Purpose and Scope

This report documents the development and calibration of a three-dimensional, transient numerical model of the Long Island aquifer system and the use of that model to improve understanding of how the system will respond to changing natural and anthropogenic stresses during different time scales. The report describes the compilation and analysis of a diverse set of climatic, physiographic, geologic, hydrologic, and water-use data that underlie the numerical model and how those data are synthesized into a numerical model. The model simulation of historical hydrologic conditions from 1900 to 2019, as determined using historical water levels, streamflows, and chloride concentrations, and the calibration of model inputs using emerging inverse methods to best fit observed conditions, are also discussed in the report.

An updated overview of the geologic framework of western Long Island is presented in Stumm and others (2024), including the mapped extents and surface altitudes of the major glacial, preglacial Pleistocene, and Cretaceous units that comprise the Long Island aquifer system. The methods used to develop a three-dimensional hydrogeologic framework from the mapped framework are discussed in this report, along with modifications to an existing islandwide, grid-independent texture model developed from lithologic logs and methods used to incorporate that data as water-transmitting properties into the numerical model. Natural recharge to the aquifer determined by use of a soil-water balance model that utilized climate, soil, and land-use data is presented for human-influenced and natural landscapes from 1900 to 2019. Sources of anthropogenic recharge also are presented, including potential recharge from impervious surface runoff, leaky water-supply infrastructure, and wastewater return flow in sewered and unsewered areas, as are the methods and data sources used to estimate those recharge components. Historical groundwater withdrawals, the underlying data sources, and methods used to fill in data gaps are also described for 1900 to 2019. Historical water levels, streamflows, and chloride concentrations are discussed, along with the data sources and methods used in analyzing the data for use in calibration and verification of the numerical model.

The report documents the synthesis of the hydrologic data into a regional-scale, transient numerical model of the Long Island aquifer system. The model grid extent and discretization are presented, as are the locations and types of simulated hydrologic boundaries, the distribution of simulated recharge, and the locations of simulated pumping wells. The initial values of hydraulic parameters, including boundary leakances, horizontal and vertical hydraulic conductivity, and specific capacity and storage are presented, as well as the assumptions underlying them and comparisons to values in the literature. The use of a recently developed inverse calibration method to adjust these initial values to best match observed annual averaged water levels and streamflows is also discussed in detail. This discussion includes the parameterization of the model, implementation of inverse techniques, and fit between observations of water levels, streamflows, and their simulated equivalents. The agreement between simulated chloride concentrations for current and historical observations of chloride from water samples and estimates from borehole and surface geophysics is also presented in the report.

Important considerations when utilizing the model to make predictions of future hydrologic conditions are discussed, including potential errors associated with uncertainties in estimated input parameters and structural errors in the model design, hydrogeologic interpretations, and data gaps in estimates of pumping and anthropogenic recharge. The effect of parameter uncertainties on model predictions are quantified and illustrated using probabilistic methods and examples of the practical use of uncertainties in predicting hydrologic responses. Examples of structural error arising from data gaps in historical pumping are also presented. Descriptions are included on the modifications to the regional model that were implemented for the prediction of future hydrologic responses to changes in natural and anthropogenic stresses.

The report also includes analyses of changes in hydrologic conditions in the Long Island aquifer system in response to historical changes in sea level, recharge, groundwater withdrawals, and changes in water infrastructure from 1900 to 2019. Annual hydrologic budgets for the aquifer system that include changes in recharge, pumping, surface water discharge, and storage are discussed. The report also includes a discussion of the relative importance of historical changes in natural and anthropogenic stresses on the water table, streams, and the position of the freshwater/saltwater interface. This discussion includes maps of the current [2019] and historical changes in water table altitude and hydrographs of water levels and streamflows that illustrate time-varying hydrologic conditions at selected locations. Changes in the position of the boundary between fresh and saline groundwater (also known as the freshwater/saltwater interface) between 1900 and 2019 also are presented in maps and cross sections. The report also includes a discussion of general limitations associated with the use of numerical models, and limitations specific to the regional model of the Long Island aquifer system.

Hydrogeology

Long Island is underlain by unconsolidated sediments of late Pleistocene and late Cretaceous age that are, in turn, underlain by relatively impermeable crystalline bedrock of pre-Cambrian age (fig. 2). The maximum thickness of this wedge of unconsolidated sediments is more than 2,000 feet (ft) beneath the south-central part of the island. Long Island is at or near the southernmost advance of the continental ice sheet during the Wisconsinan glaciation, about 18,000 years before present (Smolensky and others, 1989). Glacial moraine sediments were deposited during periods of ice advance and retreat. These sediments formed morainal ridges that trend east-west along the spine of the island, which is bifurcated at the eastern end where two morainal ridges diverge to form the North and South Forks. The moraine deposits consist of poorly sorted sand, silt, and clay deposited marginal to the advancing ice sheet. Moraines are bounded generally to the south by glacial outwash deposits and to the north by ice-contact deposits.

Glacial outwash comprises fluvial sediments deposited in front of the ice margin and generally consist of well-sorted sand and gravel. Ice-contact deposits, which consist of sand, silt, and clay, were deposited within or near the margin of the retreating ice sheet. Pleistocene-age sediments on Long Island are underlain by upper Cretaceous-age sediments that were deposited in deltaic (fluvial and marsh) environments about 65 million years before present (Smolensky and others, 1989). These sediments represent the northern extent of the Northern Atlantic Coastal Plain aquifer system, a large wedge of mostly marine Cretaceous-age sediments that underlie the coastal plain of the mid-Atlantic region (Masterson and others, 2016). Only 3 of the 18 geologic units that compose the Northern Atlantic Coastal Plain aquifer system—the Magothy, Potomac, Potomac-Patapsco formations—underlie Long Island. The overlying units either were not deposited or were subsequently removed by erosion. The hydrogeologic equivalents of the Magothy, Potomac, and Potomac-Patapsco on Long Island are known as the Magothy aquifer, the Raritan confining unit, and the Lloyd aquifer, respectively. Cretaceous-age sediments are absent along the northern shore of Long Island where the units either pinch out or have been removed by erosion. Glacial sediments in these areas are either underlain by bedrock or by older Pleistocene-age fluvial and marine sediments deposited during interglacial periods, generally during the Illinoian period (Stumm, 2001; Stumm and others, 2002, 2004, 2024).

Glacial sediments comprise the surficial upper glacial aquifer, which has been an historically significant aquifer in western Long Island and remains an important source of water on eastern Long Island. The underlying Magothy aquifer and the contiguous Jameco aquifer in western Long Island make up an important aquifer that is the primary source of water in Nassau and Suffolk Counties. Two extensive fine-grained hydrogeologic units—the Gardiners clay and Raritan confining units—have hydraulic conductivity values several orders of magnitude lower than surrounding aquifers, and where present, the two confining units separate the groundwater flow system into three major aquifers: the Lloyd, the Jameco-Magothy, and the upper glacial aquifers (fig. 2; Franke and Cohen, 1972).

The Gardiners clay unit restricts vertical flow and creates confining conditions between the upper glacial aquifer and the Jameco and Magothy aquifers, predominately along the southern shore of the island. The Raritan confining unit restricts vertical flow between the Jameco and Magothy aquifers and confines the Lloyd aquifer in most areas of the island. The crystalline bedrock underlying the unconsolidated deposits is much less permeable than the overlying deposits, and the bedrock surface is considered the lower extent of the aquifer system (Smolensky and others, 1989).

The major hydrogeologic units (and their geologic equivalents) in descending order, are the upper glacial aquifer, Gardiners clay unit (Gardiners Clay), Jameco aquifer (Jameco Gravel), Monmouth greensand unit (Monmouth Group), Magothy aquifer (Magothy Formation and Matawan Group, undifferentiated), Raritan confining unit (clay member of the Raritan Formation), and the Lloyd aquifer (Lloyd Sand Member of the Raritan Formation; Smolensky and others, 1989). In addition to these major units, other Pleistocene-age units—the North Shore aquifer and North Shore confining unit—underlie Wisconsinan-age glacial sediments in some areas where Cretaceous-age units are absent (Stumm, 2001), and there are local confining units within the upper glacial aquifer (Doriski and Wilde-Katz, 1982; Krulikas and Koszalka, 1982; Schubert and others, 2004).

The Long Island aquifer system is a freshwater aquifer system that is bounded below by relatively impermeable bedrock, above by the water table, and laterally by saline groundwater. The position of the freshwater/saltwater interface in the largely unconfined upper part of the aquifer system generally represents a hydrostatic balance between freshwater and denser saltwater and likely is close to the shoreline (fig. 2). The freshwater/saltwater interface in the deep, confined Lloyd aquifer can be displaced seaward of the shoreline beneath the Raritan confining unit.

Precipitation, which is the sole source of natural recharge to the Long Island aquifer system, averaged about 47 inches per year (in/yr) ranging from about 28 in/yr to about 56 in/yr from 1900 to 2019 (Finkelstein and others, 2022). On average, about half of the precipitation on Long Island reaches the water table as aquifer recharge. The upper glacial aquifer is unconfined, as are the underlying Magothy and Jameco aquifers where the Gardiners clay is absent; the water table is a free surface that changes in response to changes in recharge. Water table altitudes exceed 60 ft in two areas, to the east and west of major surface-water drainages in the central part of the island (fig. 1). Areas of locally high water table altitudes also occur in parts of New York City, on necks and peninsulas in northern Nassau County, and on the North and South Forks in eastern Suffolk County.

Water levels and streamflows vary naturally with time in response to changes in hydrologic stresses, particularly recharge from precipitation (fig. 3). The correlation between natural recharge and hydrologic conditions (water levels and streamflows) is most pronounced in the unconfined aquifers and in generally undeveloped areas, such as eastern Suffolk County. The lowest water levels were recorded in well S5517 in eastern Suffolk County in the mid-1960s, which coincides with the largest drought for the period of record (1900 to 2019 for the purposes of this report); the highest water levels were observed in the early 1980s (fig. 3B), coinciding with a period of high precipitation and recharge (fig. 3A).

