrol. Process., 35, e14396, 10.1002/hyp.14396, 2021.Hughes, D., Birkinshaw, S., and Parkin, G.: A method to include reservoir operations in catchment hydrological models using SHETRAN, Environ. Model. Softw., 138, 104980, 10.1016/j.envsoft.2021.104980, 2021.International Joint Commission: An Overview of Great Lakes Diversions, International Joint Commission, https://ijc.org/en/lsbc/watershed/great-lakes-diversions (last access: 25 October 2026), 2020.Jorde, K., Burke, M., Scheidt, N., Welcker, C., King, S., and Borden, C.: 23 Reservoir operations, physical processes, and ecosystem losses, in: Gravel-Bed Rivers VI: From Process Understanding to River Restoration, Developments in Earth Surface Processes, Vol. 11, edited by: Habersack, H., Piégay, H., and Rinaldi, M., Elsevier, 607–636, 10.1016/S0928-2025(07)11151-2, 2007.Kern, J. D., Characklis, G. W., Doyle, M. W., Blumsack, S., and Whisnant, R. B.: Influence of deregulated electricity markets on hydropower generation and downstream flow regime, J. Water Res. Pl., 138, 342–355, 10.1061/(ASCE)WR.1943-5452.0000183, 2012.Kratzert, F., Klotz, D., Herrnegger, M., Sampson, A. K., Hochreiter, S., and Nearing, G. S.: Toward Improved Predictions in Ungauged Basins: Exploiting the Power of Machine Learning, Water Resour. Res., 55, 11344–11354, 10.1029/2019WR026065, 2019a.Kratzert, F., Klotz, D., Shalev, G., Klambauer, G., Hochreiter, S., and Nearing, G.: Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets, Hydrol. Earth Syst. Sci., 23, 5089–5110, 10.5194/hess-23-5089-2019, 2019b.Kumar, H., Hwang, J., Devineni, N., and Sankarasubramanian, A.: Dynamic Flow Alteration Index for Complex River Networks With Cascading Reservoir Systems, Water Resour. Res., 58, e2021WR030491, 10.1029/2021WR030491, 2022.Lehner, B., Liermann, C. R., Revenga, C., Vörösmarty, C., Fekete, B., Crouzet, P., Döll, P., Endejan, M., Frenken, K., Magome, J., Nilsson, C., Robertson, J. C., Rödel, R., Sindorf, N., and Wisser, D.: High-resolution mapping of the world's reservoirs and dams for sustainable river-flow management, Front. Ecol. Environ., 9, 494–502, 10.1890/100125, 2011.Lehner, B., Beames, P., Mulligan, M., Zarfl, C., De Felice, L., van Soesbergen, A., Thieme, M., Garcia de Leaniz, C., Anand, M., Belletti, B., Brauman, K. A., Januchowski-Hartley, S. R., Lyon, K., Mandle, L., Mazany-Wright, N., Messager, M. L., Pavelsky, T., Pekel, J.-F., Wang, J., Wen, Q., Wishart, M., Xing, T., Yang, X., and Higgins, J.: The Global Dam Watch database of river barrier and reservoir information for large-scale applications, Sci. Data, 11, 1069, 10.1038/s41597-024-03752-9, 2024.Lewis, A., Hatfield, T., Chilibeck, B., and Roberts, C.: Assessment Methods for Aquatic Habitat and Instream Flow Characteristics in Support of Applications to Dam, Divert, or Extract Water from Streams in British Columbia, Final Version, British Columbia Ministry of Water, Land and Air Protection and Ministry of Sustainable Resource Management, https://www.env.gov.bc.ca/wld/documents/bmp/assessment_methods_instreamflow_in_bc.pdf (last access: 25 October 2026), 2004.Lueders, M. B., McManamay, R. A., Oliver, J. A., Young, D., Bean, P., and Mayes, K.: Reconstructing Long-Term Historical Hydrologic Alteration to Support Environmental Flows in Texas, River Res. Appl., 1–17, 10.1002/rra.70172, 2026.Maavara, T., Chen, Q., Van Meter, K., Brown, L. E., Zhang, J., Ni, J., and Zarfl, C.: River dam impacts on biogeochemical cycling, Nat. Rev. Earth Environ., 1, 103–116, 10.1038/s43017-019-0019-0, 2020.Messager, M. L., Lehner, B., Grill, G., Nedeva, I., and Schmitt, O.: