Land cover and land use in the MHRB ranges from shrub and scrub lands to evergreen forests to developed and agricultural lands. The primary land covers in the MHRB, based on the 2016 National Land Cover Database (Dewitz, 2019; Jin and others, 2019), were shrub/scrub (59.2 percent), herbaceous (32.0 percent), and evergreen forest (2.9 percent). Other land cover types were the remaining 6 percent of total land cover. The lowest percentage of land-cover types (each with less than 1 percent of the total land cover in the MHRB) were low, medium, and high-intensity developed lands, open water, and mixed forest (table 3). Cultivated cropland cover is 1.1 percent of the total area of the MHRB and was mostly in Paradise Valley, Antelope Valley, middle Reese River Valley, Grass Valley (near Winnemucca), Winnemucca, and parts of Boulder Flat (table 3). Evergreen forest land cover is prominent in the high-elevation southern parts of the MHRB.

In the MHRB, the Humboldt River and its floodplain occupy a valley that ranges from less than 1,000 ft wide several miles west of Carlin, Nevada, to several miles wide in the Battle Mountain, Nevada, area. The Humboldt River mainstem is approximately 262 mi long through the MHRB between the Carlin and Imlay gages (fig. 2). In general, the Humboldt River receives flow from a combination of tributary inflows and groundwater discharge from its headwaters near and upstream from Elko, Nevada, downstream to the Palisade gage (10322500; Prudic and others, 2006). The Humboldt River at Palisade generally has the greatest streamflow, but streamflow begins decreasing below Palisade because of infiltration to shallow aquifers and irrigation diversions (Prudic and others, 2006). The mean streamflow of the Humboldt River at Palisade is about 282,000 acre-ft/yr (about 390 ft³/s) for the period of record from WY 1903 through 2016 (U.S. Geological Survey, 2021).

The main tributaries to the Humboldt River in the MHRB are Maggie Creek, Marys Creek, Susie Creek, Rock Creek, and Pine Creek on the eastern half of the MHRB (streamgage locations are provided in decimal degrees, NAD 83, and streamgage altitudes are provided in feet above North American Vertical Datum of 1988 [NAVD 88] or National Geodetic Vertical Datum of 1929 [NGVD 29]; fig. 2; table 4). The Reese River, south of the Humboldt River, has a large contributing drainage area but is considered a small and insignificant tributary that rarely contributes flow to the Humboldt River because of climate and underlying geology with high permeability. The Little Humboldt River to the northwest also has a large contributing drainage area, but the river rarely contributes flow to the Humboldt River because of water use in Paradise Valley for irrigation, climate, and the permeable geology underlying the riverbed. During periods of excess runoff, however, the Little Humboldt River can contribute flow to the Humboldt River.

Large-scale mining of low-grade gold deposits in northern Nevada started in the mid-1960s and increased for the next several decades (Muntean and others, 2017). Many of these deposits were in the MHRB, and by the 1990s, some of the surface mines in the MHRB began to require dewatering operations. This study included 10 mine operations with dewatering activities during the 1990s, 2000s, and 2010s (fig. 2). The Lone Tree mine operation north of Battle Mountain, Nevada (fig. 2), operated from about 1990 through 2007. The Phoenix mine operation south of Battle Mountain, Nevada (fig. 2), was an expansion of previous mine operations in the area but began its production of gold, silver, and copper in 2001 and was still in operation as of 2016. The Twin Creeks mine operation in the northwest part of the study area (fig. 2) began gold production operations in about 1993 and has remained in operation since 2016. The Carlin trend area north of Palisade, Nevada (fig. 2) includes the Carlin north and south mine operations. The Carlin north mine operation began gold production in the mid-1960s, and the south mine operation began in the mid-1980s. Both mines were operating as of 2016. The mine operation in Cortez Hills is a gold producer and, in this study, consisted of the Cortez Hills mine operation and the Pipeline Mine operation in the south part of Crescent Valley, and collectively are referred to as the Cortez Hills Mine Complex in this report (fig. 2). The Cortez Hills operation began in the early 1940s, with large-scale ore processing starting in the late 1990s, and remained in operation as of 2016 (Muntean and others, 2017). The Goldstrike mine operation north of Dunphy, Nevada (fig. 2), operates the Betze-Post open pit, one of the largest in the world, and several adjacent underground operations. Production at the Betze-Post open pit began and stopped in the late 1970s and started again in the mid-1980s. The pit remained in operation as of 2016. Mine operations in Turquoise Ridge (or Getchell), near the Twin Creeks mine operation (fig. 2), began large-scale production in about 1989, but parts of the mine had operated since the late 1930s. The Cove-McCoy mine operation area south of Battle Mountain (fig. 2) began production in about 1986 and ended production in about 2006. The Pinson mine operation was placed into care and maintenance status at the end of 2015 because of lack of ore (Muntean and others, 2017).

Diversion of flow from the Humboldt River and its tributaries for irrigating meadows and crops has been the principal use of surface water in the HRB. The first large ranching operations in the HRB were established between about 1861 and 1872 (Horton, 2000a). The completion of the Transcontinental Railroad in 1869, which went through Nevada, was particularly important to the establishment of these operations because the railroad allowed for transportation of cattle to markets.

During high flows of spring and early summer, the middle Humboldt River can overtop its banks, flow into abandoned channels and oxbows, and naturally irrigate low-lying meadows. In the 1860s, ranchers and farmers began diverting water by impounding the river and allowing the river to flood its banks to irrigate crops and meadows. The first diversions, based on year of priority, were put into use in the Lovelock area in 1861, in the Elko area in 1862, and in the Winnemucca and Imlay areas in 1863 (Hennen, 1964b). By the late 1800s, diversions on the upper Humboldt River were so numerous that streamflow rarely reached the Lovelock area in average or dry years (Horton, 2000a). Conflicts among upstream and downstream users intensified in the early 1900s, causing the adjudication of Humboldt River water rights in the 1930s (Mashburn and Mathews, 1943) and the construction of the Rye Patch dam and Reservoir (fig. 2) in 1935–36.

The process of water-rights adjudication on the Humboldt River began with the filing of the Nevada State Engineer’s Final Order of Determination in the District Court of Humboldt County in 1923. In 1943, Mashburn and Mathews (1943) summarized the order and follow-on court decrees and decisions in a document titled “The Humboldt River Adjudication.” The Humboldt River Adjudication, collectively called the “Humboldt Decree,” included the Edwards Decree and Bartlett Decree, intervening orders by two other judges and a final decision by the Nevada Supreme Court ending litigation (Mashburn and Mathews, 1943). In addition to establishing priorities by year for individual water rights, the adjudication also established two districts (upper and lower) on the river with irrigation seasons of differing length and identified three land classes and irrigation requirements for each.

The upper district on the Humboldt River is the part of the HRB upstream from the Palisade gage (table 4), and the lower district extends from Palisade to Lovelock Valley (Malone, 1932). The irrigation season in the upper district extends from April 15 to August 15 each year and in the lower district from March 15 to September 15. The three recognized land classes and court approved water requirements are (1) harvest lands—3 feet per year (ft/yr); (2) meadow pasture—1.5 ft/yr; and (3) diversified pasture—0.75 ft/yr. The different water requirements are because of different irrigation seasons for each land class with harvest lands having a full season, meadow pasture having a half season, and diversified pasture having the first quarter of a season.

The Bartlett Decree recognized that the total cultivated area using water from the Humboldt River was 285,238 acres (Mashburn and Mathews, 1943). The decree further recognized that 698,379 acre-feet (acre-ft) would be required to satisfy all water rights with a priority of 1928 or earlier. Because the mean annual flow of the river at Palisade at the time of the decree was 255,650 acre-ft, serving all water rights in a year of mean streamflow depended on irrigation return flows to the river (Hennen, 1964a) and receiving adequate streamflow for the duration of the irrigation season. In 1963, decreed water rights of 666,680 acre-ft were under irrigation in the HRB for 265,791 acres of land (Hennen, 1964a). In 2000, decreed water rights of 674,581 acre-ft were under irrigation for 270,978 acres of land (Horton, 2000a). The irrigated lands within the MHRB consisted of about 109,319 acres in 1931 (Mashburn and Mathews, 1943) and about 76,544 acres in 1963 (Hennen, 1964a). The substantial reduction in irrigated acreage in the MHRB between 1931 and 1963 mostly was because of the purchase and transfer of approximately 32,182 acres by the Pershing County Water Conservation District later in 1931 (Horton, 2000b), with the remainder attributed to tabulation errors in the Bartlett Decree (Hennen, 1964a). The acreage and water rights values for 1963 and 2000 do not include lands irrigated by diversions from the Reese River and the Little Humboldt River because they are not included in the Bartlett Decree; water use in the Little Humboldt River Basin is governed by the Little Humboldt Decree (Mashburn and others, 1935), and the Upper Reese River has not been adjudicated. Additionally, both rivers are rarely tributary to the Humboldt River, having no substantial effect on Humboldt River streamflow except on years with extraordinary runoff.

Before 1910, the Humboldt River followed a river course between Dunphy and Battle Mountain that was 1–3 mi north of its present channel. The river moved to its present (as of 2016) channel because of flooding during February and March 1910 (Foster, 1933). After 1910, the present channel was well-defined, except for a marshy area near Argenta (fig. 2) that was called the “Argenta Marsh.” In this area, the river channel was shallow and widened into an area of marshes and wetlands. Estimates of the total area of the Argenta Marsh ranged from less than 1,000 acres (Malone, 1932) to 12,000 acres and then to 15,000 acres (Horton, 2000a). The river exited the Argenta Marsh in two channels that joined a few miles east of Battle Mountain (Foster, 1933). The effect of the Argenta Marsh was that it impeded the streamflow of the river, especially in the spring, and delayed the arrival of the snowmelt runoff to the Lovelock area. Streamflow losses in this reach of the river were estimated to range from 4,000 to 12,000 acre-ft/yr because the river had to provide enough water to saturate the shallow water table in the marsh before streamflow could continue downstream (Malone, 1932). In the 1930s, irrigators near Lovelock, Nevada, and the Bureau of Reclamation considered draining the marshes, but a functioning drainage ditch was not completed until the early 1950s. As of 2020, the ditch remains visible from Interstate–80 as a straight section of river channel north of the highway and immediately west of the Argenta railroad siding.

There are 10 reservoirs that store water in the HRB: (1) 3 reservoirs are used to store groundwater pumped for mine-dewatering purposes and (2) 7 reservoirs impound streamflow for irrigation, recreation, and flood control purposes. The MHRB study area encompasses 5 of the 10 reservoirs in the HRB (fig. 2). The Maggie Creek Reservoir, T-S Ranch Reservoir, and Lone Tree cooling ponds (no longer active as of 2016) store excess groundwater pumped from the Carlin, Goldstrike, and Lone Tree mine operations, respectively (fig. 2). Water from Maggie Creek Reservoir was released to the Humboldt River by way of Maggie Creek since April 1994, with total annual volumes ranging from about 8,000 to 19,000 acre-ft. Water from the T-S Ranch Reservoir was released to the Humboldt River by way of a lined canal and pipeline at annual volumes ranging from about 8,000 to more than 48,000 acre-ft. Water from the Lone Tree cooling ponds was released to the Humboldt River by way of a lined ditch, Iron Point Relief Canal, and Herrin Slough since June 1992, at total annual volumes ranging from about 13,000 to more than 45,000 acre-ft.