Long Island is surrounded by saltwater from the Atlantic Ocean to the south and Long Island Sound to the north. The mainland of the island encompasses the area to the west of North and South Forks, which are separated by the Peconic and Gardiners Bays (fig. 1). The northwestern shore of the island has numerous peninsulas (or necks) with small intervening embayments, whereas the northeastern shore of Long Island generally is smooth, with few bays or coastal landforms. The southern shore of Long Island is separated by Great South Bay from Fire Island, the barrier island that lies off the southern coast mainland Long Island. Most surface drainage features on Long Island are along the southern shore. The largest surface drainages—the Carmans, Connetquot, Nissequogue, and Peconic Rivers—are in the central and eastern parts of the island. There are few natural ponds on the island, the largest of which is Lake Ronkonkoma in the central part of the island.

Long Island received, on average, about 2,600 cubic feet per second (ft³/s) of aquifer recharge from precipitation from 2005 to 2015 (Walter and others, 2020). Groundwater flows away from regional, inland groundwater divides toward streams and coastal receiving waters; these divides generally trend east-west on the main body of the island and extend onto the North and South Forks. About 28 percent of groundwater discharged to streams and fresh wetlands, and little less than half (about 46 percent) discharged to salt marshes and coastal receiving waters.

Groundwater is the sole source of drinking water for the residents of Nassau and Suffolk Counties. An annual average of about 660 ft³/s of groundwater, or about 20 percent of total recharge, was withdrawn for public, agricultural, and industrial uses from 2005 to 2015 (Walter and others, 2020). Most of the water that was withdrawn during this period (about 60 percent) was returned to the aquifer system as anthropogenic recharge.

Historical Background and Human Effects on the Aquifer System

Long Island was a generally densely populated island with a population of 8.1 million people in 2020 (U.S. Census Bureau, 2021a). Most of the population (about 64 percent) resided within New York City, in Kings and Queens Counties; the remainder resided east of New York City, in Nassau and Suffolk Counties. The first census of population on Long Island showed that it had a total population of about 37,000 people in 1790, with about three quarters of the population residing in Nassau and Suffolk Counties. The population of Long Island increased to about 1.5 million people by 1900, generally as part of the industrial revolution and rapid urbanization in and around New York City in the 19th century (fig. 4A). More than 90 percent of the Long Island population resided in New York City, in Kings and Queens Counties, in 1900.

Agriculture was prevalent in Nassau and Suffolk Counties in 1900; the two rural counties combined had only about 9 percent of the total population of the State. Population increased steadily through the first part of the 20th century, with a net increase in population in Nassau and Suffolk Counties to about 13 percent of the State total. The population in these two counties more than doubled to about 30 percent of the total population of the State between 1940 and 1960 primarily due to suburban development following the end of World War II (WWII) in 1945. The largest population increase was in Nassau County where the population increased by about 890,000 people. The population on Long Island increased throughout the remainder of the 20th century—the largest rate of increase was in Suffolk County—but decreased slightly in Kings County between 1960 and 2000. Suffolk County, which has exceeded the population in Nassau County since the mid-1980s, remains the fastest growing county on Long Island with an average rate of increase of about 5 percent per year since 2000. An increase in population affects the aquifer system in several ways, including withdrawal of groundwater for public supply, changes in the amount and distribution of recharge, and physical changes to the landscape and land-use activities.

Groundwater is the sole source of potable water for Nassau and Suffolk Counties; about 450 million gallons per day (Mgal/d) of groundwater was withdrawn from the Long Island aquifer system in 2015 (Walter and others, 2020). Water supply for Kings and Queens Counties comes from an extensive system of reservoirs in upstate New York, which began with major expansions of water infrastructure in 1917 and 1936 (Buxton and Shernoff, 1999). Groundwater was an important source of water in New York City, in Kings and Queens Counties, for much of the 20th century, before the importation of reservoir water.

Groundwater from wells supplemented surface water and ponded spring water as the source of potable water in Kings and Queens Counties in the late 19th century. Groundwater withdrawals from public-supply wells in New York City began in the early part of the 20th century and peaked in the early 1930s, when about 150 Mgal/d of groundwater was pumped from the aquifer system underlying New York City (fig. 4B). Groundwater withdrawals for public supply in Kings County ceased in 1947 because saline groundwater was discovered to have intruded (what is known as saltwater intrusion) into the aquifer system. Pumping in Queens County continued and exceeded 50 Mgal/d between the late 1940s and the early 1980s, though pumping generally moved eastward within the county due to saltwater intrusion during that period. Pumping for public supply varied annually but was generally constant at more than 50 Mgal/d between the 1910s early 1980s, and by the mid-1990s, the entirety of New York City was supplied from upstate reservoirs. Pumping of groundwater for other uses continued in Kings and Queens Counties until about 1960 and 2007, respectively.

Groundwater withdrawals for public supply in Nassau County began around 1920 and increased to about 50 Mgal/d by the mid-1940s, after which withdrawals increased rapidly to about 200 Mgal/d by 1970 (fig. 4B). The increase in pumping corresponds to the period of rapid suburbanization following WWII when the population of Nassau County more than doubled (fig. 4A). Pumping in Nassau County generally has been more constant since 1970, averaging about 200 Mgal/d (fig. 4B). Pumping from public-supply wells in Suffolk County began in the mid-1950s and increased at a generally steady rate to more than 100 Mgal/d by 1980. Groundwater withdrawals from public-supply wells nearly doubled to about 200 Mgal/d by 1990, likely due to the transition from onsite, private wells to large capacity, public-supply wells during that time. Suffolk County has had the largest groundwater withdrawals on Long Island since 2000; pumping in the county generally has fluctuated between 200 and 250 Mgal/d during that time.

Human activities also have affected the aquifer system through urbanization, changes to the landscape, and the development of water infrastructure. Monti and others (2024) estimated spatially variable historic land use on Long Island for the period 1900-2019 using land cover projections from Sohl and others (2014) and Sohl and others (2018). About 350 mi² of the land area of Long Island was considered developed in 1900, which represented about 27 percent of the total land area. About 22 percent of the landscape was used for agriculture, either as crop land or pasture, and the remaining 50 percent was undeveloped land, generally maritime forests of pitch pine and red oak (fig. 5A). By 2019, a total of about 880 mi² of Long Island was considered developed, or about 69 percent of total land area. Undeveloped land and land used for agriculture represented a total of about 5 and 26 percent, respectively, and was generally in eastern Suffolk County (fig. 5B).

The history of development on Long Island since 1900 varies by county. High-density residential development, defined as a population density exceeding about 11,200 people per square mile, in 1900 was generally limited to densely populated areas of Kings County (fig. 5A ; Finkelstein and others, 2022). Remaining developed areas generally had low-density development (less than about 5,600 people per square mile). About 83 percent of land in Kings County was considered developed in 1900; most (about 70 percent) was considered high density development (fig. 5A). About 77 percent of land in Queens County was considered developed, most of which was considered low density development. The amount of developed land increased slightly since 1900 (fig. 4D) and in 2019, both counties were about 95 percent developed (fig. 5B), most of which (about 71 percent) is considered high-density development.

About 50 percent of Nassau County was considered developed, and about 20 percent of the county was used for agriculture in 1900 (fig. 5A). The largest rate of development was generally between the late 1960s and early 1990s when developed land increased from about 60 percent to about 85 percent (fig. 4D) due to a rapid suburbanization and associated conversion of forested and agricultural land to low- and medium-density residential development (fig. 5B). About 86 percent of Nassau County was developed in 2019, with undeveloped areas generally in the northeastern part of the county (fig. 5B). About 51 percent of developed land in Nassau County in 2019 was considered low density residential development (fig. 5B). Medium- and high-density residential development made up about 33 and 16 percent of developed in Nassau County, respectively.

Suffolk County is the most rural county on Long Island. About 90 mi² of land in the county was considered developed in 1900, or about 10 percent of total land area (fig. 5A). About 27 percent of land in the county was used for agriculture and most land (about 62 percent) was undeveloped, likely forested, at that time (fig. 5A). Development in Suffolk County increased from less than 20 percent in the early 1960s to more than 50 percent of total land area in the early 1990s because of rapid suburbanization (fig. 4D). About 60 percent of Suffolk County was considered developed in 2019. Most of the developed land (about 85 percent) was considered low-density residential development (fig. 5B), and about 14 percent of developed land was considered medium-density residential development. Less than 1 percent of developed land in the county was high-density residential development. About 7 percent of the county was in agricultural use by 2019, primarily on the North Fork in eastern Suffolk County (fig. 5B). About 33 percent of Suffolk County remained forested in 2019, primarily in eastern Suffolk County (fig. 5B).

Natural recharge is the sole source of freshwater to the Long Island aquifer system and is affected by several characteristics of the landscape, including soil type and water capacity, land slope, and vegetative cover. Development affects recharge in different ways: impervious surfaces can impede the movement of water into the unsaturated zone and lower recharge to the water table in urbanized areas (fig. 6A), whereas the conversion of forested land to low-density residential development can increase recharge (Yang and others, 2015). Impervious surfaces generally coincide with areas with high-density residential development, including almost all of Kings and Queens Counties in New York City (fig. 4B).

Finkelstein and others (2022) estimated the average recharge on Long Island from 1900 to 2019 by use of a soil-water balance model (Westenbroek and others, 2010) that utilizes climate and spatial data. Natural recharge from 2010 to 2019 estimated using developed landscape characteristics averaged 20.5 in/yr across Long Island and varies spatially (fig. 6B). The lowest recharge rate is in Kings County where the average rate was about 6.6 in/yr, and the highest is in Suffolk County where the average recharge rate was about 23.2 in/yr. These soil-water balance estimates of recharge assume 2010–19 climate conditions. Estimated recharge using a soil-water balance model for the period before development in the 20th century (known as predevelopment), which assumes a uniformly forested landscape, averaged 19.2 in/yr, only slightly (about 1.3 in/yr) lower than the natural recharge estimated for 2010 to 2019.

