Introduction

Streamflow on many rivers in the United States is regulated by the use of dams and their reservoirs, which store and release water, altering the natural flow regime. The effects of this regulation vary depending on the size of the reservoir as well as the purpose of the dam, such as irrigation, hydropower, or flood control (Collier and others, 1996). Scientists frequently make decisions about the use of long-term streamgage records in hydrologic studies based on available regulation information. If a study has a goal of understanding the effects of natural climate variability or anthropogenic climate change on flooding, the set of streamgages used should reflect those not modified by reservoirs or substantial land-use change. In studies focused on peak streamflow (peak) values, the peaks included in the study should be unaltered by regulation.

Sources used for information on the degree of regulation on a particular river reach or streamgage basin vary by study and include information from the peak-flow file (PFF) published by the U.S. Geological Survey (USGS), using institutional knowledge from local analysts familiar with the study area, and using reservoir storage and streamflow information to develop a method unique to a specific study. Few of these methods are systematic; thus, consistent application across a large region would be difficult. Sources of data on the presence of dams and reservoirs, like the National Inventory of Dams (NID), are not all-inclusive and often are missing some dams and smaller structures like stock ponds.

One source of information on regulation is the PFF, which is available as part of the USGS National Water Information System (NWIS) database (U.S. Geological Survey, 2021b). In addition to annual peak values, the PFF dataset may contain qualification codes that are applied to some peak values to provide additional information about the forces that generated the peak value. Qualification code definitions can be viewed in NWIS by visiting the page for a specific streamgage, clicking on “Peak streamflow,” and then clicking on “Tab-separated file” under “Output formats” (U.S. Geological Survey, 2021c). For sites with an active Next Generation Monitoring Location page, a user can get to the same location by selecting “Data Inventory” and then “Peak Streamflow” (U.S. Geological Survey, 2021b).

The qualification codes describing regulation are code 5, indicating that streamflow was affected to an “unknown degree by Regulation or Diversion,” and code 6, indicating that streamflow was “affected by known Regulation or Diversion” (U.S. Geological Survey, 2021c). These codes have important implications for streamgage selection in studies and for flood-frequency analysis because PeakFQ, the USGS program used for flood-frequency analysis, uses peaks qualified with code 5 but excludes peaks qualified with code 6 unless otherwise specified by the user; however, these descriptions are open to interpretation. Code 5 peaks have been considered to have less than 10 percent of the peak streamflow affected by regulation (Ryberg and others, 2017) whereas peaks qualified with code 6 were affected by known and planned regulation affecting peaks by more than 10–15 percent (Ryberg, 2008; Ryberg and others, 201714). Throughout the history of the USGS PFF, technical reviews, regional-flood studies, and user inquiries have indicated inconsistencies among USGS Water Science Centers and over time in the use of these and other qualification codes, and efforts have been made to improve the consistency of the application of the codes (Ryberg, 2008; Ryberg and others, 2017). However, the definition of codes 5 and 6 have not been revised to be more specific and “the difference between code 5…and code 6…is admittedly fuzzy and subject to differing interpretations. The difference is mainly one of certainty that there is a regulation effect on the peak discharge” (Ryberg and others, 2017, p. 9).