Estimating the volume and age of water stored in global lakes using a geo-statistical approach, Nat. Commun., 7, 13603, 10.1038/ncomms13603, 2016. Metcalfe, R. A., Mackereth, R. W., Grantham, B., Jones, N., Pyrce, R. S., Haxton, T., Luce, J. J., and Stainton, R.: Aquatic Ecosystem Assessments for Rivers, Aquatic Research Series 2013-06, Science and Research Branch, Ministry of Natural Resources, Peterborough, Ontario, ISBN 978-1-4606-3215-4 (PDF), 2013.Ministry of Natural Resources and Forestry: Maintaining Water Management Plans: Technical Bulletin, Ministry of Natural Resources and Forestry, https://files.ontario.ca/technical-bulletin-maintaining-wmps.pdf (last access: 25 October 2026), 2016.Montanari, A., Young, G., Savenije, H. H. G., Hughes, D., Wagener, T., Ren, L. L., Koutsoyiannis, D., Cudennec, C., Toth, E., Grimaldi, S., Blöschl, G., Sivapalan, M., Beven, K., Gupta, H., Hipsey, M., Schaefli, B., Arheimer, B., Boegh, E., Schymanski, S. J., Di Baldassarre, G., Yu, B., Hubert, P., Huang, Y., Schumann, A., Post, D. A., Srinivasan, V., Harman, C., Thompson, S., Rogger, M., Viglione, A., McMillan, H., Characklis, G., Pang, Z., and Belyaev, V.: “Panta Rhei – Everything Flows”: Change in hydrology and society – The IAHS Scientific Decade 2013–2022, Hydrolog. Sci. J., 58, 1256–1275, 10.1080/02626667.2013.809088, 2013.Nilsson, C., Reidy, C. A., Dynesius, M., and Revenga, C.: Fragmentation and flow regulation of the world's large river systems, Science, 308, 405–408, 10.1126/science.1107887, 2005.Ontario Ministry of Natural Resources: Ontario Dam Inventory, Ontario Ministry of Natural Resources [data set], https://geohub.lio.gov.on.ca/maps/mnrf::ontario-dam-inventory/about (last access: 25 October 2026), 2014.Ontario Ministry of Natural Resources and Forestry: Lakes and Rivers Improvement Act Administrative Guide, Ontario Ministry of Natural Resources and Forestry, https://www.ontario.ca/page/lakes-and-rivers-improvement-act-administrative-guide (last access: 25 October 2026), 2017.Ouarda, T., Rasmussen, P., Cantin, J., Bobée, B., Laurence, R., and Hoang, V.: Identification d'un réseau hydrométrique pour le suivi des modifications climatiques dans la province de Québec, Rev. Sci. Eau J. Water Sci., 12, 425–448, 10.7202/705359ar, 1999.Palmer, M. and Ruhi, A.: Linkages between flow regime, biota, and ecosystem processes: Implications for river restoration, Science, 365, eaaw2087, 10.1126/science.aaw2087, 2019.Pellerin, J.: Updating the Canadian Reference Hydrometric Basin Network to detect climate-related trends in streamflow, Master's Thesis, University of Waterloo, https://hdl.handle.net/10012/14574 (last access: 25 October 2026), 2019.Petts, G. E.: The Adjustment of River channel Capacity Downstream from Reservoirs in Great Britain, PhD thesis, University of Southampton, https://eprints.soton.ac.uk/458889/ (last access: 25 October 2026), 1978.Petts, G. E.: Complex response of river channel morphology subsequent to reservoir construction, Prog. Phys. Geogr., 3, 329–362, 10.1177/030913337900300302, 1979.Petts, G. E.: Long-term consequences of upstream impoundment, Environ. Conserv., 7, 325–332, 10.1017/S0376892900008183, 1980. Petts, G. E.: Impounded rivers: perspectives for ecological management, Wiley, ISBN 9780471103066, 1984.Poff, N. L., Allan, J. D., Bain, M. B., Karr, J. R., Prestegaard, K. L., Richter, B. D., Sparks, R. E., and Stromberg, J. C.: The natural flow regime, BioScience, 47, 769–784, 10.2307/1313099, 