The difference in natural recharge rates between developed and undeveloped landscapes varies across Long Island (fig. 6C). Natural recharge rates from 2010 to 2019 in Kings and Queens Counties (in New York City) and parts of southern Nassau County were substantially lower (exceeding a 20 in/yr deficit) than for predevelopment conditions owing to the high-density residential development and impervious surfaces in urbanized landscapes typical of that area. The average natural recharge rates in Kings and Queens Counties decreased by about 13.6 and 11.0 in/yr, respectively, due to development patterns. Conversely, recharge rates in parts of Nassau and Suffolk County were substantially higher (exceeding 8 in/yr) owing to the conversion of forested and agricultural land to low-density residential landscapes that generally are more efficient for aquifer recharge (Yang and others, 2015). Average natural recharge rates in Nassau County increased by about 1.6 in/yr from predevelopment conditions, and average recharge rates in Suffolk County increased by about 3.9 in/yr. Recharge rates estimated from the soil-water balance model in areas forested in 2019, which were generally in eastern Suffolk County, were the same for both developed and undeveloped landscapes (fig. 6C).

Most of the groundwater withdrawn from the aquifer for public supply (about 60 percent) was returned as anthropogenic recharge, either as wastewater return flow or as loss from compromised water-supply infrastructure. In 2019, about 84 percent of wastewater return flow on Long Island was discharged into sanitary sewers. Kings County has been fully sewered since the beginning of the 20^th century; Queens County was about 90 percent sewered in 1900 and fully sewered by the mid-1950s (fig. 4C). Construction of sewers began in Nassau County in the mid-1930s, and about 10 percent of the county was sewered by 1950. The extent of sewered areas in Nassau County increased rapidly between 1950 and the mid-1980s, corresponding with the period of rapid suburbanization following WWII. About 70 percent of Nassau County was sewered by the mid-1980s, after the last phase of sewer construction. The remaining unsewered areas are in the northern part of the Nassau County.

The fraction of wastewater return flow that enters sanitary sewers increased in Nassau County from less than 2 percent in the mid-1930s to more than 90 percent by 1980. Construction of sewers in Suffolk County generally began in the mid-1930s, but the total of sewered area in the county was less than 1 percent by the mid-1950s and did not exceed 5 percent until 1980 (fig. 4C). Sewer expansions in the early 1980s doubled the fraction of sewered areas to about 10 percent by the mid-1980s. In 2019, the fraction of sewered areas remained low, at about 12 percent, and a total of about 85 percent of wastewater return flow recharged the aquifer in unsewered areas (Shaffer and Runkle, 2007). About 36 percent of Long Island was sewered in 2019, mostly in the densely populated areas in New York City, southern Nassau, and southwestern Suffolk County (fig. 6A). Most wastewater (about 84 percent) was discharged into sanitary sewers; the remaining 16 percent of wastewater was discharged into onsite septic systems, primarily in central and eastern Suffolk County and along the northern shore of both Nassau and Suffolk Counties. About 90 percent of wastewater discharged into sanitary sewers was treated and discharged into surface waters; the remaining 10 percent was treated and discharged onto land, mostly in inland areas of Suffolk County. Recharge from wastewater return flow was primarily in unsewered areas but also could occur in sewered areas through leaky or compromised sewer lines.

Pumped groundwater can also recharge the aquifer from leaky water-supply infrastructure; the amount of recharge from infrastructure is a function of the density and age of water lines. Misut and Monti (1999) reported that recharge from aging or compromised water-supply infrastructure in New York City was approximately 10 percent of the total water supplied to the area.

Anthropogenic recharge refers to the combination of recharge from wastewater return flow and leaky infrastructure and depends on several factors, including population density, the location of water infrastructure, and the age and mechanisms of return flow in sewered and unsewered areas. The average rate of anthropogenic recharge on Long Island was about 4.7 in/yr and varies across Long Island (fig. 6D). Anthropogenic recharge generally is largest in Kings and Queens Counties where high-density residential development results in substantial recharge from leaky infrastructure and also in central Suffolk County where extensive areas of low and medium density residential areas are not sewered and wastewater return flow recharges the aquifer system. Anthropogenic recharge is lower in rural areas of eastern Suffolk County and in sewered areas in Nassau County. Anthropogenic recharge accounts for more than 90 percent of total recharge in parts of Kings County. About 51 percent of total recharge in Kings and Queens Counties and about 13 percent in Nassau and Suffolk Counties is from anthropogenic sources.

Withdrawal of groundwater can affect the aquifer system by lowering the water table, and potentiometric surfaces in confined aquifers and depleting streamflows. The drawdown of the water levels from pumping has the potential to reverse hydraulic gradients and induce saltwater intrusion into freshwater portions of aquifers. Pumping in the northern part of Kings County and to the south in the Flatbush franchise resulted in a cone of depression as early as the mid-1930s (fig. 7A; Buxton and Shernoff, 1999). Water table altitudes in this area were as low as 30 ft below mean sea level, representing a drawdown of about 40 ft from nonpumping conditions. Although pumping in the county ended in 1947 in response to saltwater intrusion, water levels remained below sea level until the early 1960s. Water table altitudes below sea level were first observed in the western part of Queens County in the late 1930s due to pumping in the Woodhaven franchise (fig. 7B). The water table decreased to an altitude of 15 ft below mean sea level by the early 1960s, representing a drawdown of about 30 ft.

Water table altitudes below sea level were first observed in central Queens County by the early 1970s (fig. 7C; Buxton and Shernoff, 1999). Pumping in western Queens County ended in 1974 due to saltwater intrusion, after which water levels recovered to near sea level. Continued pumping from the Jamaica franchise in central Queens County resulted in water table altitudes 15 ft below sea level (fig. 7C; Buxton and Shernoff, 1999), which represents a drawdown of about 50 ft from nonpumping conditions. Pumping in central Queens County declined following an expansion of water-supply infrastructure to allow for the importation of water from upstate New York throughout Kings and Queens Counties and to preserve fresh groundwater resources in central Queens County.

Historical chloride data indicate that saltwater intrusion was observed in Kings County before 1920, and currently [2022] large areas along the coast in the aquifers underlying the county are intruded with saltwater (Stumm and others, 2024). Saltwater intrusion was observed along the entire southern coast of Queens County north of Jamaica Bay before 1960, and aquifers underlying the area remain affected by saltwater intrusion. Saltwater intrusion also has been observed in aquifers underlying southwestern Nassau County since the 1940s, and in some areas along the northern coast of the county since the 1960s. Before 1990 along the northern coast of Queens County, saltwater intrusion was first observed primarily in the Flushing area (Stumm and others, 2024). Parts of the Lloyd, Magothy, and upper glacial aquifers in New York City and western Nassau County remain intruded by saline groundwater, though large-scale pumping in New York City ceased by the early 1980s. The Lloyd and Magothy aquifers in Kings County remain intruded along the western and southern shores. In Queens County, groundwater with chloride concentrations exceeding 1,000 milligrams per liter (mg/L) in the Lloyd aquifer extends northward about 1 mi inland from Jamaica Bay, as well as southward by about 0.5 mi near Flushing Bay. The Magothy aquifer in Queens County also is intruded along the southern and northern coasts. Parts of the Magothy and Lloyd aquifers also are intruded beneath peninsulas along the northern shore of Nassau County due to historical and ongoing groundwater withdrawals (Stumm and others, 2024).

Anthropogenic perturbation of the western Long Island aquifer system by groundwater withdrawals and changes in recharge also is reflected in time-varying water levels and streamflows (fig. 8). Water levels and streamflows are strongly correlated with precipitation and recharge under natural conditions (fig. 3). Large groundwater withdrawals and the subsequent cessation of those withdrawals result in long-term trends in both water levels and streamflows that are not strongly correlated with recharge (fig. 8). Wells K1235 and K1263 in Kings County show an increase in water levels starting in the late 1940s (fig. 8B); this increase in water levels coincides with the cessation of pumping in that area (fig. 4B). Water levels in the wells increased by about 15 ft between the late 1940s and 1970 when recharge fluctuated but did not appear correlated to water level fluctuations (fig. 8A). Water levels in well Q1252 showed a steady decrease from about 27 ft in the early 1950s to 2 ft below sea level by 1970. The decline in water levels coincided with increases in withdrawals in Queens County to meet water demand following the cessation of pumping and an expansion of sewer systems to the west (fig. 4D). Groundwater withdrawals also affect streamflows. Flow in Valley Stream in southwestern Nassau County (fig. 1) decreased from between 5 and 10 ft³/s in the late 1950s to less than 2 ft³/s by the early 1960s (fig. 8B). This decrease in flows generally follows an increase in pumping and an expansion of sewer systems in western Nassau County starting in the mid-1950s (fig. 4D).

Human activity has also affected the hydrologic system by altering natural hydrologic features, including the impounding of freshwater streams and the filling of coastal marshes and streams. A total of about 30 mi² of coastal marshes and streams was filled before 1920. Most infilled areas are former coastal marshes along Jamaica Bay in southeast Kings and southern Queens Counties and along Flushing Bay and Newtown Creek in northwestern Queens County (fig. 6A). Additional infilled areas include freshwater bodies in central Queens County. Infilling of natural waters can affect groundwater discharge patterns, and filled areas near the coast that have been subsequentially developed could be at a higher risk of groundwater inundation and sea level rise.

Physiography

The surface-water hydrography and geometry of the coast were determined from geographic information system (GIS) linework digitized from 1:24,000-scale topographic quadrangles (U.S. Geological Survey, 2020a). Most surface drainages are along the southern shore of the island, and many of those streams have artificial impoundments. The largest surface drainages—the Connetquot, Nissequogue, and Peconic Rivers—are in the central and eastern parts of the island (fig. 1). There are few natural ponds on the island. The northern shore of Nassau and western Suffolk Counties generally has few streams with large peninsulas, referred to locally as necks, separated by small embayments. There are numerous coastal streams along the southern shore, and a series of barrier islands are south of the western and central parts of the Long Island mainland.