When it comes to streamgage selection and better determination of regulation effects, several tactics have been used. The USGS Hydro-Climatic Data Network (HCDN) was identified in 1992 as a subset of all USGS streamgages for which the streamflow primarily reflects prevailing meteorological conditions for specified years (Slack and Landwehr, 1992). The HCDN was updated in 2009 (hereafter referred to as “HCDN-2009”; Lins, 2012) to address questions concerning the number of original streamgages that maintained their status as sensitive to climate, streamgages that no longer remained operational, and streamgages that did not qualify as HCDN sites in 1988 (the last year of the original data evaluation) because their records were too short and did not have sufficiently long streamflow records for climate-sensitivity studies by 2009. Many of those same questions would be beneficial to revisit if one were to use HCDN-2009 sites. In related efforts, the USGS National Water-Quality Assessment Program analyzed reference sites to identify those streamgages minimally affected by direct human activities in the conterminous United States (Falcone and others, 2010). The list of reference and nonreference sites and associated basin characteristics was called Geospatial Attributes of Gages for Evaluating Streamflow (GAGES) in 2010 and updated to GAGES-II in 2011 (Falcone and others, 2010; Falcone, 2011). GAGES-II has been widely used because it provides geospatial data for each streamgage in the dataset; however, GAGES-II has two shortcomings for streamgage selection. First, as with HCDN-2009, the dataset has not been updated. Because reference status can change over time, some streamgages may have been discontinued and some not initially included because of lack of data may now have sufficient data. In addition, if part of the drainage area upstream from a streamgage was in Canada or Mexico, the streamgage was not included in GAGES-II because of the difficulty in obtaining or harmonizing data across international boundaries. The exclusion of transboundary watersheds left out some streamgages with long-term records of good quality data, which was particularly of concern for Montana and North Dakota. Given the shortcomings of the peak qualification codes and the HCDN-2009 and GAGES-II datasets and the lack of updates, many alternative methods of streamgage selection have been used.

One option for selecting streamgages, or selecting individual peaks from a streamgage, despite known shortcomings, is to assume the peak qualification codes are correct and represent the best judgement of USGS subject matter experts. In this case, analysts may remove peak values qualified with codes 5, 6, and other codes indicating unnatural conditions or uncertainty in the data. Code 3 is often removed from analyses because it represents “Discharge affected by Dam Failure” (U.S. Geological Survey, 2021c). Code C may also be removed, depending on the purposes of the study, because it represents peaks that have been “Affected by Urbanization, Mining, Agricultural changes, Channelization, or other” (U.S. Geological Survey, 2021c). Code A indicates that the year of occurrence is unknown or not exact and code B indicates that the month or day of occurrence is unknown or not exact (Ryberg, 2008; Ryberg and others, 2017). Depending on the goal of the study, some analysts may remove peaks qualified with codes A and B, whereas others may use them; for example, within Ryberg and others (2016), individual peak values qualified with codes 3, 5, 6, A, B, and C were removed from the study and then streamgages were required to have least 15 remaining peak values for inclusion in the study.

Another option is initial screening of peaks based on the peak qualification codes and subsequent inclusions or exclusions of peaks based on institutional knowledge and discussions with subject matter experts (for example, Hirsch and Ryberg, 2012; Ryberg and others, 2014). This option requires the analysts to know who the subject matter experts are and can require substantial time, depending on the extent of the study.

In other cases, analysts develop their own regulation criteria for streamgage selection or to categorize results, such as calculating a flow-based normalized dam storage above a streamgage and picking a threshold value to indicate a regulated basin (Hodgkins and others, 2019; Ryberg and others, 2020). Sando and McCarthy (2018) developed major and minor regulation definitions. Major dam regulation status was applied to streamgages where a single upstream dam had a drainage area that exceeded 20 percent of the drainage area of the streamgage. Minor dam regulation status was applied to streamgages where the cumulative drainage area of all upstream dams exceeded 20 percent of the drainage area of the streamgage, but no single upstream dam had a drainage area that exceeds 20 percent of the drainage area of the streamgage (Sando and McCarthy, 2018). Depending on data availability and institutional knowledge, others have used a variety of methods including the comparison of peak values pre- and postreservoir operations to determine the effect of regulation and the examination of additional datasets, including data describing regulation location and purpose, the fraction of total basin drainage area that is controlled, the flood-storage capacity of dams in the basin, and other management practices (Asquith, 2001; Sando and others, 2008).