Of the seven non-mine related reservoirs, four impound streamflow on tributaries to the Humboldt River and three impound water along or near the main stem of the Humboldt River (fig. 2). The four reservoirs on tributaries are (1) Bishop Creek Reservoir (east of the MHRB and not shown), with a potential capacity of 30,000 acre-ft and constructed in 1912, which as of 2016 was not used because of structural problems; (2) South Fork Reservoir (east of the MHRB study area), with a capacity of 42,000 acre-ft and constructed in 1987, which is used for recreation and fisheries purposes; (3) Willow Creek Reservoir, with a capacity of 18,000 acre-ft and constructed during 1910–25, which is used for irrigation; and (4) Chimney Reservoir, with a capacity of 35,000 acre-ft and constructed in 1974, which is used for irrigation and recreation on the Little Humboldt River. The three reservoirs on or near the Humboldt River are downstream of the Imlay gage and MHRB and are used for irrigation in the Lovelock area. The Upper and Lower Pitt–Taylor Reservoirs have a combined capacity of 46,400 acre-ft and were constructed during 1907–11 (Rush and Rice, 1972). The two reservoirs are filled by way of the Pitt–Taylor Canal, and streamflow diversion from the Humboldt River to the Pitt–Taylor Canal is upstream from the Imlay gage (fig. 2). The Rye Patch Reservoir is the only impoundment on the main stem of the Humboldt River, and the reservoir also is used for recreation. Rye Patch Reservoir was constructed during 1935–36 and has a storage capacity of 194,300 acre-ft. From 1935 to 1936, the Pershing County Water Conservation District acquired lands and appurtenant water rights from the Battle Mountain area for filling of Rye Patch Reservoir and providing an increased supply of irrigation water to Lovelock Valley (U.S. Department of Agriculture, Nevada Cooperative Survey, 1965). Total usable impoundment capacity in the HRB is about 326,000 acre-ft; however, this capacity is achieved only during the infrequent periods of greater precipitation and runoff in the HRB.

Groundwater is pumped in the HRB to meet the demands for five principal uses: (1) municipal (production) and domestic, (2) power generation, (3) irrigation, (4) mining, and (5) stock (Prudic and others, 2006). The development history of groundwater resources in the HRB from the late 1800s to the 1940s is not known because few records were maintained. Census records starting in the late 1800s, and driller’s logs and groundwater permits starting in the early 1900s from the NDWR (State of Nevada Division of Water Resources, 2020b, 2020c; U.S. Census Bureau, 2021), provide some of the earliest data for estimating groundwater use in the HRB.

Based on decadal census data, populations in the counties in the HRB were nearly static from 1940 to 1960 (Prudic and others, 2006). Because of little population growth, few changes in groundwater development were necessary during that time. Populations in all counties in the HRB increased slowly from 1960 to 1980. Populations increased dramatically after 1980 because of the development of large, low-grade gold deposits in the area and the resultant employment opportunities.

Groundwater was used for mining purposes throughout much of the 1900s; however, this use was relatively minor compared to other uses during most of that century. Until the 1980s, groundwater for mining was used mostly for milling purposes and dust control (Prudic and others, 2006). By the 1980s, mining companies began developing mine operations for the large, low-grade gold deposits in northern Nevada (Prudic and others, 2006). Starting in the early 1990s, groundwater pumping began increasing dramatically because of the need to dewater some of the surface and underground mines as they expanded operations below the water table. Annual pumping increased from 10,000 acre-ft in 1991, to more than 120,000 acre-ft in 1992, to a high of nearly 264,000 acre-ft in 1998 (Prudic and others, 2006). Groundwater use for mining related purposes has remained substantial since 1998, with variations in pumping whenever mine operations complete pit excavations and allow pit lakes to form (if the water table is above bottom of the pit) or when new mines or dewatering operations begin (Jon Benedict, Nevada Division of Water Resources, written commun., 2020).

The number of wells drilled for municipal and mining use declined from 1998 to 2003, which corresponded to a decline in the number of new gold mining operations and declines in population for Elko, Humboldt, and Lander counties during that time (U.S. Census Bureau, 2021; State of Nevada Division of Water Resources, 2020b). The decline in new gold mining operations also corresponded to a decline in the number of wells drilled for domestic, monitoring, and test purposes during that period. The populations of Elko and Humboldt counties rebounded to exceed or approach previous high levels by the late 1990s, although Lander County population level remains less than the 1998 peak. Overall, the number of wells drilled for municipal and domestic use was relatively stable since around 2003 (State of Nevada Division of Water Resources, 2020c).

Groundwater was used for irrigation purposes since the late 1940 and 1950s but was a relatively minor use until the early 1960s (Prudic and others, 2006). From the early 1960s through the early 1980s groundwater use for irrigation rapidly increased (Prudic and others, 2006). From 1983 through the mid-1990s, there was a slight reduction in groundwater used for irrigation, but use increased again through the early 2000s and remained steady through the mid-2000s (Prudic and others, 2006; Davis and others, 2026).

Carbonate rocks form a large groundwater flow system of regional extent in eastern Nevada and western Utah (Harrill and Prudic, 1998). This area, referred to as the “carbonate-rock province of the eastern Great Basin,” is underlain by as much as 30,000 ft of limestone, dolomite, and interbedded sandstone and shale of Middle Cambrian to Lower Triassic age (Harrill and Prudic, 1998). The approximate western boundary of the carbonate-rock province crosses the model area from southwest to northeast, so carbonate rocks underlie the eastern third of the MHRB.

In the northern Boulder Flat HA and the Maggie Creek Area HA, carbonate rocks are the primary hydrogeologic unit dewatered by groundwater pumping operations at Goldstrike mine, Carlin north mine, and Carlin south mine operations. In this part of the study area, carbonate rocks are extensively faulted and fractured and have secondary porosity because of dissolution by groundwater. Near the mines, the hydraulic conductivity of carbonate rocks can be as much as 100 feet per day (ft/d; Prudic, 2007). Regionally, however, the hydraulic conductivity of the carbonate units is probably about 3–10 ft/d (Prudic, 2007).

Clastic sedimentary rocks of marine origin, categorized as clastic rocks for this study, are largely composed of siltstones and are the predominant rock types in mountain ranges of the model area. Clastic rocks are either exposed over large areas (fig. 7) or overlain by varying thicknesses of volcanic rocks and basin-fill deposits. Where underlying younger rocks and deposits north of the Humboldt River, clastic rocks are at depths of 3,000 ft or less (Plume and Ponce, 1999). South of the Humboldt River, depths to clastic rocks range from 1,000 ft to more than 6,000 ft in several basins (Plume and Ponce, 1999). Although clastic rocks are extensively fractured, they are not as permeable as carbonate rocks because they are not as susceptible to solution widening. Over large areas, the hydraulic conductivity of these rocks is about 0.01 ft/d (Prudic, 2007). The low permeability of the clastic rocks is evident by the presence of perennial streams and significant drainage development in watersheds where these are the predominant rock type (fig. 7). In contrast, the watersheds in the MHRB underlain by carbonate rocks seldom have perennial streams, and drainage development is immature.

Crystalline rocks of Jurassic and Tertiary age include granitic and metamorphic rocks and are exposed at scattered locations in the model area (fig. 7). Crystalline rocks generally impede the movement of groundwater because they have low permeability, extend to substantial depths, and are more spatially extensive at depth than their outcrops. Crystalline rocks on the south side of the Goldstrike mine operation in the southern Tuscarora Mountains were informally named the “Goldstrike intrusive,” and they function as a barrier to groundwater flow (Plume, 2009). The hydraulic conductivity of the crystalline rock generally ranges from 0.0001 to 5 ft/d, with higher values occurring where the rock is fractured (Plume and Ponce, 1999).

Volcanic rocks are exposed extensively in mountain ranges in the model area and can be interbedded with older basin-fill deposits (fig. 7). In areas north of the Humboldt River, volcanic rocks consist of relatively permeable basalt flows and related rocks that overlie less permeable tuffaceous volcanic rocks. Thicknesses of these rocks range from less than 500 ft to more than 3,000 ft (Plume and Ponce, 1999). Numerous springs approximately define the contact between the two types of volcanic rocks in the Snowstorm Mountains in the northern part of the study area (fig. 7). Perennial streams are rare in areas above the contact but are common in areas along and below the contact. Volcanic rocks in areas south of the Humboldt River are dominantly of tuffaceous composition and generally are thicker over larger areas than in northern parts of the model area (Plume and Ponce, 1999). Like the clastic siltstone rocks, the presence of perennial streams and riparian vegetation in watersheds underlain by tuffaceous volcanic rocks indicate that these rocks are relatively impermeable. Groundwater occurs in fractures and interflow zones in volcanic rocks. The hydraulic conductivity of the volcanic rock units generally is between 0.5 and 25 ft/d (Prudic, 2007) but can range between 3x10⁻⁷ and 1,300 ft/d for various types of volcanic rock (table 5).

Deposits of the basin-fill aquifer system, composed of unconsolidated and semi-consolidated sediments, form the most extensive aquifers in the MHRB (fig. 7). Combined with underlying or interbedded volcanic rocks, total thicknesses of the deposits range from less than 500 ft to about 3,000 ft west of the Tuscarora Mountains to as much as 4,000–8,000 ft east of the mountains (Plume, 1995; Plume and Ponce, 1999). In basins south of the Humboldt River, combined thicknesses of basin-fill deposits and volcanic rocks are as much as 9,000 ft in Crescent and Antelope Valleys and as much as 18,000 ft in Pine Valley (Plume and Ponce, 1999).

Basin-fill aquifer deposits were divided into older semi-consolidated and younger unconsolidated basin-fill units because the two differ lithologically and, consequently, have differing ranges in hydraulic properties (table 5; Plume and Ponce, 1999; Maurer and others, 2004). Older basin-fill deposits consist of semi-consolidated clay, silt, sand, gravel, and boulders that range from well sorted to unsorted, depending on the depositional environment, which includes lacustrine (lake) and fluvial (stream) environments. In addition, older basin-fill deposits commonly contain interbedded volcanic rocks. Although the older basin-fill deposits can be permeable, the permeability generally is less than the overlying younger basin-fill deposits due to partial cementation of its matrix.

The younger basin-fill aquifer, which consists of alluvial slope, valley floor, and fluvial deposits, generally is less than 500 ft thick and consists of surficial deposits of poorly sorted and unconsolidated clay, silt, sand, and gravel. Alluvial slope deposits, also referred to as “alluvial fans” or “piedmont slopes,” have lithologies ranging from sands to gravels and cobbles (Maurer and others, 2004). Generally, alluvial slope deposits are poorly sorted and grade from coarse-grained near the top of the alluvial slope to fine-grained near the bottom of the slope where the slope merges with the valley floor. Valley floor deposits generally consist of interbedded layers of clays, silts, and sands that are relatively permeable in a horizontal direction and much less permeable in a vertical direction.

The width of the Humboldt River flood plain ranges from less than 1,000 ft in the eastern third of the study area, to 6 mi in Boulder Flat, and to 4 mi west of Battle Mountain. Well-sorted and highly permeable gravels underlie the Humboldt River flood plain along the length of the river in the study area and constitute the fluvial deposits in the MHRB (Bredehoft, 1963; Cohen, 1964c); they generally coincide with the modern-day meanders of the Humboldt River within the flood plain. In the eastern part of the study area, the thickness of the gravels probably does not exceed 100 ft. In the Battle Mountain area, the gravels are as much as 300 ft in the area 20 mi east of the town and abruptly thin to 30–40 ft thick west of the town (Bredehoft, 1963). In the western part of the study area, the thicknesses of the gravels are largely unknown and variable and could be as much as 100 ft thick (Cohen, 1964c). High permeability of the gravel unit can be inferred from streamflow and stilling well (Sauer and Turnipseed, 2010) data at the Imlay gage when flows in the Humboldt River are observed after a prolonged period of no flow. On February 6, 2016, the water level in the stilling well started to rise 1 hour and 15 minutes before streamflow arrived (U.S. Geological Survey, 2021; fig. 8). The stilling well is open at the bottom and completed in the gravel unit at a depth of about 1.3 ft below the streambed bottom. Highly permeable sediments beneath the Humboldt River are necessary for this relationship between discharge and stilling well water levels where increases in water levels in wells are observed before stream flow increases are measured at the same location.

The presence of a persistent gravel unit at depth also is evident in Paradise Valley (Bredehoft, 1963). Bredehoft (1963) reported that the gravel unit is thickest at almost 100 ft at the northeast end of Paradise Valley where Martin Creek enters the valley. The gravel unit generally is thickest along a north-south axis of the valley coinciding with the stream traces of Martin Creek and Little Humboldt River and thins toward the valley margins as well as to the south. The gravels are underlain by blue clay west of Battle Mountain and yellow, silty and sandy clay to the east (Bredehoft, 1963).