The altitudes of land surface and the seabed were derived from a 1-m (3.3-ft) topobathymetric digital elevation model of the coastal regions of Massachusetts, Rhode Island, Connecticut, and New York (Danielson and Haines, 2017). The dataset is a collaboration between the USGS and the National Oceanic and Atmospheric Administration and combines data from 321 sources to create a seamless, internally consistent representation of land and seabed altitudes at a spatial resolution of 1 m (3.3 ft). These data were upscaled and mapped to the regional model grid; model cells have a horizontal discretization of 500 ft, and there are about 23,100 individual altitude measurements within each model cell. The mean and minimum values of these measurements were computed for each model cell to a distance 3 mi seaward of the coast (fig. 9). Mean land-surface altitudes range from sea level to more than 370 ft above sea level and exceed 200 ft above sea level in the north-central part of the island, in isolated areas in the eastern part of the Long Island mainland, and on the South Fork. Land-surface altitudes are highest in areas underlain by glacial moraines that have an east-west trend in the northern part of the Long Island mainland and extend onto the North and South Forks.

Topography along the northern shore of the island generally is steep, particularly in areas where moraines are adjacent to the shore. The topography to the south of the moraines is gently sloping to the southern shore of the island, generally corresponding to glacial outwash plains (fig. 9). Mean seabed altitudes range from sea level to more than 350 ft below sea level. The lowest seabed altitudes (deepest water) are in offshore areas of Long Island Sound, along the North Fork. The highest seabed altitudes (shallowest water) generally are in estuaries and bays along the southern shore of the island. The northern shore of Long Island generally is characterized by steep coastlines resulting from coastal erosion, whereas the southern shore is characterized by gently sloping topography with numerous small embayments.

The highest water table altitudes in 2010 exceeded 70 ft in two areas on the mainland of the island (fig. 1; Monti and others, 2013). The mean thickness of the unsaturated zone thickness, defined as the difference between land surface and the water table, is about 47 ft, and the unsaturated zone can be as thick as 300 ft in some areas underlain by moraines. The thickness of the unsaturated zone in 2010 was 10 ft or less across about 250 mi², generally along the southern coast, near streams, and on barrier islands (fig. 9).

Hydrogeology

New and existing geologic data were collected, compiled, and analyzed to better define the hydrogeology of the Long Island aquifer system for synthesis into a groundwater flow model. The sources of the data include historical literature, compilation of existing data (lithologic and borehole geophysical logs), and collection of geologic data from ongoing (2017–2023) field efforts. These data were used to develop a hydrogeologic framework of the Long Island aquifer system and to estimate the water-transmitting properties of sediments of the major unconfined aquifers, including the upper glacial, Jameco, and Magothy aquifers.

Previous investigations of the extents and surface altitudes of major hydrogeologic units were used to develop a three-dimensional model of the hydrogeologic framework of the Long Island aquifer system (Walter and others, 2020). This information included maps of the surficial (Wisconsinan) glacial geology of Long Island (Cadwell and Muller, 1986; Cadwell, 1989), locally significant Wisconsinan confining units (Doriski and Wilde-Katz, 1982; Krulikas and Koszalka, 1982), pre-Wisconsinan Pleistocene marine, lacustrine, and fluvial units (Stumm, 2001; Stumm and others, 2002, 2004; Schubert and others, 2004), and the surfaces and extents of the bedrock and overlying Cretaceous formations (Smolensky and others, 1989).

A hydrogeologic framework was developed from existing lithologic and geophysical data and new data collected as part of an ongoing drilling and geologic mapping effort as part of the Long Island Groundwater Sustainability Project (LISUS; U.S. Geological Survey, 2024a). The field effort, to date [2024], focused on the western part of Long Island, including Kings, Queens, and Nassau Counties. This effort used lithologic and geophysical data from 643 existing wells of varying depths across Long Island (fig. 10). These data were augmented with information from 23 new (2017–22) wells, including 18 boreholes drilled to bedrock as part of the project and an additional 5 wells drilled as part of other investigations. The results of this investigation, including the extents and surfaces of major hydrogeologic units and the extent of saltwater intrusion, are documented in detail in Stumm and others (2024). The hydrogeologic framework also was updated with the results of mapping of the bedrock surface of New York City (DeMott and others, 2023).

Hydraulic properties of aquifer sediments of importance to groundwater flow include horizontal and vertical hydraulic conductivity, specific yield in unconfined aquifers, specific storage in confined aquifers, and sediment porosity and dispersivity. These properties vary by hydrogeologic unit and within each unit. Direct measurements of hydraulic properties are focused primarily near major pumping centers, and consequently, field measurements of hydraulic properties are concentrated in areas of large groundwater withdrawals and for those particular aquifer units with the most use (McClymonds and Franke, 1972; Franke and Getzen, 1976; Lindner and Reilly, 1983; Prince and Schneider, 1989; Stumm, 2001; Stumm and others, 2002, 2004, 2024; Williams and others, 2020); as such, hydraulic property data are relatively sparse compared with the apparent heterogeneity of the regional hydrogeologic units.

Estimates of hydraulic properties made in previous numerical model investigations include Buxton and Smolensky (1999), Kontis (1999), Misut and Monti (1999), Misut and others (2004), Schubert and others (2004), and Monti and others (2009). Buxton and Smolensky (1999) provided a summary of the previous investigations of the hydraulic properties of the aquifer system. That synthesis of previous work represents the most comprehensive summary to date of the existing information on hydraulic properties of the major hydrogeologic units of the Long Island aquifer system. Stumm and others (2024) present a summary of the water-transmitting properties of major hydrogeologic units in western Long Island as determined from slug tests and nuclear magnetic resonance logs.

The relation between sediment characteristics and measured hydraulic conductivity, as compiled from previous investigations, was used to develop a three-dimensional texture model of hydraulic conductivity within the upper glacial, Jameco, and Magothy aquifers—the principal aquifers on Long Island—using lithologic logs from 1,769 wells across Long Island (Walter and Finkelstein, 2020). An augmented network of 1,846 wells was used to develop an updated texture model of those three principal aquifers for this investigation (fig. 10). The texture model defines a three-dimensional distribution of hydraulic conductivity that represents the heterogeneity of sediments in the major aquifers. The workflow used to develop the three-dimensional texture model from borehole data is presented in Walter and Finkelstein (2020) and consists of four steps:

The hydraulic conductivity of deep hydrogeologic units, including the Raritan confining unit and Lloyd aquifer, were more generalized, and were determined from previous investigations owing to the paucity of lithologic data in those parts of the aquifer. Specific yield, specific storage, and porosity of aquifer sediments in the major hydrogeologic units also were estimated from previous investigations.

Natural and Anthropogenic Recharge

Long Island has a temperate climate that is moderated by its proximity to ocean waters and typified by warm, humid summers and cool winters. Precipitation is the sole natural source of water to the Long Island aquifer system, with about half of it reaching the water table as aquifer recharge, and the remainder lost to evapotranspiration and surface runoff. Aquifer recharge is a function of climatological conditions (precipitation and temperature) and landscape characteristics (vegetation, soil properties, and land use); variations in these factors affect the rates and distribution of recharge. Total recharge to the aquifer system includes natural recharge from precipitation and from anthropogenic sources, such as wastewater return flow and leakage from water infrastructure.

Finkelstein and others (2022) used a soil-water balance (SWB) model code to estimate spatially variable, natural recharge for Long Island from 1900 to 2019. The SWB model developed by Westenbroek and others (2010) is based on a modified version of the Thornthwaite-Mather analytical method (Thornthwaite and Mather, 1957) that incorporates land slope, soil properties, land use, and daily climate data to account for processes that occur as water moves through unsaturated soils and sediments to the water table.

Precipitation on Long Island from 1900 to 2019 ranged from 28.3 in/yr in 1965, during a period of extreme drought, to 65.2 in/yr in 1983 and averaged about 47 in/yr. Recharge is correlated with precipitation and has a similar temporal pattern (fig. 15A). The estimated fraction of precipitation that recharges the aquifer at the water table was between about 25 and 57 percent and averaged about 42 percent from 1900 to 2019 (fig. 15B).

Estimated recharge from precipitation ranged from about 8.9 to about 34.7 in/yr and averaged about 19.8 in/yr during that period. The fraction of precipitation recharging the aquifer between 1900 and 2019 generally was lowest in Kings County where the average was about 16 percent, and highest in Suffolk County where the average was about 46 percent (fig. 15B). The fraction of precipitation recharging the aquifer decreased in the early 20th century in New York City (Kings and Queens Counties) due to urbanization and the expansion of impervious surfaces, and currently [2023] is 14 and 18 percent, respectively (fig. 15B). Kings and Queens Counties are the most urbanized and densely populated part of Long Island; impervious surfaces can intercept precipitation, resulting in more surface runoff and evaporation and less recharge at the water table (fig. 15C). The fraction of precipitation recharging the aquifer was similar in Nassau and Suffolk Counties until the mid-1980s when the fraction generally varied between 30 and 60 percent and averaged 44 percent in both areas (fig. 15B). The fraction of precipitation recharging the aquifer was higher in Suffolk County after the mid-1980s when the average fraction in Nassau and Suffolk Counties was about 41 and 49 percent, respectively. Suffolk County generally is characterized by large areas of low-density residential land use (fig. 5B), which can result in higher recharge rates (fig. 15B).

Recharge from precipitation varies spatially as a function of the distribution of precipitation and natural and human-influenced land characteristics. Recharge in 2019, when precipitation islandwide was about 54 in/yr, generally was less than 10 in/yr in parts of New York City and more than 30 in/yr in parts of north-central Suffolk County and on the South Fork (fig. 16A). Undeveloped land had recharge rates of about 25 in/yr. Recharge generally is lower in urbanized areas with impervious surfaces than in undeveloped areas and higher than undeveloped areas in areas with low-density residential development (Yang and others, 2015). Recharge to the aquifer in Kings and Queens Counties averaged 7.4 and 11.2 in/yr, respectively, during the period from 1900 to 2019. The recharge rates in Nassau and Suffolk County were substantially higher than in New York City and averaged 20.4 and 21.6 in/yr, respectively, during the same period (Finkelstein and others, 2022).