This exercise in determining regulated and (relatively) unregulated streamgages for studies is repeatedly performed and revised by USGS scientists and others that use USGS data. Recent efforts have been made to determine the effects of dams on a more systematic scale by defining a dam impact/disturbance metric for flowlines in the conterminous United States (Wieczorek and others, 2021) using information from the NID (U.S. Army Corps of Engineers, 2018). This impact metric was used to identify a threshold in an eight-State study area and apply it to streamgages in a nine-State area to identify those that are largely unregulated. The method removes some of the subjectivity from the process of identifying regulated streamgage basins, is reproducible, and could be used to develop more quantitative descriptions of peak-streamflow qualification codes in the future.

Description of Study Area

The study area includes the States of Illinois, Iowa, Michigan, Minnesota, Missouri, Montana, North Dakota, South Dakota, and Wisconsin (fig. 1). The study area includes basins outside of those State boundaries whose streamflow flows into any of the nine States in the study area, including streamgages in Arkansas, Idaho, Indiana, Kansas, Nebraska, Washington, and Wyoming. These States are largely within the upper Mississippi River Basin and vary in topography, ecoregions, land use, land cover, and climate. Dams are prevalent throughout the study area but are unevenly distributed throughout the region. The application of code 6 peak qualification codes varies throughout the study area; for example, in contrast with several other States in the region, it is more common to apply code 5 to peaks in Montana streamgage basins with some regulation, and therefore Montana has very few streamgages with a code 6.

Methods

An impact metric value denoting a threshold above which a streamgage basin would be considered regulated was established using streamgages with known regulation (streamgages with peaks qualified with code 6). This threshold was developed using streamgages from only eight States, excluding Michigan, because the Michigan Department of Transportation joined the Transportation Pooled Fund study after this portion of the analysis had concluded. The threshold was then applied to a selection of streamgages from the full nine-State study area that do not have peaks qualified with code 6 to determine which of these streamgages had a dam impact metric equal to or greater than the threshold at some point in their periods of record.

Results

An impact metric value denoting a threshold above which a streamgage basin would be considered regulated was established using 140 streamgages and applied to 1,017 streamgages without known regulation (table 1). Results from this assessment are provided in the following sections.

Identification of Regulated Basins

Regulated basins in the study area were identified by applying the established threshold of 0.1 to each noncode 6 streamgage. Of the 1,017 noncode 6 streamgages (some with code 5 peaks, denoting an unknown degree of regulation or diversion) evaluated, 136 (about 13 percent) had an impact metric over 0.1 at some point in their periods of record. Of those 136 streamgages, a minimum impact metric of 0 and maximum of 5,085 was observed. The large range of the impact metric indicates there is a large variation in the degree of impact of regulation greater than the identified 0.1 threshold. For the 136 streamgages identified as regulated, the median impact metric of the entire active period of record is presented in table 2 and figure 3. A small number of regulated streamgages had a median impact metric less than 0.1 during their entire active period of record, despite their maximum impact metric value being greater than the 0.1 threshold.

Table 2.    

U.S. Geological Survey streamgages without annual peak streamflow values qualified as regulated (code 6) that were determined to be regulated in this evaluation.

[YYYY, year; MM, month; DD, day]

Table 2.    U.S. Geological Survey streamgages without annual peak streamflow values qualified as regulated (code 6) that were determined to be regulated in this evaluation.