The blue clay underlying much of the fluvial deposits along the middle reach of the Humboldt River near Battle Mountain is a persistent lacustrine clay unit (figs. 7, 9). The blue clay was originally recognized by Phoenix (1949) through test drilling near the Argenta marsh area near Argenta, Nevada, and later described in more detail by Bredehoft (1963). Bredehoft (1963), however, did not define the extent and thickness of this unit. The extent and thickness of the blue-clay unit in the model area was determined by inspecting well drillers’ reports (State of Nevada Division of Water Resources, 2020c) for descriptions of blue clay. The top and bottom depths below land surface of blue clay or blue-clay lenses were mapped if the clay thickness was greater than or equal to 10 ft. The boundary of the blue clay (fig. 9) was estimated by manually tracing the extent of wells with blue clay listed in the lithology and by considering the topography and extent of pre-Pleistocene lakes (Reheis, 1999). Depths from each well were linearly interpolated using geographic information system software (ArcMap; Esri, 2011) to determine the depth to the top and bottom of the blue-clay layer. A minimum thickness of 10 ft was enforced, resulting in blue-clay thickness ranging from 10 to about 150 ft (fig. 9). Hydrologically, the blue clay is like the playa unit (table 5) and is relatively impermeable (Bredehoft, 1963). Estimated extents and thicknesses of the blue-clay layer are uncertain in the areas with minimal data (fig. 9).

The proportions of older and younger basin fill in most basins are difficult to determine because the older fill is often concealed by the younger fill, and the two cannot be easily distinguished at depth with well driller’s logs; however, for this study, an attempt was made to differentiate older from younger basin fill by recording changes in lithology noted in drillers’ logs among finer clays, sands, gravels, and alluvial deposits and the coarser sand, gravel, boulders, or rock implying older basin fill. The description of the contact between older and younger basin fill was rated as poor, fair, or good, and the depths corresponding to fair and good ratings were used for interpolation and mapping. Older basin fill is exposed throughout Pine Valley, and it constitutes, by far, most of the total thickness of basin fill in that valley. If Pine Valley is used as a general model for other basins, then older basin fill likely constitutes most of the total thickness of fill in the other basins of the model area. This assumption is reasonable because older fill was accumulating from as long as 34 million years ago to as recently as 6 million years ago, whereas younger basin fill has been accumulating mostly during the past several million years (Stewart, 1980). In the northeastern part of the study area, the total composite thickness of older basin-fill deposits is about 7,600 ft (Maurer and others, 1996); therefore, older basin-fill deposits likely constitute most of the total basin-fill thickness in most basins, and the thickness of younger basin-fill deposits probably does not exceed 500 ft.

Geologic structural features that potentially affect groundwater movement in the MHRB include horsts and grabens, rift zones, and faults. Horst (uplifted) and graben (down-dropped) features are large-scale structural blocks formed from extensional tectonics and faulting that have created the characteristic basin and range landforms across Nevada. Rift zones are sets of rifts, which are linear geologic features also formed from extensional tectonics, where extensive fissures create conduits associated with volcanic activity. A fracture or zone of fractures that has caused movement between blocks of rock (faults) not only displace geologic material but also can have significant hydrogeologic effects. Faults can act as barriers to groundwater flow when (1) the material in the fault plane itself is mechanically or chemically altered to reduce the permeability of the zone of faulting or (2) the movement along the fault juxtaposes a permeable hydrogeologic unit against a low-permeability unit. Faults also can act as conduits of groundwater flow along the fault planes by causing mechanically and chemically induced open spaces along fracture zones. Alternatively, faults may have no measurable effect on groundwater flow (Maurer and others, 2004).

The model area contained several geologic structural features (fig. 7). Horsts and grabens were represented in the model as abrupt lateral changes in hydrogeologic units and were generally orientated southwest to northeast. Geologic rifts extend from north or northwest to south or southeast across the model area, and several regional-scale rifts form the northern Nevada rift zone (fig. 7; Zoback and others, 1994; Damar, 2018). The northern Nevada rift is defined by prominent linear aeromagnetic anomalies, which indicate the presence of highly magnetic volcanic basaltic rocks along a 2–3-mi-wide zone and extending to depths of 6 mi or more (Zoback and others, 1994). Hydrologic effects of the northern Nevada rifts are not well understood because groundwater-level data are not readily available for hydrogeologic units on either side of the rifts; however, the rifts could impede the northeast to southwest flow of groundwater from the Sheep Creek Range and Snowstorm Mountains to the Clovers Area (HA 064) and Kelley Creek Area (HA 066) HAs and possibly across the western part of Pine Valley (Maurer and others, 1996; fig. 2).

Normal faults are common basin and range related structural features in the MHRB and are generally orientated from southwest to northeast. The effects of faults on groundwater flow in the MHRB are variable and largely unknown. Faults may function as conduits for groundwater flow, although evidence of this in the MHRB is sparse; some faults in the MHRB may act as barriers to flow, but others may not affect groundwater flow. Four faults in the MHRB, located near large gold mines, were assumed to act as barriers to groundwater flow because of large differences in water levels across the faults. These four faults include the Post and Siphon faults near the Carlin north and Goldstrike mine operations and the Cortez and Crescent faults near the Cortez Mine Complex (fig. 7). Fault extents and locations in the study area were derived from previous studies by Plume (2005) and Itasca Denver, Inc. (2016).

The Post and Siphon faults cause abrupt changes in properties across hydrogeologic unit boundaries. The Post fault juxtaposes carbonate rocks and clastic rocks south of the Carlin north mine operations, and the Siphon fault juxtaposes volcanic rocks, which are concealed by younger basin-fill deposits, and the same block of carbonate rocks south of the Carlin north and Goldstrike mine operations (fig. 7). Plume (2005) mapped water-level declines resulting from mine-dewatering operations for 1991–2003 near the Post and Siphon faults; however, water-level increases were observed to the southwest from 1993 to 2003. Although the clastic rocks (siltstones) to the northeast of the Post fault are extensively fractured, they are not as permeable as the carbonate rocks they are juxtaposed with because they are not susceptible to solution widening (Plume, 2005). Evidence of the differing permeability of the units is shown by water-level variation in dewatered carbonate rocks near the Goldstrike mine operations in the Tuscarora Mountains (fig. 7). Water levels in siltstones are perched above water levels in the underlying dewatered carbonate rocks (Paul Pettit, Newmont Mining Corp., written commun., 2004), and in 2004, water levels differed by more than 1,000 ft across the Post fault (Plume, 2005). Before mining activities, the difference in water levels across the Siphon fault was about 500 ft, with water levels in the carbonate rocks above water levels in the volcanic rocks. The difference in water levels across the Siphon fault in 2004 was about 1,100 ft, with water levels in the carbonate rocks below water levels the volcanic rocks (J. Zhan, Barrick Management Corp., written commun., 2004). These large differences in water levels among the clastic, volcanic, and carbonate units indicate the absence of hydraulic connection or an extreme contrast in hydraulic properties among aquifer units spanning across the Post and Siphon faults (Plume, 2005).

The Crescent fault and the Cortez fault also affect groundwater flow in the study area. The Crescent fault (fig. 7), on the north side of the Cortez Mountains, trends northeast to southwest and is estimated to have as much as 10,000 ft of vertical displacement (Gilluly and Masursky, 1965), creating a deep bedrock basin in Crescent Valley. The juxtaposition of bedrock and basin fill, and any gouge zones (pulverized rock) created by this fault, generally act as barriers to flow, as evidenced by discontinuities in water levels on either side of the fault. The Cortez fault trends north-northwest to south-southeast at the southern end of the Cortez Mountains (fig. 7). At least 3,800 ft of vertical displacement along the Cortez fault have created sequences of gouge between the hanging wall and footwall and impedes flow across the fault (Itasca Denver, Inc., 2016). Other faults that are approximately normal to the Cortez fault intersect it in various places and may create permeable pathways that facilitate groundwater flow across the Cortez fault. Collectively, these numerous faults have led to variable groundwater levels and compartmentalization of the groundwater flow system (Itasca Denver, Inc., 2016).

Numerous other faults exist in the MHRB that may affect groundwater flow in similar ways as the faults previously discussed. However, hydrologic data are unavailable to determine the effects of all faults on groundwater flow. The faults previously discussed were identified only after wells were installed as part of hydrologic monitoring programs around large mines. The potential effects of other faults on groundwater flow are difficult to predict without measuring water levels on both sides of a fault or measuring water-level responses to stresses occurring across the fault.

The Humboldt River in the MHRB is about 262 mi long (river valley length) from the Carlin gage on the east to the Imlay gage between Winnemucca and Rye Patch Reservoir on the west. The mean annual discharge of the Humboldt River at the Carlin, Palisade, Battle Mountain, Comus, and Imlay gages for the available period of record through WY 2016 for each station was about 258,000; 282,000; 258,500; 237,100; and 191,300 acre-ft/yr (356, 389, 356, 327, and 264 ft³/s), respectively (table 4). Long-term (WYs 1960–2015) mean monthly streamflow characteristics of the Humboldt River at the Carlin, Palisade, Battle Mountain, Comus, and Imlay gages are shown as bar graphs on figure 10 for three periods of record: (1) WYs 1960–70 before large-scale development of groundwater for irrigation (fig. 10A); (2) WYs 1971–93, during which development of groundwater for irrigation started increasing substantially (fig. 10B); and (3) WYs 1994–2015 after mine dewatering began (fig. 10C). All graphs indicate that the months of lowest flows are August through December when runoff and evapotranspiration are at annual minimum rates. The onset of cooler temperatures and the resulting decrease in evapotranspiration, generally in late October, results in a slight increase in flows in November and December, although runoff remains minimal. Flows then begin a steady increase in January and continue through July because of winter storms (December–February), low-altitude snowmelt in some years (February–March), and high-altitude snowmelt in most years (April–June). After June, flows rapidly decrease to late summer minimums when stream diversions are occurring and annual evapotranspiration is at maximum rates.

Month-to-month variations in streamflow are mostly consistent among the five streamgages and the three periods (fig. 10). The lowest monthly flows were during WYs 1960–70 (fig. 10), when annual precipitation was below normal more often than it was above normal (fig. 4). Flows at the five stations were consistently higher, especially in March–May, in the WYs 1971–93 period than the WYs 1960–70 and 1994–2015 periods. Part of the reason for this increase is the effect of the extremely wet years in WYs 1982–84 (fig. 4). The Carlin, Palisade, Battle Mountain, and Comus gages had similar relative monthly flows during the WYs 1960–70 and 1971–93 periods, with Battle Mountain streamflows generally less than Carlin or Palisade streamflows; however, in WYs 1994–2015, some monthly flows at the Battle Mountain gage were close to or exceeded flows at the other stations during spring and summer months of March through July (fig. 10) because mines were discharging water to the Humboldt River or Maggie Creek at various times during the period.

Annual streamflow in the Humboldt River measured at the Palisade gage for WYs 1903–06 and 1912–2016 (no data for WYs 1907–11) varied from about 25,200 (in WY 1934) to about 1,340,400 acre-ft/yr (in WY 1984; fig. 11). The mean annual streamflow from WYs 1903 to 1906 and 1912 to 2016 at the Palisade gage was about 282,000 acre-ft/yr. Before large-scale development of irrigation pumping in the 1960s and mine pumping in the 1990s, the mean annual streamflow from WYs 1903 to 1906 and 1912 to 1958 was about 264,200 acre-ft/yr. From WYs 1959 to 2016, the mean annual streamflow was about 297,600 acre-ft/yr, an increase of about 13 percent partly because of increased precipitation and partly from discharge of water from mine-dewatering operations to the Humboldt that started in the 1990s. Mean annual streamflow from WYs 1946 to 1958 was about 265,300 acre-ft/yr and representative of the pre-development conditions of streamflow at the Palisade gage. The period of WYs 1946–58 was selected to represent pre-development streamflow conditions for the Humboldt River because (1) this period was before major groundwater development in the MHRB; (2) at the Palisade gage, mean daily streamflow for this period closely matched mean daily streamflow for the period of record before WY 1961; and (3) most of the other streamgages in the MHRB had daily streamflow measurements for this period.