The effects of human-influenced landscapes on recharge were determined by use of a modified SWB model (Finkelstein and others, 2022); the same climate and natural landscape data were used in combination with open forested (predevelopment) landscape characteristics, and the difference represents the effects of human landscapes on aquifer recharge. Predevelopment recharge rates between 1900 and 2019 were similar to estimated recharge at an islandwide scale, ranging from about 7.3 in/yr to about 34.5 in/yr and averaging about 18.5 in/yr (fig. 15A). However, the assumption of natural landscape characteristics resulted in a more uniform distribution of recharge that would generally be a function of spatially variable precipitation and temperature. The average recharge rates in all four counties were between 17.8 and 18.8 in/yr from 1900 to 2019 (Finkelstein and others, 2022). The difference in recharge between predevelopment and that estimated using developed landscape characteristics ranged from about −25 in/yr, indicating a deficit of recharge, to about 10.9 in/yr, indicating enhanced recharge resulting from development (fig. 16B). Recharge rates in Suffolk County generally were enhanced because of development, which is characterized by low-density residential development in central and eastern Long Island (fig. 5B). Recharge rates in natural, forested landscapes are unchanged (fig. 16B). Areas with a deficit of recharge from development generally are in Kings and Queens Counties and parts of Nassau County.

The largest recharge deficits from impervious surfaces are in the urbanized areas of New York City and southern Nassau County; however, much of the recharge potentially lost to impervious surfaces in Nassau County can recharge the aquifer through sumps, storm drains, and a network of more than 600 recharge basins. The deficit of recharge from impervious surfaces in urbanized areas was considered to be a potential source of recharge to the aquifer through stormwater infrastructure, although the details on the recharge mechanisms are unavailable. This potential component of recharge is referred to as rejected recharge (fig. 15C). The amount of potential rejected recharge from redirected impervious runoff ranged from 0 to about 25 in/yr in parts of New York City (fig. 16B) and could be a significant portion of recharge in urbanized areas. Recharge generally increased by between 10 and 30 percent as a result of changes in undeveloped land use in Nassau and Suffolk Counties, likely as a result of conversion of undeveloped land to low-density residential development (fig. 15C). Conversely, the fraction of recharge potentially lost to impervious surfaces in Kings and Queens Counties increased steadily between 1900 and about 1960 and was about 60 and 50 percent, respectively, in 2019. This water represents an additional source of potential recharge through stormwater infrastructure.

Long Island is a densely populated region with natural recharge enhanced by recharge from human sources. This enhanced recharge is referred to as anthropogenic recharge and includes recharge from wastewater return flow and from leaky infrastructure. These additional components vary spatially as a function of the type and age of water infrastructure and population density during time. Water that recharges the aquifer after human use is referred to as wastewater return flow and depends on population density and the location of sanitary sewers (fig. 16D).

Wastewater return flow from about 84 percent of Long Island’s population is discharged in areas served by sanitary sewers, which include all of New York City and most (about 90 percent) of Nassau County. About 23 percent of the population of Suffolk County is served by sanitary sewers, mainly in the southwestern part of the county (fig. 16D). Wastewater return flow in areas served by sewers is treated and generally discharged into surface waters, effectively removing it from the hydrologic system. However, a small fraction of this return flow likely recharges the aquifer in areas with aging infrastructure.

Wastewater return flow in unsewered areas, primarily in Suffolk County, is discharged into onsite septic systems and recharges the aquifer at the water table. The total amount of return flow available for recharge of the aquifer was estimated from spatially variable population density and an assumed per capita water use. Potential recharge from wastewater return flow was calculated from the number of people in each cell in a regular grid coincident with the numerical model and a per capita water use of 70 gallons per day (Walter and others, 2020).

The potential wastewater return flow was calculated for each year from 1900 to 2019 from U.S. Census Bureau data spatially distributed across the island by census block and building footprint data (U.S. Census Bureau, 2021b). The result is a gridded dataset of recharge values of the total potential recharge from wastewater across the entirety of Long Island that reflects population density (fig. 16D). Potential wastewater return flow needed to be adjusted by assumed loss terms to better estimate recharge likely at the water table. It was assumed that there was a consumptive loss of 10 percent before the discharge of wastewater (Shaffer and Runkle, 2007). It also was assumed that 5 percent of wastewater discharged into sewers, such as New York City, most of Nassau County, and southwestern Suffolk County (fig. 16C), leaked into the aquifer through aging or compromised infrastructure and that 90 percent of discharged wastewater in unsewered areas recharged the water table. The resulting distribution of recharge from wastewater return flow reflects both population density and wastewater infrastructure (fig. 16E). Wastewater return flow in populated areas was lowest in sewered areas of Nassau County where return flow generally was less than 2 in/yr but was slightly higher in sewered areas in New York City because of a higher population density and water use. Recharge from wastewater exceeded 20 in/yr in parts of Suffolk County in areas with low and medium density residential development not served by sanitary sewers.

Almost all (more than 99 percent) of the population of Long Island was served by public water supply; private water supplies were limited to small areas of eastern Suffolk County, the South Fork, and the northern shore of Nassau County (fig. 16F). Water from leaky water-supply infrastructure was an additional source of potential recharge to the aquifer system. Recharge from leaky infrastructure is a function of the spatial distribution of water-supply infrastructure and the amount of water pumped and distributed within each county and varies in space and time. The distribution of public-supply lines was estimated from the length of roads in populated areas, expressed as the fraction of road length in each cell to the total length of roads in the county containing the cell and the assumption that water lines and road length are correlated in developed areas. Road length density was largest in Kings and Queens County where population density was largest (fig. 16C). Areas with large road length density were also in densely populated areas in southern and central parts of Nassau County and in western and central parts of Suffolk County (fig. 16F).

The amount of potential recharge from leaky infrastructure in each cell was calculated by apportioning the total amount of pumping in the county containing the cell by the estimated fraction of road length. The potential recharge from leaky infrastructure was calculated for each year from 1900 to 2019. Pumping data from Nassau and Suffolk Counties was obtained from the NYSDEC (New York Department of Environmental Conservation, 2024), augmented with additional sources discussed in the “Water Use” section of this report.

Kings and Queens Counties imported water from reservoirs in upstate New York. Water use rates to calculate recharge from leaky infrastructure in New York City were estimated from population density; a per capita water use of 70 gallons per day per person was used because data on the amount of water imported into the two counties were insufficient or unavailable. The potential recharge rate was adjusted assuming a loss of 10 percent from water supply infrastructure. Potential recharge from leaky infrastructure generally reflected patterns of population and development and was largest in New York City (fig. 16G). About 700 Mgal/d of water was imported from a reservoir system in upstate New York and, assuming a 10 percent loss of water from the water distribution system, may result in about 70 Mgal/d of additional recharge to the water table in Kings and Queens Counties (Misut and Monti, 1999). The potential for recharge from leaky infrastructure in 2019 exceeded 10 in/yr in large parts of Kings County, which was the most urbanized and densely populated part of Long Island (fig. 16G). Potential recharge from leaky infrastructure exceeded 5 in/yr in most parts of Queens County and 3 in/yr in populated areas in parts of Nassau and Suffolk Counties.

Total recharge to the aquifer system is the sum of natural recharge and the two components of anthropogenic recharge. The fraction of anthropogenic recharge large in Kings and Queens Counties was large, and it increased steadily between 1900 and 1940. The fraction of total recharge derived from anthropogenic recharge was highest in Kings County where it averaged about 58 percent from 1900 to 2019 (fig. 15D). Anthropogenic recharge in Queens and Nassau Counties was, on average, about 30.9 and 14.2 percent of total recharge, respectively, during the same period. The fraction of total recharge from anthropogenic sources is lowest in Suffolk County, where it averaged 5.7 percent. The map of total recharge for 2019 (fig. 16H) reflects the combined spatial variability of the three components of recharge.

The lowest total recharge in 2019, about 15.5 in/yr, was in Queens County (fig. 16G) where natural recharge generally was low in urbanized areas (fig. 16A). Kings County had a low natural recharge rate (fig. 16A) but a high recharge rate from leaky infrastructure, resulting in a high total recharge rate (about 17.1 in/yr; fig. 16G). Total recharge rates were similar in Nassau County (23.8 in/yr) and Suffolk County (23.2 in/yr; fig. 16H). Natural recharge rates also were similar in the two counties. Recharge from leaky infrastructure was larger in developed areas of Nassau County, whereas recharge from wastewater return flow was larger in Suffolk County, resulting in similar total recharge rates. The highest total recharge rates generally were in central Suffolk County in unsewered areas with low to medium density residential development (fig. 16H). Total recharge from natural and anthropogenic sources was augmented in urbanized impervious areas by potential recharge from runoff through stormwater infrastructure, primarily in New York City (fig. 16B). Total recharge to the aquifer, therefore, was the sum of natural recharge from precipitation, recharge from wastewater return flow and leaky infrastructure, and recharge from stormwater runoff. Loss terms for potential anthropogenic recharge components—wastewater return flow, leaky infrastructure, and rejected recharge—were adjusted during model calibration to better match observed hydrologic conditions.

Water Use

Groundwater was the sole source of drinking water for Nassau and Suffolk Counties in 2019 and was an important source of drinking water for Kings and Queens Counties before the mid-1990s (Buxton and Shernoff, 1999). Data on the amount and location of groundwater withdrawals from 1900 to 2019 were compiled from several sources to develop a detailed, continuous record of the location, depth, volumetric rate, and purpose of groundwater withdrawals from the Long Island aquifer system. Thompson and Leggette (1936) compiled water use data for Kings and Queens Counties for the early part of the 20th century. Johnson and Waterman (1952) updated and extended public-supply water use in New York City from 1906 to 1952. Kilburn (1981) reported water-use data for Nassau County from 1920 to 1977. Chu and others (1997) compiled groundwater-withdrawal data from western Long Island from 1921 to 1995. Public-supply pumping data were compiled and augmented with industrial and commercial pumping data from 1906 to 2015 for New York City (Matthew Gamache, CDM Smith, written commun., 2017). Pumping data from Nassau County from 1980 to 2014 were compiled by the Nassau County Department of Public Works (NCDPW; Mike Flaherty, NCDPW, written commun., 2018). The Suffolk County Water Authority (SCWA) compiled monthly pumping data in Suffolk County from 1988 to 2015. (Daniel DeSalvo, Suffolk County Water Authority, written communication, February 24, 2011).