U.S. Geological Survey streamgage number	U.S. Geological Survey streamgage name	Record start date (YYYY–MM–DD)	Record end date (YYYY–MM–DD)	Peak code 5	Median impact metric (dimensionless)
03382100	South Fork Saline River near Carrier Mills, Illinois	1966–04–28	2020–01–12	False	0.290
04024000	St. Louis River at Scanlon, Minnesota	1953–08–07	2020–04–03	True	0.127
04037500	Cisco Branch Ontonagon River at Cisco Lake Outlet, Michigan	1945–07–06	2020–05–01	True	0.117
04040000	Ontonagon River near Rockland, Michigan	1942–08–22	2020–03–30	True	0.146
04056500	Manistique River near Manistique, Michigan	1938–04–01	2020–04–05	False	0.466
04071000	Oconto River near Gillett, Wisconsin	1921–04–29	2020–03–31	False	0.571
04078500	Embarrass River near Embarrass, Wisconsin	1921–03–21	2020–03–30	False	0.597
04079000	Wolf River at New London, Wisconsin	1968–07–05	2020–04–03	False	0.262
04097500	St. Joseph River at Three Rivers, Michigan	1954–03–29	2020–05–20	False	0.406
04099000	St. Joseph River at Mottville, Michigan	1924–03–31	2020–05–20	True	0.300
04101500	St. Joseph River at Niles, Michigan	1931–03–30	2020–05–20	False	0.156
04170000	Huron River at Milford, Michigan	1949–02–15	2020–05–21	True	0.572
04172000	Huron River near Hamburg, Michigan	1952–04–17	2020–01–15	False	0.284
04174500	Huron River at Ann Arbor, Michigan	1968–06–26	2020–05–19	True	0.135
04175600	River Raisin near Manchester, Michigan	1970–03–05	2020–03–29	False	0.252
05017500	St. Mary River near Babb, Montana	1921–05–27	2020–07–02	False	0.130
05030500	Otter Tail River near Elizabeth, Minnesota	1992–08–13	2019–05–22	True	0.719
05066500	Goose River at Hillsboro, North Dakota	1931–04–07	2020–04–02	False	0.393
05076000	Thief River near Thief River Falls, Minnesota	1922–05–12	2020–04–09	False	0.230
05079000	Red Lake River at Crookston, Minnesota	1922–05–13	2020–04–08	False	0.546
05084000	Forest River near Fordville, North Dakota	1940–04–17	2020–04–07	True	0.108
05129115	Vermilion River near Crane Lake, Minnesota	1980–04–21	2020–04–11	False	0.003
05229450	Pine River near Pine River, Minnesota	1986–03–28	2020–04–03	False	0.367
05244000	Crow Wing River at Nimrod, Minnesota	1931–06–12	2020–04–04	False	0.163
05245100	Long Prairie River at Long Prairie, Minnesota	1972–07–22	2019–10–06	False	0.756
05270500	Sauk River near St. Cloud, Minnesota	1930–05–16	2019–10–27	False	0.195
05275000	Elk River near Big Lake, Minnesota	1932–04–11	2020–04–04	False	0.182
05288500	Mississippi River at Highway 610 in Brooklyn Park, Minnesota	1931–06–16	2020–04–06	True	0.207
05293371	Pomme De Terre River below Elbow Lake, Minnesota	1986–06–23	2019–10–23	False	0.401
05294000	Pomme De Terre River at Appleton, Minnesota	1931–03–26	2019–10–24	False	0.155
05315000	Redwood River near Marshall, Minnesota	1940–03–30	2020–03–29	False	0.195
05320480	Maple River near Rapidan, Minnesota	1973–03–14	2019–10–06	False	0.134
05331000	Mississippi River at St. Paul, Minnesota	1921–06–16	2019–03–31	True	0.122
05344500	Mississippi River at Prescott, Wisconsin	1929–03–25	2019–04–01	False	0.174
05355024	Cannon River at Northfield, Minnesota	1980–03–19	2019–03–22	False	0.126
05369000	Red Cedar River at Menomonie, Wisconsin	1921–03–28	2020–06–30	True	0.526
05378500	Mississippi River at Winona, Minnesota	1929–04–03	2019–04–02	False	0.291
05404000	Wisconsin River near Wisconsin Dells, Wisconsin.	1935–03–27	2020–04–01	True	0.127
05407000	Wisconsin River at Muscoda, Wisconsin	1921–05–06	2020–04–04	True	0.109
05426000	Crawfish River at Milford, Wisconsin	1931–11–26	2019–10–09	False	13.0
05430500	Rock River at Afton, Wisconsin	1921–05–02	2019–10–15	True	3.05
05437500	Rock River at Rockton, Illinois	1937	2019–10–07	False	1.58
05443500	Rock River at Como, Illinois	1921–04–28	2019–10–05	False	1.14