1997.Poff, N. L., Richter, B. D., Arthington, A. H., Bunn, S. E., Naiman, R. J., Kendy, E., Acreman, M., Apse, C., Bledsoe, B. P., Freeman, M. C., Henriksen, J., Jacobson, R. B., Kennen, J. G., Merritt, D. M., O'Keeffe, J. H., Olden, J. D., Rogers, K., Tharme, R. E., and Warner, A.: The ecological limits of hydrologic alteration (ELOHA): a new framework for developing regional environmental flow standards, Freshwater Biol., 55, 147–170, 10.1111/j.1365-2427.2009.02204.x, 2010. Richter, B. D., Postel, S., Revenga, C., Scudder, T., Lehner, B., Churchill, A., and Chow, M.: Lost in development's shadow: The downstream human consequences of dams, Water Altern., 3, 14–42, 2010.Salwey, S., Coxon, G., Pianosi, F., Lane, R., Hutton, C., Bliss Singer, M., McMillan, H., and Freer, J.: Developing water supply reservoir operating rules for large-scale hydrological modelling, Hydrol. Earth Syst. Sci., 28, 4203–4218, 10.5194/hess-28-4203-2024, 2024.Schmadel, N. M., Harvey, J. W., Alexander, R. B., Schwarz, G. E., Moore, R. B., Eng, K., Gomez-Velez, J. D., Boyer, E. W., and Scott, D.: Thresholds of lake and reservoir connectivity in river networks control nitrogen removal, Nat. Commun., 9, 2779, 10.1038/s41467-018-05156-x, 2018.Schmadel, N. M., Harvey, J. W., Schwarz, G. E., Alexander, R. B., Gomez-Velez, J. D., Scott, D., and Ator, S. W.: Small Ponds in Headwater Catchments Are a Dominant Influence on Regional Nutrient and Sediment Budgets, Geophys. Res. Lett., 46, 9669–9677, 10.1029/2019GL083937, 2019.SEI International: Strategic Environmental Assessment of the Hydropower Master Plan in the Context of the Power Development Plan VI: Final Report, Stockholm Environment Institute, https://mediamanager.sei.org/documents/Publications/SEI-SEA-Hydropower-Vietname-FullReport-2009.pdf (last access: 25 October 2026), 2009.Shen, H.: Streamflow Alteration Index (SAI) calculation tools (v1.0.0), Zenodo [code], 10.5281/zenodo.23602080, 2026.Shen, H., Han, M., Tolson, B., and Metcalfe, R. A.: The Ontario Lake and River Routing Product Version 2 (OLRRP V2), Zenodo [data set], 10.5281/zenodo.14225103, 2024.Shen, H., Tolson, B., Craig, J., Metcalfe, R. A., Romero-Cuellar, J., and Luce, J.: Streamflow Alteration Index (SAI) v1.0 Database for Ontario, Canada, Zenodo [data set], 10.5281/zenodo.19547932, 2026.Shrestha, P. K., Samaniego, L., Rakovec, O., Kumar, R., Mi, C., Rinke, K., and Thober, S.: Toward Improved Simulations of Disruptive Reservoirs in Global Hydrological Modeling, Water Resour. Res., 60, e2023WR035433, 10.1029/2023WR035433, 2024.Smokorowski, K. E.: The ups and downs of hydropeaking: a Canadian perspective on the need for, and ecological costs of, peaking hydropower production, Hydrobiologia, 849, 421–441, 10.1007/s10750-020-04480-y, 2022. Stainton, R. T.: Hydrologic Zone of Influence Downstream of Waterpower Facilities: Literature Review and Analysis of Potential Methods to Support Prediction in Ontario Rivers, IWS Report No. 02-2012, Institute for Watershed Science, Trent University, Peterborough, Ontario, Canada, 2012.Steyaert, J. C., Sutanudjaja, E. H., Bierkens, M., and Wanders, N.: Data derived reservoir operations simulated in a global hydrologic model, Hydrol. Earth Syst. Sci., 29, 6499–6527, 10.5194/hess-29-6499-2025, 2025.Tharme, R. E.: A global perspective on environmental flow assessment: emerging trends in the development and application of environmental flow methodologies for rivers, River Res. Appl., 19, 397–441, 10.1002/rra.736, 2003.Tursun, A., Xie, X., Wang, Y., Peng, D., Liu, Y., Zheng, B., Wu, X., and Nie, C.: Streamflow Prediction in Human-Regulated Catchments Using Multiscale Deep Learning Modeling With Anthropogenic