Groundwater-withdrawal data also were compiled from water purveyors, as scanned data, from 1920 to 2019. Data on public-supply, commercial, industrial, golf course, and agricultural pumping, including withdrawals and completion reports, were compiled by the NYSDEC from 1985 to 2015 (Jennifer Pilewski, NYSDEC, written communs., February 2017 to June 2022). NYSDEC data are available through the DEC Info Locator interactive map (New York State Department of Environmental Conservation, 2024). Well-construction data, including location, screen depth, and age were compiled from published sources; these data were augmented with data from other sources, including the USGS National Water Information System (NWIS) database (U.S. Geological Survey, 2020b). Georeferenced well locations were adjusted as needed using aerial photographs, and well-screen depths were estimated from well depth when that information was unavailable. These data on pumping rates and screen locations within the aquifer system were used to develop 86,939 average annual pumping records for a total of 2,942 wells from 1900 to 2019 (fig. 17). Gaps in data and possible missing pumping records from sequential years in some wells, particularly in the early part of the 20^th century, were filled by linear interpolation between the previous and subsequent annual pumping estimates.

Groundwater has been withdrawn from wells, at different times, throughout Long Island since about 1906; the period of operation and amount withdrawn has varied historically among wells as a function of local water-supply demand, water quality, and the importation of water from surface supplies in upstate New York. A total of 1,900 wells have been in operation at different times since 2000, and another 1,034 wells were abandoned before 2000 (fig. 17). Wells abandoned before 2000 primarily were in New York City and coastal parts of Nassau County and were abandoned as the result of saltwater intrusion into the freshwater aquifer system. Wells abandoned in the interior parts of Nassau and Suffolk Counties generally were abandoned as a result of contamination of the upper glacial and shallow parts of the Magothy aquifers from human activities.

Public-supply pumping in Kings County ceased in 1947 because of saltwater intrusion near major pumping centers but continued in Queens County until saltwater intrusion caused a cessation of public-supply withdrawals in the mid-1980s (fig. 18A). Groundwater withdrawals in Nassau County increased substantially after the mid-1940s because of rapid suburbanization and became generally constant after about 1980. Pumping for public supply in Suffolk County increased rapidly after the late 1960s because of suburbanization and the conversion of domestic water supplies to public service areas. Pumping in Suffolk County generally became more constant by the late 1990s. The largest amount of groundwater withdrawn from the Long Island aquifer system was in 1999 when about 475 Mgal/d of water was withdrawn from 1,184 wells in Nassau and Suffolk Counties (fig. 18A). In 2019, about 390 Mgal/d of groundwater was withdrawn from 1,115 wells.

Most groundwater withdrawn from Long Island aquifers before the early 1920s was for industrial or commercial uses in New York City; public supply was the largest use for groundwater after that period. About 88.0 percent of groundwater withdrawn from the aquifer system was used for public supply since pumping began in about 1904 (fig. 18B). About 10.8 percent of groundwater withdrawn was for commercial and industrial uses, with the remainder (about 1 percent) for other uses, such as agriculture or remedial systems. Pumping from private wells in 2019 accounted for less than 1 percent and was limited to rural parts of eastern Suffolk County (fig. 16F). The upper glacial aquifer was the surficial unconfined aquifer across Long Island and was the source of most (54.5 percent) groundwater withdrawn from the aquifer system since pumping began near the start of the 20th century (fig. 18C). About 28.0 percent was withdrawn from the underlying Magothy aquifer, and about 12.6 percent was withdrawn from the contiguous Jameco aquifer. The Lloyd aquifer, which is the deepest aquifer and is confined throughout its extent, supplied about 1.1 percent of the total groundwater withdrawn from the aquifer system. The North Shore aquifer, which is a confined aquifer contiguous with the Lloyd aquifer and is an important local aquifer along the northern shore for New York City and Nassau County, supplied about 3.8 percent of the total.

Hydrologic Conditions

Data on current and historical hydrologic conditions were compiled and processed for use in model calibration (history matching) and verification of simulation results, including water levels in wells, flow in streams, and chloride concentrations in groundwater. Groundwater levels and streamflows have been measured throughout Long Island since the early 1900s (U.S. Geological Survey, 2004). A total of 253,033 water-level measurements from 1,870 wells (fig. 19) were obtained from the NWIS database (U.S. Geological Survey, 2020b); the earliest measurements were made in 1907 in 14 wells in Nassau and Suffolk Counties. More extensive measurement of water levels began in the early 1930s and increased steadily until about 1950, when water levels were measured in 375 wells (fig. 20A). Measurement of water levels increased steadily after a drought in the mid-1960s and peaked in 1983 when 4,489 measurements were made in 785 wells (fig. 20).

Data collection frequency and duration differed among the sites, with differing periods of record and total numbers of individual observations. These data were examined to identify observation wells with enough information to suitably represent average annual hydrologic conditions from 1900 to 2019, the period needed for calibration of the transient model to hydrologic conditions at an annual time scale. It was determined that inclusion of a mean water level in a well as an observation in the model calibration required a minimum of 12 measurements in a given year and a minimum of 5 consecutive years meeting that minimum threshold.

Streamflows have been measured in Long Island streams since the early 1900s. A total of 497,999 values of daily mean streamflow from 111 streamgaging sites were obtained from the NWIS database (U.S. Geological Survey, 2020b). The first streamflow measurement was in 1924. More extensive measurement of streamflows began in the late 1930s; daily mean streamflows were computed at more than 50 streamgage sites by the late 1950s. The number of daily mean values peaked in 1992 at 6,798 across 94 streamgage sites (fig. 20B). The number of daily values and the periods of record vary by streamgage site, and the data were examined to determine the suitability of data from each site for inclusion in the calibration of the average annual transient model. A threshold of a minimum of 100 measurements each year at a given site was used to identify those suitable for model calibration. The number of streamgages satisfying this criterion in a given year increased from 5 in the late 1930s to 20 in the early 1990s.

Measured and estimated chloride concentrations were compiled from several sources; a total of 43,205 chloride concentrations were compiled from 4,099 sites for model verification. Chloride concentrations in groundwater on Long Island have been measured since 1903. A total of 38,380 concentration measurements from 3,747 sites compiled from the NWIS database were augmented with 4,467 measured concentrations at 443 sites from historical sources. A total of 222 estimates of chloride concentrations using borehole geophysical techniques (electromagnetic logging) were compiled from 31 sites and from 6 sites using surface geophysical techniques (time-domain electromagnetic measurements; Stumm and others, 2021). These data were used to map chloride as isochlors aggregated by the upper glacial, Magothy, and Jameco aquifers above the Raritan clay and in the Lloyd and North Shore aquifers below the Raritan clay (Stumm and others, 2024). These mapped isochlors were used in a qualitative comparison with model-simulated chloride concentrations to verify that the numerical model reasonably represents saltwater intrusion in the aquifer system as currently understood and matches known chloride concentrations.

Model Design

The two principal components of the numerical model of the Long Island aquifer system are (1) the spatial extent and discretization of the model and (2) the representation of the intrinsic aquifer properties and hydraulic stresses in the model. The discretization of the model determines the resolution of the model with regards to the representation of the physical system, including hydrologic boundaries and the mapped hydrogeologic framework. Intrinsic aquifer properties—hydraulic conductivity, storage, and porosity—and hydraulic stresses, including recharge and pumping, are averaged within model cells; therefore, the resolution of the model affects the fidelity between modeled and actual properties. The resolution of the model also determines the appropriate uses of model predictions.

Simulation of Groundwater Flow

Variable-density groundwater flow is simulated by use of the MODFLOW 6 Buoyancy package, which solves the variable-density from of Darcy’s law (Langevin and others, 2020). The Buoyancy package is coupled with the MODFLOW 6 Groundwater Transport (GWT) model simulating salinity concentrations, which allows for fluid density to be dynamically updated with those concentrations (Langevin and others, 2022). The variable density flow equations are solved within each stress period, followed by the solution of the transport equations. The Buoyancy package (Langevin and others, 2008) updates the fluid densities, and the model solution advances to the next stress period.

A seawater salinity of 35 mg/L (as chloride) was assigned to constant-head boundary cells (as represented by use of the CHD package) by use of the Sources and Sink Mixing (SSM) package. Porosity values varied by unit: glacial sands were assigned porosity values of 0.35, sands in Cretaceous units were assigned values of 0.25, and clays were assigned porosity values between 0.40 and 0.50. The Advection (ADV) package was used to simulated advective transport, using a finite-difference method to solve the transport equation. The use of a mass-balance method, such as finite difference, in advection-dominated systems such as Long Island allows for conservation of mass but can result in numerical dispersion.

Dispersion is a measure of the spread of a solute within porous sediments that occurs in all aquifers. It is a function of natural heterogeneity and is scale dependent. Numerical dispersion is a function of spatial discretization, the length of a time step, and groundwater velocity and can be approximated from the following equation:

where α is numerical dispersion, $\underset{\_}{\Delta }$X is cell length, V is velocity, and $\underset{\_}{\Delta }$T is time-step length (Fletcher, 1991). No dispersion was explicitly simulated beyond the numerical dispersion arising from the finite-difference solution within the ADV package. The inherent assumption is that the numerical dispersion in the model is a reasonable approximation of that seen in the real system at the scale of the analysis. Given a spatial discretization of 500 ft and a time-step length of 121.75 days, and groundwater velocities calculated by the numerical model, equation 1 yields a mean numerical dispersion of about 250 ft. This estimated dispersion is consistent with literature values at the regional scale of this analysis (Gelhar and others, 1992; Schulze-Makuch, 2005)

Estimation of Model Input Parameters

Model inputs were adjusted to produce an optimal fit to observed hydrologic conditions using inverse parameter-estimation techniques (Doherty and Hunt, 2010). Parameter estimation, as implemented in this effort, is a four-step process: (1) trial-and-error manual adjustments to initial parameter values, (2) an inverse algorithmic update of parameters resulting in an optimal fit to observed hydrologic conditions and associated data-quality weights, (3) manual adjustments, based on expert judgement, of inversely fit parameters to address local misfits between observations (measured values) and their model-simulated equivalents, and (4) conditioning of all model inputs to ensure all values are within ranges considered to be reasonable based on prior knowledge of the Long Island aquifer system.