05446500	Rock River near Joslin, Illinois	1940–08–30	2019–10–07	True	1.04
05464000	Cedar River at Waterloo, Iowa	1941–04–21	2020–06–24	False	0.969
05464500	Cedar River at Cedar Rapids, Iowa	1921–06–01	2020–06–25	False	0.759
05465000	Cedar River near Conesville, Iowa	1929	2020–06–27	False	0.630
05486000	North River near Norwalk, Iowa	1940–08–01	2019–10–04	False	0.949
05487980	White Breast Creek near Dallas, Iowa	1963–03–19	2018–10–09	False	0.055
05507600	Lick Creek at Perry, Missouri	1969–10–12	2020–05–24	False	0.272
05507800	Salt River near Center, Missouri	1973–04–22	2020–02–12	True	0.409
05520500	Kankakee River at Momence, Illinois	1921–03–25	2020–05–18	False	5.65
05527500	Kankakee River near Wilmington, Illinois	1921–03–30	2020–05–18	False	2.54
05531300	Salt Creek at Elmhurst, Illinois	1960–03–30	2020–05–17	False	0.117
05543830	Fox River at Waukesha, Wisconsin	1963–03–21	2020–05–19	False	0.160
05568500	Illinois River at Kingston Mines, Illinois	1941–04–24	2020–05–21	True	0.867
05586100	Illinois River at Valley City, Illinois	1921–05–05	2019–05–10	True	0.522
05587450	Mississippi River at Grafton, Illinois	1986–10–10	2019–06–07	True	0.308
05594000	Shoal Creek near Breese, Illinois	1946–08–18	2020–01–13	False	0.156
06012500	Red Rock River below Lima Reservoir near Monida, Montana	1926–04–21	2020–05–01	True	0.189
06016000	Beaverhead River at Barretts, Montana	1921–06–17	2020–07–23	True	0.108
06018500	Beaverhead River near Twin Bridges, Montana	1936–08–12	2019–10–14	True	0.132
06038500	Madison River below Hebgen Lake near Grayling, Montana	1940–07–25	2020–06–05	True	0.334
06038800	Madison River at Kirby Ranch near Cameron, Montana	1959–10–23	2020–06–07	True	0.313
06041000	Madison River below Ennis Lake near Mcallister, Montana	1943–06–02	2020–06–01	True	0.190
06066500	Missouri River below Holter Dam near Wolf Creek, Montana	1946–06–03	2020–07–04	True	0.199
06078200	Missouri River near Ulm, Montana	1958–06–15	2020–07–01	True	0.170
06089000	Sun River near Vaughn, Montana	1934–06–08	2020–06–02	False	0.114
06090300	Missouri River near Great Falls, Montana	1952–05–04	2020–07–01	True	0.164
06090800	Missouri River at Fort Benton, Montana	1921–06–16	2020–07–01	True	0.150
06101500	Marias River near Chester, Montana	1921–05–27	2020–06–13	True	0.396
06108000	Teton River near Dutton, Montana	1955–07–17	2020–07–02	False	0.101
06109500	Missouri River at Virgelle, Montana	1935–06–16	2020–07–02	True	0.173
06115200	Missouri River near Landusky, Montana	1934–06–12	2020–07–03	True	0.149
06126500	Musselshell River near Roundup, Montana	1946–07–04	2020–05–23	True	2.62
06127500	Musselshell River at Musselshell, Montana	1929–06–05	2020–05–24	True	2.37
06130500	Musselshell River at Mosby, Montana	1929–06–19	2020–05–25	False	1.44
06135000	Milk River at Eastern Crossing of International Boundary, Montana	1921–04–04	2020–04–25	True	0.320
06139500	Big Sandy Creek near Havre, Montana	1947–03–21	2020–05–16	False	0.181
06140500	Milk River at Havre, Montana	1921–04–06	2020–06–03	True	0.299
06154100	Milk River near Harlem, Montana	1960–03–27	2020–04–10	True	0.180
06155030	Milk River near Dodson, Montana	1983–07–12	2019–03–23	False	0.158
06164510	Milk River at Juneberg Bridge near Saco, Montana	1978–04–03	2020–03–31	True	0.110
06177000	Missouri River near Wolf Point, Montana	1929–06–04	2019–10–01	True	0.526
06185500	Missouri River near Culbertson, Montana	1942–08–30	2020–04–03	True	0.488
06204050	West Rosebud Creek near Roscoe, Montana	1966–07–15	2020–06–30	True	0.157
06235500	Little Wind River near Riverton, Wyoming	1941–06–08	2020–06–07	True	0.196
06287000	Bighorn River near St. Xavier, Montana	1935–06–16	2020–07–03	True	0.183
06294400	Andresen Coulee near Custer, Montana	1963–06–04	2019–03–21	False	0.162
06294500	Bighorn River above Tullock Creek near Bighorn, Montana	1982–07–05	2020–04–14	True	0.163