Similarities, Water Resour. Res., 60, e2023WR036853, 10.1029/2023WR036853, 2024.Wada, Y., Bierkens, M. F. P., de Roo, A., Dirmeyer, P. A., Famiglietti, J. S., Hanasaki, N., Konar, M., Liu, J., Müller Schmied, H., Oki, T., Pokhrel, Y., Sivapalan, M., Troy, T. J., van Dijk, A. I. J. M., van Emmerik, T., Van Huijgevoort, M. H. J., Van Lanen, H. A. J., Vörösmarty, C. J., Wanders, N., and Wheater, H.: Human–water interface in hydrological modelling: current status and future directions, Hydrol. Earth Syst. Sci., 21, 4169–4193, 10.5194/hess-21-4169-2017, 2017.Wang, J., Walter, B. A., Yao, F., Song, C., Ding, M., Maroof, A. S., Zhu, J., Fan, C., McAlister, J. M., Sikder, S., Sheng, Y., Allen, G. H., Crétaux, J.-F., and Wada, Y.: GeoDAR: georeferenced global dams and reservoirs dataset for bridging attributes and geolocations, Earth Syst. Sci. Data, 14, 1869–1899, 10.5194/essd-14-1869-2022, 2022.Water Survey of Canada: National Water Data Archive (HYDAT), 14 October 2025 release, Water Survey of Canada [data set], https://www.canada.ca/en/environment-climate-change/services/water-overview/quantity/monitoring/survey/data-products-services/national-archive-hydat.html (last access: 25 October 2026), 2025. WCD: Dams and Development: A New Framework for Decision-making: the Report of the World Commission on Dams, Earthscan, ISBN 9781853837982, 2000.Wieczorek, M. E., Wolock, D. M., and McCarthy, P. M.: Dam impact/disturbance metrics for the conterminous United States, 1800 to 2018, U.S. Geological Survey [data set], 10.5066/P92S9ZX6, 2021.Winemiller, K. O., McIntyre, P. B., Castello, L., Fluet-Chouinard, E., Giarrizzo, T., Nam, S., Baird, I. G., Darwall, W., Lujan, N. K., Harrison, I., Stiassny, M. L. J., Silvano, R. A. M., Fitzgerald, D. B., Pelicice, F. M., Agostinho, A. A., Gomes, L. C., Albert, J. S., Baran, E., Petrere, M., Zarfl, C., Mulligan, M., Sullivan, J. P., Arantes, C. C., Sousa, L. M., Koning, A. A., Hoeinghaus, D. J., Sabaj, M., Lundberg, J. G., Armbruster, J., Thieme, M. L., Petry, P., Zuanon, J., Vilara, G. T., Snoeks, J., Ou, C., Rainboth, W., Pavanelli, C. S., Akama, A., Soesbergen, A. V., and Sáenz, L.: Balancing hydropower and biodiversity in the Amazon, Congo, and Mekong, Science, 351, 128–129, 10.1126/science.aac7082, 2016.Yang, X., Pavelsky, T. M., Ross, M. R. V., Januchowski-Hartley, S. R., Dolan, W., Altenau, E. H., Belanger, M., Byron, D., Durand, M., Van Dusen, I., Galit, H., Jorissen, M., Langhorst, T., Lawton, E., Lynch, R., Mcquillan, K. A., Pawar, S., and Whittemore, A.: Mapping Flow-Obstructing Structures on Global Rivers, Water Resour. Res., 58, e2021WR030386, 10.1029/2021WR030386, 2022. Yassin, F., Razavi, S., Elshamy, M., Davison, B., Sapriza-Azuri, G., and Wheater, H.: Representation and improved parameterization of reservoir operation in hydrological and land-surface models, Hydrol. Earth Syst. Sci., 23, 3735–3764, 10.5194/hess-23-3735-2019, 2019.Yin, Z., Lin, P., Riggs, R., Allen, G. H., Lei, X., Zheng, Z., and Cai, S.: A synthesis of Global Streamflow Characteristics, Hydrometeorology, and Catchment Attributes (GSHA) for large sample river-centric studies, Earth Syst. Sci. Data, 16, 1559–1587, 10.5194/essd-16-1559-2024, 2024.Yue, S. and Gan, T. Y.: Simple scaling properties of Canadian annual average streamflow, Adv. Water Resour., 27, 481–495, 10.1016/j.advwatres.2004.02.019, 2004.Zajac, Z., Revilla-Romero, B., Salamon, P., Burek, P., Hirpa, F. A., and Beck, H.: The impact of lake and reservoir parameterization on global streamflow simulation, J. Hydrol., 548, 552–568, 10.1016/j.jhydrol.2017.03.022, 2017.