Between the Carlin and Palisade gages (fig. 2), the Humboldt River gains streamflow from Susie, Maggie, and Marys Creeks and from groundwater discharge into the river channel (Plume, 1995; Maurer and others, 1996). Mean annual streamflow gains along this reach of the river have been estimated at 38,000 acre-ft/yr (52.5 ft³/s; Plume, 1995). This reach of the river has always had base flow, even during extended droughts, because of groundwater discharge (Maurer and others, 1996). From WYs 1951 to 1958, the Carlin to Palisade reach gained about 9,900 acre-ft/yr (12.4 ft³/s; table 6) which was less than the estimated gains by Plume (1995) because of the effects of extended drought from WYs 1951 to 1962 (fig. 4).

Downstream from the Palisade gage to the Imlay gage, Humboldt River discharge generally decreases, even with stream inflows from Pine Creek and Rock Creek in years of above mean runoff, and rarely, with streamflow contributions from the Reese River and Little Humboldt River. The decreased streamflow discharge for the Humboldt River is from numerous irrigation diversions and from infiltration of streamflow to the shallow gravels that underlie the Humboldt River flood plain where groundwater evapotranspiration is substantial below the Dunphy gage. Mean annual streamflow decreases from diversions and infiltration losses on Humboldt River between Palisade and Battle Mountain, Battle Mountain and Comus, and Comus and Imlay during available periods of record were about 20,000, 24,000, and 46,000 acre-ft/yr (about 27, 34, and 63 ft³/s), respectively. Humboldt River streamflow gains and losses in the MHRB during WYs 1951–58 and WYs 1959–62 also were recorded and analyzed for reaches shorter than those described earlier (Cohen, 1963b). Mean annual streamflow losses for WYs 1951–58 in the four river reaches between Palisade and Comus were Palisade to Argenta, 46,000 acre-ft/yr (63.5 ft³/s); Argenta to Battle Mountain, 34,800 acre-ft/yr (48.0 ft³/s); Battle Mountain to Valmy, 19,000 acre-ft/yr (26.2 ft³/s); and Valmy to Comus, 9,000 acre-ft/yr (12.4 ft³/s; table 6).

Several tributaries contribute streamflow to the Humboldt River in most years. Major tributaries for this study were defined as those that are perennial in most years and had continuous streamgage records for most years. Minor tributaries for this study were defined as those that do not consistently contribute streamflow to the Humboldt River nor have long-term continuous streamgage records. The major and minor designation for tributaries in this report primarily was used to differentiate between gaged and ungaged streams in the MHRB to aid in the conceptualization of streams in the model.

Gumboot Lake is an ephemeral surface-water feature that forms during times of high runoff in the lower part of Paradise Valley (Harrill and Moore, 1970; Prudic and Herman, 1996; fig. 12). Water becomes impounded behind sand dunes south of the confluence of Big Cottonwood Creek, Martin Creek, the Little Humboldt River, and other minor tributaries to create Gumboot Lake. When Gumboot Lake forms, it infiltrates and recharges the valley floor sediments in the area. Gumboot Lake remains until the water seeps into the ground, evaporates, or a channel is dredged through the sand dunes to allow flow in Little Humboldt River to continue to the Humboldt River (Prudic and Herman, 1996). The first documented report of the lake was in 1875, and additional lake formations were recorded in 1890, 1907, 1910, 1914, 1953, 1958, 1969, 1983, and 1984 (Harrill and Moore, 1970; Prudic and Herman, 1996). After 1984, the formation of Gumboot Lake was not documented, but the lake was known to continue to form during years with substantial runoff. Dredging operations of the sand dunes were done south of the lake, which allowed the lake to drain, and were documented in 1953, 1958, and 1969 (Harrill and Moore, 1970). In 1983, and almost certainly in 1984, when Gumboot Lake filled to its largest recorded depth and extent, sand dunes below the lake likely were dredged; however, documentation confirming dredging operations in 1983 or 1984 were not found. On average, Gumboot Lake forms every 5 years (Prudic and Herman, 1996), with large impoundments forming about every 15 years (Harrill and Moore, 1970).

A geospatial information systems analysis was used to estimate the Gumboot Lake surface altitude and lake area for three approximate lake sizes: 2,000 acres; 5,000 acres; and 10,000 acres (fig. 12; Medina, 2021), although only the 2,000 acre and 10,000-acre lake sizes were used in this study. At its likely historical maximum extent, Gumboot Lake had an area of as much as 10,000 acres and a depth of 4–5 ft (Harrill and Moore, 1970), with a lake surface altitude of about 4,322 ft. For an area of 5,000 acres, Gumboot Lake has a depth of about 2–3 ft, with a lake surface altitude of around 4,320 ft. For an area of 2,000 acres, the lake has a depth of about 0.5–2 ft, with a lake surface altitude of 4,317.5 ft. Drainage from the lake discharged to the Little Humboldt River and Humboldt River generally during the months of April, May, and June (Harrill and Moore, 1970).

Statutes regulating diversion and use of surface water in Nevada have existed since 1866 (Welden, 2003). After the early 1930s, court decrees established diversion locations, rates, and priority dates that were enforced and managed by the Nevada Office of the State Engineer (Malone, 1932). The Edwards Decree and the Bartlett Decree govern the use of Humboldt River and most tributary waters for irrigation, and together with other later rulings, collectively are referred to as the “Humboldt Decree” (State of Nevada Office of the State Engineer, 2021). The use of surface water from streams contributing to the Little Humboldt River for irrigation primarily in Paradise Valley are governed by the Little Humboldt Decree (Mashburn and others, 1935). The Little Humboldt Decree was interpreted into distribution tables by water commissioners with the NDWR (Steve Del Soldato, Nevada Division of Water Resources, written commun., 2015).

The Humboldt Decree (Mashburn and Mathews, 1943) governs deliveries at points of diversion (PODs) within the MHRB during the irrigation season (annually between March 15 and September 15). The distribution of water for the Humboldt River and its tributaries, regulated by the Humboldt Decree, was interpreted into distribution tables by Hennen (1964b). Diversion deliveries were determined by priority date, ranging from 1861 through 1921, based on streamflow at the Palisade gage for three irrigation subseasons: (1) early (March 15 through April 27; 44 days); (2) middle (April 28 through June 13; 47 days); and (3) late (June 14 through September 15; 90 days). The irrigation subseasons were determined based on three land classes: (1) diversified pasture; (2) meadow pasture; and (3) harvest crop. Each of these land classes can be irrigated during specific irrigation subseasons (State of Nevada Office of the State Engineer, 2021). Based on decreed water rights, diversified pasture can only be irrigated during the early irrigation subseason, with an annual volume of 0.75 acre-foot per acre (acre-ft/acre) of land; meadow pasture can be irrigated during the early and middle irrigation subseasons, with an annual volume of 1.5 acre-ft/acre of land; and harvest crop can be irrigated during the early, middle, and late irrigation subseasons with an annual volume of 3.0 acre-ft/acre of land (Mashburn and Mathews, 1943).

Water diverted in the MHRB based on the Humboldt River Decree and applied to fields for irrigation generally is used by evapotranspiration (used by crops via transpiration or evaporates), or it infiltrates below the crop root zone to shallow aquifers. In this study, the proportion of diverted water that is used by evapotranspiration is defined as the “consumptive part,” whereas the proportion that infiltrates to shallow aquifers is defined as the “nonconsumptive part.” The nonconsumptive part was assumed to return to the Humboldt River via riverbed seepage from the shallow aquifers underlying irrigated areas along the river. Consumptive and nonconsumptive percentages were estimated for diversions along the Humboldt River (Steve Del Soldato, Nevada Division of Water Resources, written commun., 2015; Davis and others, 2026). Nonconsumptive return-flow estimates varied based on the priority date at each POD and ranged from 0 to 95 percent of the total amount of water diverted. Lower return-flow percentages were for earlier priority dates and corresponded to periods of lower streamflow, generally during drier conditions. Higher return-flow percentages were associated with later priority dates during periods of greater streamflow, during which irrigated lands are likely to be fully or close to fully saturated.

In this study, bifurcations are defined as types of regulated or unregulated diversions that route streamflow from the main Humboldt River channel into sloughs or parallel channels at rates that are not specified in the Humboldt Decree. Bifurcations in the study area include White House, Rock Creek, Iron Point, CS-Slough, and Hay (fig. 2). The Iron Point bifurcation near the Lone Tree mine operation is not mentioned in the Edwards and Bartlett Decrees but conveys water to the Iron Point Relief Canal. The canal also received water discharged from dewatering operations at the Lone Tree mine operation during active mining in the 1990s through the mid-2000s (Hess, 2002). Water from the canal flows to the Herrin Slough and then discharges to the Humboldt River upstream from the Comus gage (fig. 2).

In addition to bifurcations and diversions for mining and irrigation, Humboldt River waters also are diverted for flood management at the Pitt–Taylor Diversion east of the Rye Patch dam and upstream from the Rye Patch Reservoir (fig. 2). The Pershing County Water Conservation District manages the Pitt–Taylor Diversion that redirects surface water from the Humboldt River through the Pitt–Taylor Diversion Canal to the Upper and Lower Pitt–Taylor Reservoirs, which serve as storage ponds (fig. 2). The Pitt–Taylor Diversion Canal and storage Reservoirs also are used during periods of high flow in the Humboldt River to prevent exceeding the capacity of Rye Patch Reservoir and to prevent flooding of farmland between the Pitt–Taylor Diversion and Rye Patch Reservoir (Flying M Ranch, not shown). The Pitt–Taylor Diversion is not used every year; however, available data from Rush and Rice (1972) indicate it was used for the 30 years after its construction in November 1912, except for 1935, the year when the Rye Patch dam was constructed. Records of the annual volume of water diverted from the Humboldt River at the Pitt–Taylor Diversion to the Pitt–Taylor Reservoirs are available for 1915–20 and 1922–71 (Rush and Rice, 1972), daily streamflow in the diversion canal are available for WYs 1946–76 (USGS gage 10332500 HLIL And P CO Feeder Canal near Imlay, Nev.; U.S. Geological Survey, 2021), and daily water-storage data in Rye Patch Reservoir are available from 1936 to 2015 (NV052 10334500 Rye Patch Reservoir near Rye Patch, Nev.; U.S. Geological Survey, 2021). The Pitt–Taylor Diversion likely was used during times without records; therefore, monthly mean diversion rates for years without records were estimated based on measured annual and daily Pitt–Taylor Diversion Canal discharge and daily storage in Rye Patch Reservoir. Estimated diversion rates at the Pitt–Taylor Diversion are available in the accompanying data release (Davis and others, 2026). The estimated mean annual discharge for WYs 1961–2015 was about 17,600 acre-ft/yr (24.3 ft³/s). The maximum annual mean diversion was estimated at about 89,000 acre-ft/yr (123.4 ft³/s) in WY 1983 (Davis and others, 2026).

The Little Humboldt Decree (Mashburn and others, 1935) governs deliveries of streamflow in the Little Humboldt River and its tributaries, including Martin Creek, to PODs in Paradise Valley. During the irrigation season, almost all the streamflow reaching Paradise Valley is diverted for irrigation (Loeltz and others, 1949). Martin Creek priority dates range from 1864 to 1904, and Little Humboldt River priority dates range from 1864 to 1892. In most years, available streamflow in Martin Creek and Little Humboldt River is used for surface-water irrigation; however, in some years, excess water is available and reaches the confluence with the Humboldt River (Steve Del Soldato, Nevada Division of Water Resources, written commun., 2018). In Paradise Valley, priorities for each farm are determined based on streamflow on the Little Humboldt River and tributaries using three land classifications: A (harvest crops), B (meadow pasture), and C (diversified pasture). The irrigation season in Paradise Valley spans from April 1 to September 27 within each WY. During irrigation subseasons, crops A, B, and C can be irrigated from April 1 to April 30 (subseason A, B, C), crops A and B can be irrigated from May 1 to June 29 (subseason A, B), and crop A can be irrigated from June 30 to September 27 (subseason A). During an irrigation subseason, priorities for A and B crops must be met before C crops are served. Estimates for percentages of consumptive and nonconsumptive use are not stated for Paradise Valley in the Little Humboldt Decree (Mashburn and others, 1935).