06349500	Apple Creek near Menoken, North Dakota	1947–03–26	2020–03–12	False	0.364
06407500	Spring Creek near Keystone, South Dakota	1946–05–03	2020–05–05	True	0.118
06447000	White River near Kadoka, South Dakota	1942–06–04	2020–05–16	False	1.15
06449100	Little White River near Vetal, South Dakota	1960–03–24	2020–03–07	True	8.18
06449500	Little White River near Rosebud, South Dakota	1944–05–17	2020–03–08	True	4.95
06450500	Little White River below White River, South Dakota	1930–05–11	2020–03–08	True	3.30
06452000	White River near Oacoma, South Dakota	1929–03–10	2020–05–19	False	1.09
06473000	James River at Ashton, South Dakota	1946–03–28	2020–04–20	False	26.4
06475000	James River near Redfield, South Dakota	1950–05–22	2020–03–13	False	18.4
06476000	James River at Huron, South Dakota	1922–03–22	2020–03–21	False	15.7
06477000	James River near Forestburg, South Dakota	1922	2020–03–23	False	13.7
06478500	James River near Scotland, South Dakota	1929–03–31	2020–03–15	False	11.2
06478513	James River near Yankton, South Dakota	1982–06–02	2020–03–20	False	11.0
06818000	Missouri River at St. Joseph, Missouri	1922–06–28	2019–10–03	True	5,084
06821150	Little Platte River at Smithville, Missouri	1947	2019–05–29	True	0.515
06893000	Missouri River at Kansas City, Missouri	1929–06–05	2019–10–04	True	4,206
06894000	Little Blue River near Lake City, Missouri	1948–03–20	2020–05–28	False	0.246
06895500	Missouri River at Waverly, Missouri	1929–06–05	2020–05–29	True	4,168
06909000	Missouri River at Boonville, Missouri	1926–09–25	2019–10–06	True	3,917
06934500	Missouri River at Hermann, Missouri	1929–06–08	2020–05–30	True	3,564
07010000	Mississippi River at St. Louis, Missouri	1921–05–14	2019–06–08	True	2,284
07018100	Big River near Richwoods, Missouri	1950–01–04	2020–01–11	False	0.212
07018500	Big River at Byrnesville, Missouri	1923–05–17	2020–01–13	False	0.177
07043500	Little River Ditch No. 1 near Morehouse, Missouri	1946–05–03	2020–01–11	False	0.129
07189000	Elk River near Tiff City, Missouri	1940–04–12	2019–11–07	True	0.003
12301933	Kootenai River below Libby Dam near Libby, Montana	1972–07–28	2018–12–20	True	0.358
12323250	Silver Bow Creek below Blacktail Creek, at Butte, Montana	1984–06–11	2020–06–08	False	0.985
12323600	Silver Bow Creek at Opportunity, Montana	1989–03–09	2020–06–17	False	0.373
12323770	Warm Springs Creek at Warm Springs, Montana	1984–06–21	2020–06–08	False	0.423
12323800	Clark Fork near Galen, Montana	1989–03–10	2020–06–18	False	18.0
12324200	Clark Fork at Deer Lodge, Montana	1979–05–25	2020–06–09	False	8.09
12324680	Clark Fork at Goldcreek, Montana	1978–05–24	2020–06–09	False	4.65
12334510	Rock Creek near Clinton, Montana	1972–06–01	2020–05–21	False	0.645
12334550	Clark Fork at Turah Bridge near Bonner, Montana	1986–02–25	2020–05–21	False	2.19
12340500	Clark Fork above Missoula, Montana	1930–04–28	2020–05–21	False	1.50
12342500	West Fork Bitterroot River near Conner, Montana	1941–05–28	2020–05–21	True	13.0
12344000	Bitterroot River near Darby, Montana	1938–05–29	2020–05–31	False	4.35
12352500	Bitterroot River near Missoula, Montana	1990–05–31	2020–06–01	False	1.83
12353000	Clark Fork below Missoula, Montana	1930–04–26	2020–05–21	False	1.59
12354500	Clark Fork at St. Regis, Montana	1921–05–28	2020–05–22	False	1.26
12362500	South Fork Flathead River near Columbia Falls, Montana	1921–05–26	2020–06–08	True	304
12363000	Flathead River at Columbia Falls, Montana	1928–05–26	2020–06–01	True	113
12372000	Flathead River near Polson, Montana	1921–06–11	2020–06–08	True	345
12388700	Flathead River at Perma, Montana	1984–06–29	2020–06–08	True	300
12389000	Clark Fork near Plains, Montana	1921–05–28	2020–06–02	True	154
12395000	Priest River near Priest River, Idaho	1936–05–16	2020–06–01	True	0.988