Initial parameter values were assigned based on expert knowledge, literature values, and direct measurements of the groundwater system; these were updated using a trial-and-error approach. This involved using prior knowledge, known hydrogeologic conditions, and exploratory model runs to determine what adjustments were needed to improve general agreement between the model and the known system. These changes included adjustments to the hydrogeologic framework and aquifer property assignments and revisions to assumptions made for the various recharge components.

The simulated fit (or match) to select critical observations (measurements used for model calibration) was evaluated after each exploratory model simulation. The fit between observed and simulated conditions is quantified by an objective function that represents the weighted sum of squared residuals between each observation and its simulated equivalent multiplied by the weight of the observation, which is the inverse of the error associated with the observation. The objective function is referred to as PHI (Φ) and is defined by the following equation:

where Φ is the value of the objective function, w_i is the weight assigned to the ith observation (the inverse of the associated error), y_i,obs is the ith observed value, and y_i,sim is the ith simulated equivalent. A lower objective function indicates an improvement in model fit. Model parameters were adjusted manually to improve fit, followed by another exploratory model simulation to assess improvement. This process of observation evaluation, manual parameter updating, and exploratory model testing was repeated. Once the trial-and-error adjustments were complete, parameter values were systematically adjusted to be consistent with historical observations using a methodology generally following Bayes’ theorem. The prior estimate of parameter values and their likely ranges (their uncertainty) that resulted from the trial-and-error process were updated through an inverse estimation step to be consistent with historical observations, resulting in a posterior set of parameter values and uncertainty.

This parameter updating process was performed with an iterative ensemble smoother (iES) algorithm implemented in PEST++ version 5.0 (White and others, 2018, 2020b, 2021). Following implementation of iES, some model parameters were adjusted again manually to address local misfits between observations and their model-simulated equivalents. In a final manual conditioning step, some parameters were adjusted to ensure that resulting values were within acceptable ranges that reflect expert knowledge of the aquifer system. Generally, the conditioning had a minimal effect on model residuals.

Model Limitations and Uncertainty

The use of numerical models to simulate complex regional groundwater flow systems, such as the Long Island aquifer system, has inherent limitations; however, a proper model design to address scale-appropriate questions can help minimize these limitations. Some of the primary limitations of a numerical model include (1) the simplifying assumptions inherent in the conceptual model underlying the numerical model, (2) model discretization and the representation of hydrologic boundaries, (3) representation of hydraulic stresses and intrinsic aquifer properties, and (4) the accuracy of aquifer properties estimated from the data-driven model calibration.

The Long Island groundwater model is a simplification of a very complex natural system. The model integrates large amounts of spatially varying information about the groundwater system, including geologic framework and stream-network geometries, subsurface hydraulic properties, and measurements of water levels, streamflows, and chloride concentrations. Considerable efforts have been made into developing methods for quantifying model prediction uncertainty stemming from model parameters and observation data, but it is more difficult to quantify uncertainty related to structural error (Doherty and Welter, 2010). Structural error can be caused by unavoidable oversimplification of the natural system, such as consolidating complex hydrostratigraphic units into model layers or by insufficient information, such as missing pumping well data, both of which can limit the model’s ability to accurately represent real-world conditions. The effects of structural error caused by a complex hydrogeologic framework were mitigated by applying a highly parameterized approach to model construction and history matching, using zonal, pilot-point, and cell-specific multipliers to adjust model properties with as much flexibility as was determined appropriate. Despite the limitations inherent in numerical models, the Long Island regional model provides a useful and effective tool for investigating questions related to the regional flow system and resource management.

Hydrologic Budgets

The hydrologic budget of an aquifer system refers to the balance between inflows into the aquifer system and outflows from the aquifer system. Recharge from precipitation is the only natural inflow of water into the Long Island aquifer system. Natural outflows from the system include groundwater discharge to freshwater streams and to coastal waters. An additional outflow from the system is groundwater withdrawn from pumped wells; much of this water is returned to the aquifer as anthropogenic recharge. Total recharge includes both natural and anthropogenic recharge. The only component of inflow to the aquifer system from off the island is anthropogenic recharge in Brooklyn and Queens Counties, which is from leaky infrastructure in areas that receive water imported from reservoirs in upstate New York. Storage is an additional component of the hydrologic budget and refers to the internal transfer of water into or out of the compressible aquifer matrix. Water moves into storage (from the aquifer) during wet periods, and storage is considered an outflow from the aquifer. Water moves out of storage and into the aquifer during dry periods and is considered an inflow to the aquifer.

The hydrologic budget of the Long Island aquifer system is dynamic and changes continually during time in response to natural and anthropogenic stresses (fig. 45). From 1900 to 2019, inflow from recharge varied annually with alternating wet and dry periods; 1965 and 1983 were the driest and wettest years, respectively. Recharge generally trended higher after about 1970. Groundwater discharge into coastal waters was the largest component of outflow, followed by discharge to freshwater streams. Outflow to wells increased steadily between 1900 and 2019. Storage was a small component of the budget during this period and alternated between outflow and inflow during wet and dry periods.

Total inflow to the aquifer system in 1965, during a period of extended drought, was about 1,142 Mgal/d; about 71 percent from recharge and the remaining 29 percent was from water released from storage (table 5). In contrast, the total recharge to the aquifer system in 1983, one of the wettest years on record, was almost twice (1.9 times) the total recharge in 1965. The distribution of outflow to the receiving waters and pumping wells were similar during the 2 years, with the main difference being that, in 1983, about 14 percent of the total outflow was water going into aquifer storage, whereas in 1965 water was released from storage.

Discharge to streams and urban drains as a percentage of total outflow was about 24 percent in both 1965 and 1983 (table 5) despite the fact that there was nearly twice the total outflow in 1983. Discharge to coastal waters was slightly higher as a percentage of the total outflow in 1965 (51 percent) compared with 1983 (46 percent). Groundwater withdrawals were a much higher percentage of the total outflow in 1965 (25 percent) compared with 1983 (16 percent) even though there was about 40 percent more pumping in 1983 compared with 1965.

Water Levels and Streamflows

The simulated water table for 2019 ranged from essentially 0 near the coast to more than 80 ft above NAVD88 in two water table mounds to the east and west of the major surface drainages in central Suffolk County (fig. 46A). Maximum water table altitudes in the eastern and western water table mounds on the mainland of Long Island were about 60 and 80 ft above NAVD88, respectively.

Locally, high water tables occur in the northern part of Nassau County and are associated with very low-transmissivity, possibly perched clays. The highest water table altitude of about 120 ft above NAVD88 occurs in the center of Manhasset Neck (fig. 46A). Nuclear magnetic resonance borehole logging in the area confirms that this local water table feature is in an area where the sediments are fully saturated.

The altitude of the water table changes in response to natural fluctuations in recharge and because of changes in groundwater withdrawals (fig. 46). The largest variability under natural conditions occurs near groundwater divides and areas of low hydraulic gradients, and the smallest variability is near the coast and streams and in high gradient areas where water levels are more constrained. Eastern parts of Suffolk County are the least perturbed areas on Long Island, and this pattern is evident there (fig. 46B). The range of water table altitudes in the center of the island is as high as 10 ft and is close to 0 ft along the coast.

The combination of changing natural and anthropogenic stresses has resulted in a large range of water table altitudes from 1900 to 2019 (fig. 46B). The largest range was in Kings and Queens Counties and is associated with areas with large historical groundwater withdrawals and cones of depression in the water table (fig. 7); water levels in this area have varied by more than 70 ft since 1900 (fig. 46B). The range of water table altitudes varied by more than 30 ft across large areas of central Nassau County. Locally, higher variability occurs near individual wells and in areas with low-transmissivity clays.

Water-level and streamflow variability differs between wells and stream sites (fig. 47). Water levels in well N1616 near the top of the western water table mound in the center of Nassau County varied by about 24 ft during the simulation period from 1900 to 2019; the drought of the mid-1960s is evident in the water levels (fig. 47A). By contrast, water levels in in well N9099 near the coast varied by about 7 ft (fig. 47B). The same pattern is seen in the confined Lloyd aquifer, though the variability of water levels may be large owing to the lower amount of storage in confined aquifers (fig. 47C and D), with larger variations in the center of the island and smaller variations near the coast.

Water levels in the confined Lloyd aquifer in western Long Island also show large effects from groundwater withdrawals. Water levels in well Q577 varied by more than 45 ft (fig. 47C); water levels in the well were below sea level for most of the period before the early 1980s, which is likely attributed to pumping. The effects of natural stresses, such as the mid-1960s drought, are not readily apparent in this deep, confined aquifer; however, changes in pumping stress are much more pronounced, such as the more than 20-ft increase in water level following cessation of pumping in the early 1980s (fig. 47C). Water levels in the Lloyd aquifer near the coast (well N4266) show less variability than those inland near the groundwater mound; however, the variability exceeds 15 ft at well N4266 (fig. 47D), which is much larger than the variability seen in the unconfined Magothy aquifer at nearby well N9099 (fig. 47B). The water levels in well N4266 also are affected by pumping, as evidenced by the water levels below sea level between the mid-1960s and the late 1990s (fig. 47D).