The MHRB includes mines with significant dewatering operations (fig. 2). Dewatering reduces water levels near mining operations by withdrawing substantial amounts of groundwater using a network of wells. Some of the groundwater from dewatering is consumed by mining practices, such as ore processing and dust control. Where practical, NDWR policy requires mines to return the unused or excess water to the HA from which it was withdrawn using surface application, reinjection, or as replacement water for other groundwater demands within the HA (Plume, 2003). In the MHRB, mine operations apply pumped groundwater from dewatering to reservoirs, infiltration ponds, and RIBs, or to irrigate croplands, and each of these application methods were assumed to contribute recharge to the underlying aquifers. The information for mine-related water-management practices, including recharge from mine dewatering, summarized in this section was provided by the NDWR (Jon Benedict, Nevada Division of Water Resources, written commun., 2019).

Aquifer recharge of excess water via surface application in the same HA was not always feasible because the properties of the soil and shallow alluvial aquifers may not allow for effective infiltration, or the unsaturated zones of the aquifers may not have the capacity to store the large volumes of groundwater pumped from mine dewatering (Plume, 2003). In other cases, where neither infiltration nor substitution of use had sufficient capacity for the excess water, NDWR allowed for discharge of the water directly to the Humboldt River or a tributary. In 1998, recorded releases from mine-dewatering operations to the Humboldt River or tributary peaked, and during that year, 46 percent of the total water pumped for dewatering was discharged to the river or one of its tributaries, 36 percent was returned to local aquifers by infiltration or reinjection, 16 percent was substituted for other uses (mostly irrigation), and 2 percent was lost to evaporation (Plume, 2003).

The type, timing, and location of the application of surface water from mine-dewatering operations through WY 2015 varied by mine operation. Starting in 1994, Carlin south mine operations included water application to the land surface through irrigation of croplands and to the Maggie Creek Reservoir (Plume, 2005). Water stored in the reservoir was allowed to cool before being discharged into Maggie Creek. Carlin north and Goldstrike mine operations included dewatering activities in the early 1990s. Since the early 1990s, those mine operations have combined their mine-pumped waters and applied any excess to a combination of RIBs, injection wells, and the TS Ranch Reservoir. Some of the water discharged to the TS Ranch Reservoir was used for irrigation of croplands and, to a lesser extent, for cooling water at a power plant. The remainder of the water infiltrated into the ground, with some recharging the aquifers and some being rejected and discharging to nearby springs. Much of the rejected water was recaptured and either recirculated back to the reservoir or applied to the surface elsewhere. Goldstrike mine operations also included the release of water from dewatering directly to the Humboldt River near the Argenta gage (fig. 2) via a canal between 1997 and 1999. In 1996, Cortez mine operations included the application of water from dewatering to irrigation lands and to numerous RIBs in Crescent Valley. RIBs were used in the Cove-McCoy mine operation for the surface application of water from dewatering between 1988 and 2001. When the Lone Tree mine operation was actively dewatering from the late 1990s through 2006, excess water was applied to RIBs, and starting in 1992, discharged directly to the Humboldt River near the Iron Point relief canal (fig. 2). The Twin Creeks mine operation applied excess water from dewatering to RIBs and to Rabbit Creek, and only RIBs were used by the Turquoise Ridge mine operation. Starting in 1998, water was applied to RIBs and injection wells during the Pinson Mine operation. The injection wells for the Pinson mine operation were used rarely, and the injection rate was typically low. Since the start of mining in 2011 and through 2015, water pumped at the Phoenix mine operation for dewatering has been consumed by mining practices, and no excess water has been applied to the surface. Although generally not used as a site for water application, pit lakes are surface-water bodies that form at open pit mines usually after mining and dewatering operations are discontinued. Pit lakes are discussed later in the “Evaporation from Mine-pit Lakes” section.

The movement of groundwater in the MHRB is generally toward the Humboldt River flood plain (Plume and Ponce, 1999). In HAs south of the Humboldt River, groundwater moves from recharge areas in mountains toward the central axis of each basin and then northward toward the Humboldt River. Subsurface flow between basins is through the basin fill or bedrock aquifers that underlie the surface drainages that connect the basins. The MHRB boundary was selected to minimize the likelihood of subsurface flow across the study-area boundary.

In the Susie Creek, Maggie Creek, and Marys Creek Areas, which are near major mining operations that affect groundwater flow from pumping for dewatering, groundwater flow is southward toward the Humboldt River (Plume, 1995). Groundwater flow in the upper part of the Maggie Creek Area is separate from flow in the lower part of the area and adjacent parts of the Susie Creek and Marys Creek Areas. Southward flow in the upper part of the Maggie Creek HA is in basin-fill deposits and mostly discharges in the south part of the upper Maggie Creek HA as seepage that sustains the baseflow of the lower part of Maggie Creek HA. Groundwater flow in the lower part of the Maggie Creek Area and adjacent parts of the Susie Creek and Marys Creek Areas is southward in basin-fill deposits. In this part of the MHRB, groundwater flow crosses low alluvial divides that separate the three areas. The resulting seepage to the Humboldt River channel and nearby spring discharge contribute to the baseflow of the river.

The conceptualization of the occurrence and movement of groundwater in the MHRB for the 1960s (predevelopment), 1982, and 1996 (post-development) are based on previous studies by Plume and Ponce (1999). The periods were selected for the following reasons: (1) irrigation pumping and its effects on groundwater levels were relatively minor in the early 1960s, and water-level measurements were not available until the 1960s; (2) by 1982, the effects of 20 years of irrigation pumping were becoming apparent in parts of the MHRB, and USGS and NDWR began systematic water-level monitoring in the MHRB; and (3) by 1996, the effects of 6–8 years of mine pumping, including dewatering, and continued and expanded irrigation pumping were apparent, and the USGS, NDWR, and mining companies initiated systematic water-level monitoring near major mine operations.

Water levels in the 1960s represent long-term, steady-state conditions. In many parts of the MHRB, groundwater levels did not change from the 1960s through 1996. However, water levels had started to decline by the early 1980s in areas irrigated with groundwater, such as Antelope Valley and Middle Reese River Valley HAs (Plume and Ponce, 1999). By 1982 and 1996, water-level cones of depression from pumping covered areas of 5 and 20 mi², respectively, in Antelope Valley HA and covered areas of 4 and 30 mi², respectively, in Middle Reese River Valley HA (Plume and Ponce, 1999). Groundwater also was pumped for irrigation in the Clovers Area, Kelley Creek Area, and Pumpernickel Valley HAs, but these irrigated areas were small compared to those in Antelope Valley and Middle Reese River Valley HAs, and the resulting areas of water-level declines were smaller and more difficult to define (Plume and Ponce, 1999).

Starting in the mid- to late-1990s, the largest water-level declines in the MHRB were associated with pumping for mining purposes, including mine dewatering, at several large open-pit mines. Pumping rates at these mines in 1998, which was the peak for many mines, ranged from as little as about 1,100 acre-ft/yr (about 680 gallons per minute [gal/min]) at the Pinson Mine operations on the west side of the MHRB to about 51,000 acre-ft/yr (about 32,000 gal/min) at the Lone Tree mine operation also on the west side, and about 100,000 acre-ft/yr (about 62,000 gal/min) at the Goldstrike mine operations in the northeast part (Plume, 2003). By 1996, water-level changes at the Pipeline Mine operations were mapped as sharp inflections of potentiometric contours near the mine (Plume and Ponce, 1999). At other mines, well-developed cones of depression ranged from a few square miles in extent at the Twin Creeks and Lone Tree mine operations to about 30 mi² at the Goldstrike mine operations (Plume and Ponce, 1999). Water-level declines have ranged from 100 to 200 ft at the Twin Creeks mine operations to more than 1,000 ft at the Goldstrike mine operations, Carlin north mine, and Carlin south mine operations (Plume and Ponce, 1999). Plume and Ponce (1999) provide potentiometric maps with sharp inflections of potentiometric contours indicating the effect of pumping between 1982 and 1996, and Plume (2005) provides water-level drawdown contours near the Carlin north and Carlin south mine operations that show well-developed cones of depression that resulted from mine-dewatering operations.

From 1996 through the mid-2010s, groundwater pumping to support mine operations continued but decreased from a high of 250,000 acre-ft in WY 1998 to about 150,000 acre-ft in WYs 2010–14 (Jon Benedict, Nevada Division of Water Resources, written commun., 2020; Davis and others, 2026). Mine pumping in WY 2015 was the lowest since pre-WY 1996, at about 100,000 acre-ft. Irrigation pumping, however, continued to increase from the mid-1990s, peaking in WY 2004 at about 250,000 acre-ft (Davis and others, 2026). Then, irrigation pumping declined until late 2000s, increased every year since WY 2011, and in WY 2015 was about 230,000 acre-ft (Davis and others, 2026). Because pumping volumes in WY 2015 were like those of WY 1996, the groundwater cones-of-depression near the mines likely continued to expand, except near the mines that reduced pumping after WY 1996 such as at the Lone Tree mine and Goldstrike mine operations.

Groundwater recharge can be defined as the component of precipitation or streamflow that infiltrates below the soil zone or streambeds to replenish the water table and groundwater storage. Because groundwater and surface water often are connected, the term groundwater recharge can be imprecise. In this report, groundwater recharge is not meant to describe shallow subsurface flow that returns to surface water within short periods such as days or within a year (interflow), but rather, it refers to the deeper infiltration that replenishes aquifers on a seasonal or longer-duration time scale.

The principal sources of groundwater recharge in the MHRB are (1) infiltration of precipitation that falls in mountain ranges (mountain-block recharge), (2) runoff from mountain ranges that infiltrates alluvial slopes (mountain-front recharge), and (3) infiltration of streamflow across valley floor and fluvial deposits (stream recharge). Very little groundwater recharges directly from precipitation on the valley floors because evapotranspiration generally exceeds precipitation. Previously published annual recharge estimates by HA (table 8) vary greatly, ranging from about 1,500 acre-ft/yr in Marys Creek Area (HA 052; table 8) to about 50,000 acre-ft/yr in Pine Valley (HA 053; table 8). The conceptualization and distribution of each of these sources of recharge is discussed in this section.

Groundwater discharges from aquifers in the MHRB by evapotranspiration, pumping withdrawal, discharge to streams and springs, and subsurface outflow. Groundwater evapotranspiration is a process by which shallow groundwater is either evaporated from soils or transpired by phreatophytic vegetation, which are deeply rooted plants that subsist on groundwater. Pumping includes groundwater withdrawals from aquifers by wells used for the irrigation in basin lowlands. Pumping also includes withdrawals from wells for municipal (production), power supply, mining use, domestic supplies, stock water, commercial, and industrial uses. Discharge to streams occurs when water levels in aquifers adjacent to streams are above stream stage, resulting in groundwater movement from the aquifer into the stream. Subsurface outflow is groundwater that leaves the MHRB through the subsurface and moves into areas adjacent to the MHRB.

Subsurface inflow to the MHRB is through basin-fill deposits that are near and underlie the Humboldt River on the east side of the study area near the Carlin gage. Similarly, subsurface outflow occurs through basin-fill deposits that underlie the flood plain of the Humboldt River on the western study-area boundary near the Imlay gage. The annual quantity of subsurface inflow and outflow for the study area in these two locations was estimated and is described in the “Boundary Conditions and Flow Processes” section. Subsurface groundwater inflow between the Upper Reese River Valley (HA 056) and Middle Reese River Valley (HA 058; fig. 2) was likely minimal because the extent of the basin-fill hydrogeologic units underlying the Reese River at the HA boundary is small compared to the basin-fill units underlying the Humboldt River at the Carlin and Imlay gages. Additionally, groundwater flow across MHRB boundaries in bedrock was assumed negligible because the MHRB boundary generally corresponded to surface-water divides, and it was assumed that the groundwater divides coincided with the surface-water divides. Additionally, the areas that correspond to the MHRB boundary generally were underlain by low permeability rocks.