Limitations

This analysis was conducted in an effort to establish a method that consistently identifies the degree of regulation at USGS streamgage basins over large regions and apply this method to a nine-State study area. Although this method allows for a consistent approach over a large area and integrates important information to consider when assessing potential regulation effects on peak flows, like upstream storage and watershed area, it also has limitations. Several of these limitations are mentioned in the metadata that accompanies the original impact metric dataset (Wieczorek and others, 2021). The NID contains more than 91,000 dams that meet a list of criteria: dams that would result in a loss of human life or disrupt access to critical facilities or damage facilities if there was downstream flooding, dams that equal or exceed 25 feet in height and exceed 15 acre-feet in storage, or dams with storage greater than or equal to 50 acre-feet and exceeding 6 feet in height (U.S. Army Corps of Engineers, 2018). Some dams that meet these criteria are not yet included in the NID. Additionally, dams that do not meet these criteria are not included in the NID and thus are not included in the impact metric dataset (Wieczorek and others, 2021). There are other alterations of streamflow on the landscape, such as diversions for irrigation or municipal supply and small dams associated with water rights (such as stock dams), and these alternations do not meet the criteria for inclusion in the NID (McCarthy and others, 2016). Data are sparse for small impoundments, which has made understanding their effect on all metrics of streamflow difficult, but the density of impoundments has a cumulative effect on streamflow and small impoundments alter low streamflow conditions (Brogan and others, 2021). Dams are occasionally removed, and how the removal of these structures is handled by the NID is not clear. Some dam reservoirs that function as run-of-the-river dams, which do not affect downstream streamflow in the same way as a dam that retains flow, are included in the NID dataset and may lead to overestimation on regulation impact in some cases. Additionally, in study areas such as this one, that border Canada or Mexico, there often are long-term streamgages (for monitoring of transboundary waters) with part of the drainage area outside of the United States, and extraterritorial dams are not in the NID. These streamgages are often of interest because of their long record and may require subject matter expertise, searching of databases from the other country, or extensive examination of maps looking for reservoirs. Dams with no completion date were not included in the impact metric dataset (Wieczorek and others, 2021); of more than 91,000 dams in the NID, about 17 percent of dams do not have a completion date (U.S. Army Corps of Engineers, 2018). The impact metric computation uses the maximum storage from the NID dataset, which represents the maximum amount of water that can be stored behind each dam and may lead to an overestimation of the real effects of dam storage on the streamgage basins.

The dams used in this study were snapped to NHD flowlines using automated procedures that require that the NID dam location coordinates be relatively accurate. The quality of the dam locations varies among regions and through time, and improvements on the dam locations could alter the results presented in this report. For example, in Montana, these dams were reviewed and revised for implementation in a web-based application and this effort determined that many of the dams were mislocated in the NID (Pete McCarthy, USGS, written commun., 2022). Some dams were also determined to not be on the main river system but were snapped to flowlines in the analysis. A smaller scale, State-based evaluation would likely find many of these metrics to be different, in some cases substantially.

The NHDPlus data are another input into the impact metric dataset and do include some network errors for flowlines, streamgage locations, and dam locations (Wieczorek and others, 2018). Improved flowline network representations could improve the accuracy of the impact metric used in this report. The large range of impact metric values for the streamgages evaluated indicate that in some instances the degree of regulation is far beyond the 0.1 threshold identified. Further evaluation of these instances would help the understanding of where improvements to the dataset may be made.

The limitations of this method may vary depending on the region where it is applied. In the nine-State study area used in this report, it seems that the most common result of these limitations was streamgages being incorrectly identified as regulated. For example, Andresen Coulee near Custer, Montana (USGS streamgage 06294400), which, based on local knowledge and observance on aerial imagery, has no dams within the basin, has a median impact metric of 0.162 during its period of record in this analysis. This instance may be a result of incorrect NID dam locations or incorrect snapping of those dam locations to NHD flowlines. When used as a screening tool to remove regulated basins, a more conservative approach (where unregulated basins are removed incorrectly) is less problematic than if regulated basins are incorrectly included. However, depending on where and how this method is applied, additional evaluation of results may be necessary. The large scale of this evaluation method will naturally have less detail at the local level that may otherwise be present in an approach that integrates institutional knowledge from subject matter experts. Future improvements to the data available may allow for refinements of the dam metric or the threshold selection.

Summary

Determining a set of streamgages that are not affected by regulation is a common exercise among scientists conducting various hydrologic studies. Different methods have been used to establish which streamgages are relatively unregulated in U.S. Geological Survey studies, but inconsistencies in datasets make some methods difficult to apply over a large study region. A recent U.S. Geological Survey data release provides a decadal impact metric, based on the sum of the upstream reservoir storage multiplied by the ratio of the dam drainage area to the National Hydrography Dataset flowline drainage area, divided by the volume of precipitation in the National Hydrography Dataset flowline, which characterizes the degree of regulation of a given river reach, allowing for a more systematic approach. A threshold denoting regulation was established using impact metric values from streamgages in eight States with peaks qualified with code 6, which have known regulation or diversion. This threshold was applied to streamgages in the nine-State study area to identify regulated basins. Using this dataset, 140 streamgages in the study area with known regulation (peaks qualified with code 6) were evaluated for their impact metric values in decades with code 6 peaks. A threshold value of 0.1 was identified as the value most characteristic of the regulated streamgages in the study area. This threshold was used to identify noncode 6 streamgages in the study area that had an impact metric equal to or greater than the identified threshold at some time in their active period of record. Impact metric values for 1,017 noncode 6 streamgages (some with code 5 peaks) were evaluated. Thirteen percent of streamgages examined (136 of 1,017 streamgages) had an impact metric equal to or greater than the identified regulated threshold at some point in their periods of record.