The MHRB was discretized into a 3-dimensional grid of 329 columns and 428 rows horizontally aligned north-south and 6 layers vertically, resulting in a total of 844,872 grid cells (249,996 active cells). The projection used for geospatial data management was World Geodetic System of 1984 (WGS 84), Universal Transverse Mercator, Zone 11 North, in ft. Each square cell had an area of approximately 0.14 mi² (2,000 ft on a side) with variable thickness. The model grid overlapped parts of the UHRB and LHRB model grids (Carroll and others, 2023; Nadler and others, 2023; fig. 16), but the overlap was minimal because the active model grids mostly aligned along assumed no-flow groundwater boundaries, and the boundaries were coordinated by collaboration during model development. Land-surface altitudes and the stream network were derived from a composite 30-meter (m) DEM (Gesch and others, 2002) and the National Hydrography Dataset (NHD; U.S. Geological Survey, 2017) datasets, respectively. Land-surface altitude for each of the MHRB model cells was determined by calculating the mean 30-m DEM over the area of coinciding MODFLOW grid cells. Some altitudes along streams were adjusted to ensure streambeds were below land surface and streambeds sloped from higher upstream to lower downstream altitudes.

The geometry and depths of the actual hydrogeologic units in the MHRB are largely unknown and poorly defined; therefore, a simplified approach was used to vertically define and discretize the hydrogeologic units represented in the model. The hydrogeologic framework used in the model generally followed the conceptual model and hydrostratigraphic layering discussed earlier in this report, and the six vertical numerical model layers in the model grid generally are equivalent to the six hydrostratigraphic layers from figure 5.

In this study, hydrostratigraphic zones are defined as the numerical model representation of the hydrogeologic units described in the conceptual model and in table 5. In general, basin-fill hydrogeologic units (valley floor, alluvial slope, fluvial deposits, and older basin fill; table 5) compose the primary aquifers in the MHRB and were assigned to hydrostratigraphic zones in model layers 1–4 in the numerical model (figs. 17A–D; table 10). Basin-fill hydrogeologic units that typically are characterized as aquitards include playa and blue clay and were assigned to hydrostratigraphic zones in model layers 1–3. Consolidated rock hydrogeologic units (volcanic, carbonate, crystalline, and clastic rocks; fig. 17E; table 5) that constitute the mountains and extend beneath basin-fill sediments were assigned to hydrostratigraphic zones in model layers 5 and 6 (fig. 17E; table 10). An undifferentiated hydrostratigraphic zone was assigned to layers 5 and 6 beneath the basin-fill layers where the hydrogeologic unit was largely unknown (fig. 17E; table 10) and could be an unknown combination of basin fill and consolidated rock units. The units represented by layers 5 and 6 are not generally productive aquifers (except in extensive areas of carbonate rocks or other lithologies with widespread connected fractures), and as a result, they are generally represented in the model as zones with low horizontal hydraulic conductivity.

The distribution of the hydrostratigraphic zones at land surface, which are the uppermost hydrostratigraphic zones in the model for layers 1, 2, 4, and 5, are shown on figure 16. The distribution of hydrostratigraphic zones applied in the numerical model are based on the mapped hydrogeologic units in the MHRB (fig. 7). The extent and distribution of hydrostratigraphic zones assigned to model layers 1 and 3 were based on the extent of the fluvial deposits, alluvial slope, valley floor, and playa hydrogeologic units (tables 5, 10). The extent and distribution of model layer 2 was based on the extent of the blue-clay hydrogeologic unit (table 10; fig. 9). Model layer 1 is horizontally continuous in the valleys of the MHRB where these deposits exist; however, model layer 3 is discontinuous and only exists beneath model layer 2 or where model layer 1 deposits were greater than 50 ft and do not coincide with a model cell where a stream is represented. Model layer 4 represents the older basin-fill deposits that extend beneath the younger basin-fill deposits of model layers 1, 2, and 3 but also is present at the land surface in some HAs. The distributions of hydrostratigraphic zones at the tops of model layer 5 are inferred from surface distributions of the consolidated rock hydrogeologic units (bedrock) based on Maurer and others (2004) and consisted of volcanic, carbonate, crystalline, and clastic rock hydrogeologic units (tables 5, 10; fig. 16). Hydrostratigraphic zone distributions in model layer 6 were equivalent to those represented in model layer 5. The distributions of bedrock zones below basin-fill deposits were based on the nearest exposure at land surface along basin margins or assigned to an undifferentiated hydrostratigraphic zone in the model.

The thickness of model layer 1 generally was constrained to between 25 and 50 ft (table 10); however, the thickness of model layer 1 was increased, sometimes by several hundred feet, in model cells where a stream is represented (stream cells), where needed, with a bottom elevation such that stream connectivity was maintained with model layer 1, which includes fluvial deposits. Additionally, the bottom of model layer 1 was assigned to the top of the blue clay where the blue clay exists (fig. 17). The thickness of model layers 2, 3, and 4 were determined from lithologic logs in the model area where information was available (described in the “Hydrostratigraphic Conceptualization” section). The thickness of model layer 2 was assigned the thickness of the blue clay, which ranged from about 11 to 134 ft. The thickness of model layer 3 was assigned where the thickness of younger basin-fill deposits generally exceeded 50 ft or where younger basin-fill deposits were identified in logs as existing below the blue clay. The thickness of layer 3 ranged between 5 and 332 ft. Where the thickness of model layers 2 or 3 was 0 ft, such that the bottom of model layer 1 was in direct contact with the top of either model layer 3 or layer 4, model layers 2 or 3, or both, were designated as “pass-through” cells in MODFLOW 6 (figs. 17B, C; Langevin and others, 2017). Pass-through cells allow for hydraulic connection between vertically adjacent, non-sequentially numbered model layers (fig. 17). Layer 4 thickness, determined from lithologic logs, ranged from about 50 to 1,250 ft, with an average thickness of about 100 ft. The top of model layer 5 was assigned to the land surface where the consolidated-rock hydrogeologic units were present at the land surface and to the bottom of model layer 4 beneath the basin-fill hydrogeologic units. The thickness of model layers 5 and 6 was uniformly set to 1,200 and 2,400 ft, respectively.

In each hydrostratigraphic zone in the model, hydraulic conductivity, vertical anisotropy ratio (ratio of vertical hydraulic conductivity to horizontal hydraulic conductivity), specific storage, and specific yield were assigned generally uniform values (fig. 17) based on values estimated by Prudic (2007; table 5). Localized deviations from the uniform distributions were made before model calibration for some hydrostratigraphic zones because of known aquifer heterogeneities and observed water-level distributions. Aquifer properties assigned to the model and their distribution are discussed further in the “Parameterization” section. All cells in the model were represented using convertible layers, meaning saturated thickness and transmissivity vary with changes in calculated hydraulic head.

The groundwater flow model was temporally discretized into a single steady-state stress period followed by 660 monthly transient stress periods. The steady-state stress period represented average hydrologic conditions in the MHRB from October 1, 1945 (start of WY 1946), through September 30, 1958 (end of WY 1958). This timeframe was selected because it was before significant development of groundwater resources (Cohen, 1964d), and it included continuous and overlapping streamflow records from streamgages along the middle Humboldt River (table 4). Palisade gage streamflow records also were used for the selection of the steady state model period because surface-water diversions from the Humboldt River in the MHRB are regulated based on the streamflow at the Palisade gage. The steady-state stress period was selected so that the mean annual streamflow at the Palisade gage during the steady-state stress period matched the pre-development and long-term (1903–58) mean annual flow at the gage (fig. 11). The mean annual flow for the Palisade gage for the 56-year period 1903–58 and WYs 1946–58 was about 265,000 acre-ft/yr (fig. 11).

The monthly transient stress periods of the model were temporally bounded from October 1, 1960 (start of WY 1961), through September 30, 2015 (end of WY 2015), to represent the period when groundwater development became important in the MHRB. The transient part of the model consisted of 660 monthly stress periods, each with a single time step, to represent stress changes for annual non-irrigation and irrigation seasons during the simulation period. In the model, the non-irrigation season starts on October 1 and ends on the last day of February of the following year, and the irrigation season starts March 1 and ends September 30 in any year. The length of the simulated irrigation season was not equivalent to the length of the actual MHRB irrigation season as defined by the Humboldt and Little Humboldt Decrees (Mashburn and others, 1935; Hennen, 1964a). The Humboldt Decree governs diversions on the mainstem of the Humboldt River where the annual irrigation season is between March 15 and September 15, and The Little Humboldt Decree governs diversions in Paradise Valley where the annual irrigation season is between April 1 and September 27. As a result, time-weighted rates were used in the model to ensure the simulated monthly volumes were equal to the actual monthly volumes for pumping, diversion, and irrigation recharge for the irrigation seasons. Irrigation season stresses (groundwater pumping, surface water diversions, and recharge from flood irrigation in Paradise Valley) represented in the model were specified using a constant monthly rate, and therefore, the simulated pumping, diversion, and irrigation recharge rates were time-weighted to account for the discrepancy between the actual and simulated irrigation seasons.

Stream features in the numerical model included the Humboldt River, major and minor tributaries to the Humboldt River, most surface-water diversions from the Humboldt and Little Humboldt Rivers, and supplemental discharge to streams from mine-dewatering operations. Streamflow in the MHRB that could be subjected to capture by groundwater pumping is represented in the model using multiple approaches. Interaction between surface water and groundwater in the streams and shallow aquifers in the MHRB was simulated with the DRN and Streamflow-Routing (SFR) Packages of MODFLOW 6 (Langevin and others, 2017). The DRN Package is designed to simulate the effects of agricultural drains, springs, and other features that remove water from the aquifer at a rate proportional to the difference between the head in the aquifer and some fixed head or elevation that represents the bottom of the stream channel or spring outlet. If the head in the aquifer is above that elevation, flow into the drain is proportional to the gradient; if head in the aquifer is below that elevation, there is no flow into the drain (Langevin and others, 2017). The SFR Package uses the continuity equation to route streamflow through a network of rectangular channels (Langevin and others, 2017).

Most of the non-headwater stream reaches in the MHRB lose flow, although the Humboldt River in the Carlin area and lower reaches of Pine Creek are exceptions and generally gain flow. The SFR Package was used for streams that have the potential of providing recharge (infiltration of streamflow) to or receiving discharge (groundwater discharge to streams) from the shallow aquifers in the MHRB. The DRN Package was used for streams that were conceptualized to primarily receive discharge from shallow aquifers (gaining streams). Recharge via infiltration of runoff from small tributary streams that cross alluvial slopes was represented as recharge from infiltration of precipitation in the MHRB and is discussed in the “Distribution of Recharge from Precipitation (MHRB-PRMS model)” section. Model cells used to represent streams in the MHRB using either the SFR or DRN Packages are collectively called stream cells in this report.

Steady-state and monthly streamflow in the Humboldt River, major tributary streamflow up to the first streamgage upstream from a stream’s confluence with the Humboldt River, water diverted for surface-water irrigation, and discharge to the Humboldt River or Maggie Creek from mine-dewatering operations were represented using the SFR Package. Perennial and ephemeral reaches of major tributaries above the first streamgage upstream from a stream’s confluence with the Humboldt River as well as select minor tributaries in the model area were represented using the DRN Package (fig. 18; Langevin and others, 2017). The DRN Package is used to avoid introducing additional water to the model, which is already accounted for by either recharge or specified stream inflows.

Streams in the SFR Package are represented by sections of the stream within individual model cells called reaches in MODFLOW 6. In this study, segments were defined as groups of reaches between tributary inflows or PODs that had similar stream channel characteristics. Stream segments were initially divided based on locations of tributary inflows and PODs. These stream segments were further subdivided using a combination of hydrogeology for basin-fill deposits (discussed earlier in “Discretization” section) and National Agricultural Imagery Program (NAIP) aerial imagery (U.S. Department of Agriculture and others, 2017) by identifying segments with substantial changes in adjacent hydrogeology or landforms. Segments in canyons were differentiated from segments in flood plains, and if there was a substantial change or variation in flood plain width, then a new segment was differentiated. For all reaches, evapotranspiration from the streams was assumed negligible and not included as a parameter in the SFR package.

Model boundaries were defined along watershed and HAs divides so that subsurface groundwater inflows and outflows to the MHRB were minimal. Subsurface groundwater inflow to the model area is assumed to be a small amount and occurs in the shallow aquifers beneath the Humboldt River near the Carlin gage. Subsurface groundwater outflow from the model area also is assumed to be a small amount of flow and also occurs in the shallow aquifers beneath Humboldt River near the Imlay gage. Groundwater inflow near the Carlin gage and groundwater outflow near the Imlay gage were represented as specified flux boundaries using the WEL Package (Langevin and others, 2017) in the model (fig. 18). The WEL Package is designed to simulate features that either withdraw or add water to the aquifer at a specified rate during a stress period, where the rate is independent of both the cell area and the head in the cell (Langevin and others, 2017). The inflow near Carlin was held constant during the simulation period. The outflow near Imlay was initially specified but allowed to decrease if the modeled water level in individual boundary cells was below 20 percent of the total thickness of the cell.

Groundwater inflow and outflow for the MHRB used in the model WEL Package were estimated using Darcy’s Law (Freeze and Cherry, 1979; eq. 1):Q = K*i*A(1)where

Groundwater inflow near the Carlin gage for the alluvial and underlying basin-fill aquifers was estimated using an assumed hydraulic conductivity of 1.0 ft/d (Prudic, 2007), a hydraulic gradient of 0.005 ft/ft (Plume and Ponce, 1999), and a cross-sectional flow area that is 4,000 ft wide and 77 ft deep based on the geometry of the hydrogeologic units at the model boundary. Groundwater inflow near the Carlin gage for the bedrock aquifers was estimated using an assumed hydraulic conductivity of 0.5 ft/d (Prudic, 2007), a hydraulic gradient of 0.005 ft/ft (Plume and Ponce, 1999), and a cross-sectional flow area that is 26,000 ft wide and 2,000 ft deep based on the geometry of the hydrogeologic units at the model boundary. Subsurface inflow on the east side of the model area is estimated as 0.018 ft³/s (13 acre-ft/yr) for the alluvial and basin-fill aquifers (layer 4) and 1.5 ft³/s (1,100 acre-ft/yr) for the bedrock aquifers (layers 5 and 6). Specified groundwater inflow rates near Carlin were allowed to vary during calibration using a multiplier parameter that was applied to all boundary cells in each layer.

Groundwater outflow near the Imlay gage for the alluvial and underlying basin-fill aquifers was estimated using an assumed hydraulic conductivity of 10.0 ft/d for the alluvium and 1.0 ft/d for the underlying basin fill (Prudic, 2007), a hydraulic gradient of 0.001 ft/ft (Plume and Ponce, 1999), and a cross-sectional flow area that is 40,000 ft wide and 50 ft deep for the alluvium, and 400 ft deep for the underlying basin fill. Groundwater outflow near the Imlay gage for the bedrock aquifers was estimated using an assumed hydraulic conductivity of 0.5 ft/d (Prudic, 2007), a hydraulic gradient of 0.001 ft/ft (Plume and Ponce, 1999), and a cross-sectional flow area that is 104,000 ft wide and 2,000 ft deep. Subsurface outflow on the west side of the model area was estimated as 0.23 ft³/s (170 acre-ft/yr) for the alluvium (layer 1), 0.19 ft³/s (130 acre-ft/yr) for the deeper basin-fill aquifers (layer 4) and 1.2 ft³/s (870 acre-ft/yr) for the bedrock aquifers (layers 5 and 6). Specified groundwater outflow rates near Imlay were allowed to vary during calibration using a multiplier parameter that was applied to all boundary cells in each layer.

The RCH package (Langevin and others, 2017) was used to specify groundwater recharge in the MHRB from (1) infiltration of precipitation, which may include infiltration of mountain-front runoff or from intermittent and ephemeral streams; (2) infiltration from Gumboot Lake; and (3) infiltration from surface-water applications from mine-dewatering operations. The RCH package is designed to simulate areally distributed recharge to the groundwater system (Langevin and others, 2017), but it also can be used to simulate point-source recharge, such as infiltration recharge at RIBs. The distribution of recharge from precipitation and from intermittent and ephemeral streams (mountain-block and mountain-front recharge) was estimated using GSFLOW-PRMS model (Markstrom and others, 2008) and applied as areal recharge in the MODFLOW model. During calibration of the groundwater flow model, the areal recharge distribution estimated by GSFLOW-PRMS was scaled using a multiplier for each HA. Intermittent recharge from Gumboot Lake was estimated from previous studies for variable lake sizes based on upstream streamflow and historic estimates of recharge. Infiltration from mine-dewatering operations also is a source of groundwater recharge and includes infiltration of water discharged to streams, stored in reservoirs, transferred to RIBs, or the recharge component water applied as irrigation to croplands. Infiltration rates and locations of surface-water applications from mine-dewatering operations were estimated from mine water-management records provided to the NDWR from the mine operators. Seepage (recharge from infiltration of streamflow) of the Humboldt River and the lower part of connected tributaries was simulated with the SFR package as discussed in the “Streams” section.

Groundwater evapotranspiration (ET_g) was simulated using the MODFLOW EVT Package (McDonald and Harbaugh, 1988; Harbaugh and McDonald, 1996; Langevin and others, 2017). The EVT Package removes water from the model at a variable rate controlled by an assigned maximum allowable ET_g (ET_g(max)) rate, the evapotranspiration surface (typically the land surface altitude), and evapotranspiration extinction depth. The EVT Package removes groundwater at the ET_g(max) rate when the hydraulic head in a corresponding model cell is at the land surface and decreases linearly to zero when the hydraulic head falls below the assigned extinction depth.

Simulation of ET_g in the MHRB was enabled in all areas consisting of basin-fill sediments (unconsolidated sediments; fig. 6) and disabled in areas consisting of consolidated-rock units. Spatially, allowable ET_g was zoned into two areas: (1) irrigated and (2) non-irrigated areas. The extent of irrigated areas was estimated based on land-use categories from 2016 aerial imagery (Dewitz, 2019; Jin and others, 2019), and model cells were classified as an “irrigated area" if more than 50 percent of the cell contained irrigated land based on land-cover classes from Jin and others (2019; fig. 26). Non-irrigated areas were defined by the boundaries of the basin-fill sediments that were not designated as irrigated areas.

An initial ET_g(max) rate of 3.0 ft/yr was assigned to the evapotranspiration surface, which was equivalent to the land surface or the top of model layer 1, for only basin-fill sediments in each HA. ET_g(max) in the steady-state part of the model was applied using the initial rate of 3.0 ft/yr for non-irrigated areas, and a rate of 0.6 ft/yr was used in the irrigated areas to account for ET_g during the non-irrigation season. The proportion of summer to winter annual ET_g was calculated from the mean of monthly ET_g rates in summer (April–September) and winter (October–March) months as a proportion of the annual rate measured by Berger and others (2016) in Kobeh Valley (not shown), located south of Pine Valley (HA 053). Based on monthly measurements of ET_g from Berger and others (2016), 25 percent of the annual ET_g occurs in the winter, and the remainder (75 percent) occurs in the summer. All rates applied to the model were adjusted for leap years. Transient ET_g(max) rates were initially assigned such that the summer ET_g(max) was 75 percent of the annual ET_g(max) rate and winter ET_g(max) rate was 25 percent of the annual ET_g(max) rate. The ET_g extinction depth was assigned an initial value of 25 feet to all enabled ET_g areas based on phreatophyte root depths summarized by Huntington and others (2022). The ET_g(max) rate, extinction depth, and proportion of summer to winter ET_g were allowed to vary by HA during model calibration and are discussed in the “Model Calibration Approach” section.

Groundwater pumping is a groundwater model stress that removes water directly from the aquifers and in the MHRB-included withdrawals for municipal supply (production), power generation, irrigation, and mining. Although, conceptually, groundwater withdrawn by pumping is removed from a model, in the MHRB model, parts of groundwater withdrawn for mining were not consumed and were applied as irrigation, recharged to groundwater through infiltration basins or injection, or directly discharged to streams. Groundwater pumping for mining was represented using the total use rate, and excess water applied to crop sites, at infiltration basins, or discharged to streams was simulated explicitly as discussed in the “Recharge from Mine Dewatering” section.

Groundwater pumping in the MHRB was represented with the WEL Package in MODFLOW 6 (Langevin and others, 2017). The WEL Package is designed to simulate features such as wells that withdraw water from or add water to an aquifer at a specified rate during a stress period where the rate is independent of both the cell area and the head in the cell (Langevin and others, 2017). The WEL Package includes a user-defined option to reduce the pumping rate of a well if that well dewaters a cell beyond a user-specified percentage of cell thickness. The option only applies to cells designated as convertible in MODFLOW and that become unconfined during the model simulation. The automatic flow reduction was arbitrarily set to 2 percent for all wells to attempt to simulate the full input pumping rate while still allowing MODFLOW to reduce the pumping rates, where necessary, to attempt to prevent cell drying.

Four faults in the model area that were assumed to act as barriers to groundwater flow were represented using the HFB Package (fig. 18). The HFB Package adjusts the horizontal hydraulic conductance between model cells as computed by the internal flow package (the Node Property Flow Package [NPF] in MODFLOW 6) to approximate a horizontal flow barrier (Langevin and others, 2017). The four faults represented with the HFB were the Post and Siphon faults, located near Goldstrike and Carlin north mine operations, and the Crescent and Cortez faults, near Cortez Mine (figs. 7, 18). The HFBs were assigned locations between model cells that corresponded to the mapped faults and were assumed to exist vertically in layers 5 and 6, which represent bedrock underlying basin fill. The identification and justification for simulating these features in the model is described in the “Geologic Structural Features” section. The property assigned to the HFB package that limits groundwater flow across a fault is the hydraulic characteristic of the barrier and is defined by the hydraulic conductivity of the barrier divided by the distance across the barrier in the direction of flow. Because the actual widths of the faults were largely unknown, the distance across the barrier was assigned to be 1 ft. An initial hydraulic conductivity value of 1.0x10⁻⁵ ft/d was assigned to limit groundwater flow across the faults. This value acted as a lumped parameter that was used in place of the hydraulic characteristic (units of per day [d⁻¹]) for all cells assigned to the HFB Package and was allowed to vary during model calibration. Other possible barriers of groundwater flow may exist in the model area where abrupt changes in hydrogeologic properties existed across a hydrogeologic contact, like along the Northern Nevada Rift (Ponce, 2004) or along basin and range mountain fronts. Possible barriers of groundwater flow across abrupt changes in hydrogeologic properties were not explicitly simulated with the HFB Package.

Parameters adjusted during the calibration of the numerical model were horizontal hydraulic conductivity, vertical anisotropy ratio (ratio of vertical hydraulic conductivity to horizontal hydraulic conductivity), specific storage, specific yield, recharge and ET_g rate multipliers, ET_g extinction depths, the proportion of ET_g that occurs in the winter, streambed and drainbed hydraulic conductivity, streambed roughness coefficients, streamflow bifurcation percentages, surface-water irrigation inefficiencies in Paradise Valley, pumping rate multipliers, and hydraulic characteristics for faults. Horizontal hydraulic conductivity and vertical anisotropy ratio were estimated at pilot points distributed within the MHRB (tables 16, 17). Specific storage and specific yield were applied uniformly by zones in each model layer (tables 16, 17). Recharge, ET_g, and irrigation pumping rate multipliers and ET_g extinction depths were applied uniformly by HA and were the same for every model stress period. Streambed hydraulic conductivity and streambed roughness coefficients were assigned uniformly for segments of the Humboldt River and major tributaries. Drainbed hydraulic conductivity was assigned to ephemeral and perennial sections of selected minor tributaries represented using the DRN package. Streamflow bifurcation percentages were assigned at locations where streamflow bifurcation occurred and where diversion records were not available. Pumping rate multipliers for mine and power generation pumping were applied uniformly to each mine or power plant, respectively, and multipliers for production wells were assigned for each municipality. The hydraulic characteristics for faults were applied uniquely to each fault for each model layer (layers 5 and 6 only). A total of 3,705 parameters were adjusted during model calibration, and parameters of the same type were assigned to distinct parameter groups.

Allowable parameter bounds (tables 16, 18) were determined by qualitatively assessing the accuracy of the initial parameter value and the range of potential parameter values (tables 5, 8) and at the discretion of the modelers. In general, initial parameters values with higher certainty were assigned narrower bounds (for example 0.9–1.1 times the initial value) and initial parameters values with lower certainty were assigned wider bounds (for example, 0.01–100 times the initial value). Initial parameter values and bounds for each or groups of parameters are further discussed in this section.

Hydraulic conductivity and vertical anisotropy ratio were estimated at 1,692 pilot point locations distributed throughout the MHRB (fig. 17). Vertical anisotropy ratio in MODFLOW 6 is the ratio of vertical hydraulic conductivity to horizontal hydraulic conductivity and is specified in the NPF Package. The use of pilot points allows for increased variation in the hydraulic property field applied to the model by interpolation of properties between pilot point locations when compared to the uniform application of properties within zones. Pilot points were distributed among the model layers in the hydrostratigraphic zones that were the model equivalent of the hydrogeologic units described in the “Hydrogeologic Conceptualization” and “Discretization” sections (fig. 7). Pilot points were spaced so that the heterogeneity of the aquifers was adequately represented, and a closer spacing was used in zones that had a larger number of observation wells or pumping wells (fig. 17). Pilot points were initially spaced using uniform row and column spacing in each zone except zone 1 (fluvial deposits), where pilot points were spaced every four rows and six columns (table 16). Additional pilot points were manually added to zones in areas with a relatively higher density of observation wells. Initial values at pilot point locations for horizontal hydraulic conductivity and the minimum and maximum allowable parameter values (table 16) were based on values presented in table 5 and generally varied by less than about 3 orders of magnitude from the initial value. Vertical anisotropy ratio pilot point values were initially uniformly assumed to be 0.1 but were allowed to vary between 0.01 and 0.50 during model calibration.

Specific storage and specific yield were estimated using uniform values assigned to each hydrostratigraphic zone and were allowed to vary during model calibration. Initial values for specific storage and specific yield (table 16) were based on values in table 5. Minimum and maximum allowable parameter values for specific storage were 1.0×10⁻⁷ to 1.0×10⁻⁴ ft⁻¹, respectively; and minimum and maximum allowable parameter values for specific yield were 0.01–0.25 (1–25 percent), respectively (table 16).

Estimates of recharge and ET_g were adjusted separately during the calibration process using recharge and ET_g(max) rate multipliers, ET_g extinction depth multipliers, and the proportion of ET_g that occurs in the winter. Multipliers were used for the RCH and EVT Package that applied to the rates used in the numerical model and were assigned independently by HA. Evapotranspiration extinction depths were also assigned to each HA. The recharge rate multipliers were applied uniformly to all numerical model stress periods (steady-state and transient) for the spatially varied PRMS estimates used in the RCH Package. Estimates of steady-state recharge rates by HA from table 8 were used to define the initial value and the upper and lower limits of recharge rate multipliers applied to the numerical model during calibration. If only one estimate of recharge was available for an HA, the upper and lower limits applied during calibration were set to plus or minus 10 percent of the initial value. For the EVT Package, the maximum potential evapotranspiration rate for each HA was initially assigned to 3 ft/yr and was adjusted using multipliers that were applied uniformly to all numerical model stress periods (steady-state and transient stress periods), and that could vary between 0.33 (1 ft/yr) and 1.67 (5 ft/yr). This range was assumed to represent and include possible transient variations for the model period. The evapotranspiration extinction depth that was initially assigned as 30 ft could vary between 10 and 50 ft by HA. The proportion of evapotranspiration that occurs in the winter was initially assumed to be 0.25 (25 percent) and could vary between 0.1 (10 percent) and 0.5 (50 percent) during calibration.

Streambed hydraulic conductivity parameters in the SFR and DRN Packages were adjusted during model calibration. Streambed hydraulic conductivity is a property that controls the stream flux simulated by the numerical model. Uniform values of streambed hydraulic conductivity and Manning’s roughness coefficients were assigned to the SFR Package for stream segments, and uniform values of streambed (drainbed) hydraulic conductivity were assigned to ephemeral and perennial segments of major and selected minor tributary streams for the DRN Package, as defined in the “Boundary Conditions and Flow Processes” section. Drain conductance is internally calculated by MODFLOW 6 using auxiliary multipliers that allow calibration of drainbed hydraulic conductivity (usually more uncertain), whereas geometries for streams represented by the DRN Package are generally known (and thus, more certain).

Initial values of streambed (and drainbed) hydraulic conductivity and Manning’s roughness coefficients were based on values estimated by Prudic (2007), and modeler discretion was used to define the upper and lower allowable parameter bounds applied during numerical model calibration (table 18). The percentage of upstream flow routed at stream bifurcation locations was initially set between 0.03 (3 percent) and 0.25 (25 percent) and was uniform for all model stress periods (steady-state and transient stress periods). The minimum and maximum allowable parameter bounds for stream bifurcations were set to 10 percent of the initial value (0.003–0.025) and 100 percent (1.0) of upstream flow, respectively.

Surface-water irrigation in Paradise Valley was applied using the RCH Package and calculated based on diversion priorities for Martin Creek and Little Humboldt River. Surface-water irrigation inefficiencies were used to represent the amount of water used for irrigation that is lost to (or recharges) the groundwater system (the amount of water not consumed by crops or evaporation) because of irrigation practices in the area. The irrigation inefficiency (as a percentage of water diverted for irrigation) was initially assumed to be 0.3 (30 percent) and was allowed to vary between 0.1 (10 percent) and 0.5 (50 percent) during model calibration and was based on non-consumptive percentages estimated for comparable irrigation practices on the Humboldt River (Steve Del Soldato, Nevada Division of Water Resources, written commun., 2015).

Groundwater pumping rate multipliers were assigned to the WEL Packages and were used to adjust the assigned pumping rate for each well in the numerical model to compensate for potential errors in estimated or reported pumping data. The assignment and distribution of the pumping rate multipliers depended on the type of withdrawal and the well location. For irrigation wells, groundwater pumping rate multipliers were assigned by HA. Rate multipliers for production wells, power generation, and mining were uniquely assigned and differed by the well operator. All groundwater pumping rate multipliers were allowed to vary between 0.9 (90 percent) and 1.1 (110 percent) during model calibration.

Groundwater subsurface inflow near the Carlin gage and outflow near the Imlay gage were represented using the WEL Package, and multipliers also were used in these packages. The initial rates of inflow and outflow were estimated using a Darcy flux approach (discussed in the “Boundary Conditions and Flow Processes” section). These estimates are based on an assumed value of aquifer transmissivity and approximate hydraulic gradient near each boundary. Multipliers were applied to the rates of inflow and outflow to account for errors in the estimated rates and were allowed to vary between 0.25 (25 percent) and 2.00 (200 percent) during model calibration.

The hydraulic characteristics for faults in the model area were assigned parameters during the calibration and were estimated independently by layer for layers 5 and 6 for each of the four faults represented in the numerical model. An assumed initial hydraulic characteristic of 1.0×10⁻⁵ d⁻¹ was allowed to vary between 1.0×10⁻⁷ and 1.0×10⁻⁴ d⁻¹. A wide range of hydraulic characteristic parameter values was used because estimates for the hydraulic characteristic of these faults were unknown.

Model calibration involves a process called history matching (White, 2018; White and others, 2020). History matching attempts to closely match real-world hydraulic observations or measurements of interest (calibration targets) to model-simulated values for a defined period by systematically adjusting model input parameters. In general, calibration targets (model observations) are direct measurements or estimates of water-level altitudes, flows in streams and from springs, and fluxes into and out of the groundwater system. Each observation in the history matching process is assigned a weight, and a residual is the difference between an observation and model-simulated value. History matching is done by attempting to reduce the sum of squared weighted residuals, or the objective function, for all calibration observations (Doherty and Hunt, 2010; Doherty, 2018; White, 2018). Each observation is assigned to a group of similar observation types.

Nine observation types were used for calibrating the MHRB numerical model: (1) steady-state and monthly mean streamflows, (2) cumulative monthly streamflows, (3) streambed fluxes, (4) steady-state streamflow diversions, (5) steady-state and transient water-level altitudes, (6) transient water-level changes, (7) water-level altitude inequality observations (with respect to land-surface altitudes), (8) steady-state median annual ET_g volume by HA, and (9) steady-state median annual ET_g area by HA. Observations of similar types were combined into groups for weighting purposes. Weighting adjusts the contribution of each residual to the objective function, and combining observations into groups ensures that no single observation type or group becomes a dominant contributor in the calibration process. Highly weighted observations will have greater effect during the calibration process compared to observations with lower weights because smaller changes in residuals for highly weighted observations will generate larger changes to the objective function than similar (or equal) residuals for observations with comparatively lower weights. Except for water-level altitudes and changes in water levels, datasets for each group consist of all similar observation types within the model domain. For example, the streamflow group includes steady-state and mean monthly streamflows, and the cumulative streamflow group includes cumulative monthly streamflows. Water-level groups were subdivided by HA because of their more local significance.

Observation weights can be assigned several ways, but there is no universal method for selecting observations and assigning observation weights. One method is to assign observation weights such that they are inversely proportional to the noise associated with each measurement, which associates the weights with the uncertainty of the observed values. Alternatively, observation weights can be assigned so that the residuals of the observations or groups of observations all have equal contributions to the objective function so that no observation or group of observations dominate the calibration process. Finally, weighting can be assigned to observations so that calibration favors the observations that are most closely aligned with key predictions required of the model (Doherty and Hunt, 2010).

Observation weights in the numerical model initially were assigned so that the residuals for each observation group had an approximately equal contribution to the objective function. Observation uncertainties were not considered when assigning observation weights. Weights for the monthly mean streamflow, cumulative monthly streamflow, water-level inequality, and water-level change observation groups were then adjusted such that their contributions to the objective function were about 3, 3, 14, and 2 times that of the other observation groups, respectively. These adjustments were based on results from early model calibration runs that indicated little or no reduction in the residuals from these observation groups compared to other groups. The weighting changes were selected to ensure that the calibration process would favor residual reduction for these observation groups instead of the other groups. Water-level inequality observations were weighted highest relative to all other groups to reduce the likelihood that calibrated model parameters would result in simulated water levels that were above the land surface, which was noted in early model calibration runs. This issue is discussed further in the “Water-level Altitude, Change, and Inequality Observations” section below. All observations and weights are available in the accompanying USGS data release (Davis and others, 2026).

Parameter estimates used for analysis for this study were the “base” parameters from the fifth PESTPP-IES iteration. The base parameters were selected because they represent the values that PESTPP-IES uses to develop the ensemble of parameter realizations (possible parameter combinations) to test during an iteration and because all other realizations within an iteration include PEST++-calculated observation noise to provide a more robust estimation of model uncertainty. Each realization within an iteration is a reasonable ensemble of parameters for the model given the calibration dataset provided by the modeler. However, the fifth iteration was selected because it represented the point at which successive changes to the objective function were notably smaller than previous iterations, yielding no significant reduction in the objective function after iteration five. The resulting parameters from the base realization approximates a more appropriate minimum error solution (reduction of the objective function), like traditional PEST, because manufactured observation noise is not included. In general, the calibrated base parameter values (tables 11, 16, 17, and 18) from iteration five were like previously published values (table 5). Hydrogeologic property estimates for pilot points are summarized in table 16, hydrogeologic property estimates by model layer and hydrostratigraphic zone are summarized in table 17, and other parameter calibration results are summarized in table 18. All parameter ensembles for the fifth PESTPP-IES iteration, including the “base” realization, are available in the accompanying USGS data release (Davis and others, 2026).