Streams are surface features that are hydrologic expressions of the unconfined aquifer, and as a result, show more relative variability than do water levels in wells and are more closely correlated with recharge (fig. 47E–F); the amount of variation also is a function of hydrologic setting. Annual mean baseflow in the Peconic River in eastern Suffolk County varied by more than 40 ft³/s during the period, about twice the mean annual baseflow of 21 ft³/s, with large year-to-year variations (fig. 47E). Annual mean baseflow in the Nissequogue River in north-central Suffolk County varied by about 25 ft³/s during the same period, also with large year-to-year variations (fig. 47F). The Nissequogue River is between the regional east-west groundwater divide and the coast, in an area with generally steeper hydraulic gradients than the area near the Peconic River. As a result, there is more variability in baseflow in the Peconic River than the Nissequogue River; simulated flows in the Nissequogue and Peconic Rivers had coefficients of variation of 0.18 and 0.46, respectively. Temporal patterns in recharge are readily evident in streamflows, including the mid-1960s drought and periods of high recharge in the 1950s and early 1980s (fig. 47E–H).

Movement of the Freshwater/Saltwater Interface

The three-dimensional configuration of the freshwater/saltwater interface in the Long Island aquifer system has changed from 1900 to 2019, primarily from groundwater withdrawals in western Long Island. Since 1900, saltwater intrusion has occurred in both the unconfined and confined parts of the aquifer system in several areas on western Long Island because of groundwater withdrawals and the resulting depression of water levels. The transient regional model was used to simulate the movement of the freshwater/saltwater interface since 1900 and to define the current distribution of saltwater in the aquifer system in 2019. The current simulated interface was in generally good agreement with the interface as mapped from water quality and geophysical data (Stumm and others, 2024).

In 1900, saline groundwater with chloride concentrations exceeding 1,000 mg/L in the upper glacial aquifer generally was limited to a small area north of Jamaica Bay in Kings and Queens Counties (fig. 48A). In the early 20th century, the upper glacial aquifer was an important source of water in New York City, and large groundwater withdrawals caused saltwater intrusion and the abandonment of wells as early as the early 1930s (fig. 7). As of 2019, the upper glacial aquifer was intruded with saline groundwater well inland in Kings County (fig. 48B). The intrusion likely occurred in the early part of the 20th century but remained in the aquifer likely because of low recharge and groundwater flushing rates in this highly urbanized area.

Groundwater in the Magothy aquifer is freshwater throughout much of the land area of western and central Long Island, and freshwater extends offshore along the southern shore of Nassau and Suffolk Counties where it is locally confined by the overlying Gardiners clay (fig. 48C–D). Saline groundwater with chloride concentrations exceeding 1,000 mg/L extended inland in the Magothy aquifer in a small area along the eastern shore of Jamaica Bay (fig. 48C). As of 2019, much the Magothy aquifer is intruded in most of Kings County and along Jamaica Bay in southern Queens County and southwestern Nassau County (fig. 48D). The Magothy aquifer was historically an important source of water in New York City and remains an important source of water in Nassau County. Pumping from the aquifer, particularly in the late 20th century, has led to saltwater intrusion in several areas in southern Queens and Nassau Counties (Terracciano, 1997).

Groundwater in the Lloyd aquifer also is freshwater throughout much of the land area of western and central Long Island. The Lloyd aquifer is confined by the overlying lower Raritan confining unit, and freshwater extends offshore along the southern shore of Nassau and Suffolk Counties (fig. 48E–F). In 1900, saltwater intrusion of the Lloyd aquifer was limited to a small area close to the shore of Jamaica Bay (fig. 48E). The Lloyd aquifer was an important source of water in New York City during most of the 20th century. Groundwater withdrawals from the aquifer resulted in cones of depression, saltwater intrusion, and abandonment of public supply wells by the mid-1980s. As of 2019, the Lloyd aquifer was intruded by saline groundwater with chloride concentrations exceeding 1,000 mg/L in most of Kings County and saline groundwater extended well inland in Queens County, north of Jamaica Bay (fig. 48F). The freshwater/saltwater interface was close to the shore in the Long Beach area in southwestern Nassau County. The Lloyd aquifer and the contiguous North Shore aquifer also were intruded in some areas along the northern shore of Nassau County, including Great Neck and Manhasset Neck.

The vertical position of the interface varies across Long Island (fig. 49). The interface underlying mainland Long Island is complex, with freshwater extending seaward beneath confining units, particularly along the southern shore. An inversion occurs in the Magothy aquifer, where it is confined by the Gardiners clay, and in the Lloyd aquifer, where it is confined by the lower Raritan confining unit (fig. 49A). The freshwater/saltwater interface underlying the North and South Forks approximates the bowl shape generally typical of sufficiently thick unconfined aquifers that results from density differences and the near-hydrostatic balance between fresh and saline groundwater (fig. 49B). The aquifer system along the southern shore of Long Island it is fully fresh to bedrock underneath the mainland of the island, but the Magothy and Lloyd aquifers are naturally saline underlying the eastern part of the island. The freshwater part of the upper glacial aquifer occurs in discontinuous lenses beneath the South Fork (fig. 49C). The upper glacial aquifer along the northern shore of Long Island shows is fully freshwater to bedrock underneath the western part of the island, but the Lloyd aquifer is locally intruded in several areas in Nassau and Queens Counties where the aquifer system is thin (fig. 49C–D).

Large-scale groundwater withdrawals since 1900 have caused intrusion of saline groundwater into both the unconfined and confined parts of the aquifer system, primarily in the densely populated western part of the island (fig. 48). Saltwater intrusion has been a societally important issue for several decades, and water-resources managers have had to make costly changes in infrastructure to ensure potable water for the region’s population. However, much of the regional aquifer system has been largely unaffected by saltwater intrusion.

The volume of freshwater, defined as containing chloride concentrations less than 1,000 mg/L, in the unconfined part of aquifer, which is the principal source of water for the region, has decreased only slightly since 1900, by about 15.7 cubic miles, or less than 5 percent (fig. 50A). The volume of freshwater in the confined part of the system, which is about half the volume of the unconfined system, has decreased by about 7 percent since 1900 (fig. 50A). The decadal change in the freshwater volume of the unconfined aquifer system varied slightly but was less than 1 percent for all decades (fig. 50B). The largest decadal decreases were in the early and mid-20th century, which corresponds to a period of large-scale groundwater withdrawals and documented saltwater intrusion in New York City and western Nassau County (figs. 4 and 7). The decadal change in the freshwater volume of the confined system was less variable and generally was constant between −0.4 and −0.6 percent per decade (fig. 50B). The model indicated that the effects of saltwater intrusion on the aquifer system, at an island wide scale, generally have been small. However, intrusion of saltwater into the aquifer can locally affect the sustainability of the aquifer system and the ability to utilize the aquifer as a source of potable water for individual communities. Areas that have been locally affected by saltwater intrusion include Kings County, southern Queens County, and southwestern and coastal areas of northern Nassau County (Stumm and others, 2024).

Base Case Models of Average [2010–19] Conditions

Average pumping and recharge models were developed for both average-annual pumping and recharge stresses, and for average seasonal pumping and recharge for 2010–19 conditions. Natural recharge from 2010 to 2019 varied monthly from the annual equivalent of 0 in/yr in May 2015 to 93.3 in/yr in March 2010 (fig. 51A). Pumping varied from 208 Mgal/d in February 2017 to 824 Mgal/d in July 2010 (fig. 51A). Monthly pumping and recharge were used to calculate annual and seasonal averages for 2010–19 (fig. 51B); “in-season” refers to the five-month period from May through September and “off-season” refers to the period from October to April.

The in-season and off-season averaged natural recharge rates for 2010–19, expressed as annual-equivalent rates, were about 10.5 and 27.7 in/yr, respectively. Pumping averaged 627 Mgal/d during the in-season months and 278 Mgal/d during the off-season months. The annual average natural recharge for 2010–19 was 20.5 in/yr; pumping averaged 414 Mgal/d during that period (fig. 51B). Anthropogenic recharge components were calculated for each month from 2010 to 2019 using monthly SWB estimates (Finkelstein and others, 2022) and pumping and annual population using the same approach as was used to calculate anthropogenic recharge for historical conditions. Recharge from leaky infrastructure is a function of pumping and varied seasonally; likewise, rejected recharge is a function of natural recharge and also varies seasonally (fig. 51B). Recharge from return flow is estimated from population, which was estimated from the 2010 census (U.S. Census Bureau, 2021b), and did not vary seasonally (fig. 51B).

The average annual and average seasonal recharge and pumping stresses were used to develop predictive models for the Long Island aquifer system at multidecadal time scales. The average annual model uses the recharge calculated from the monthly SWB recharge and the anthropogenic recharge components estimated from monthly pumping and population averaged during the 10-year period (2010–19); the simulation period is 30 annual stress periods. An additional annual-scale model, using the same simulated stresses, was developed that had a simulation time of 80 years. Seasonal recharge and pumping, as estimated from monthly values, also was used to develop an average seasonal baseline model of the system. In-season and off-season pumping and recharge were alternated within each simulated year, such that there was a total of 60 stress periods during the 30-year simulation period. The simulation multidecadal time scales were necessary to ensure that the transient model was at quasi-steady-state and that any evaluations of future hydrologic stresses were not affected by residual changes in storage from the initial conditions.

The initial conditions for both baseline models were heads and chloride concentrations computed after the final stress period of the historical-conditions model (the end of 2019). Natural recharge for 2019 was 23.9 in/yr, which was substantially higher than the average of 20.5 in/yr from 2010 to 2019. Groundwater withdrawals in 2019 were about 388 Mgal/d, which was about 7 percent lower than the average of 414 Mgal/d from 2010 to 2019. Both the annual and seasonal baseline models reached quasi-steady-state after a sufficient simulation time (fig. 52). Water levels reached quasi-steady-state within about 25 simulation years (fig. 52A) and streamflows reached a steady state within about 10 simulation years (fig. 52B). These stress periods represented model simulation time and were not used to predict a future conditions in any specific year.

Seasonal changes in recharge and pumping substantially affect heads and streamflows (fig. 53A). The largest seasonal changes in water table altitudes exceed 10 ft in areas with fine-grained sediments, generally in moraines and low-permeability silt and clays and, possibly, areas of perched clays in northern Nassau County. The smallest changes are near surface waters, which constrain heads in the surrounding aquifer. The changes also are generally small in New York City due to the presence of urban drains, representing subsurface infrastructure, that also constrain water levels in some urbanized areas by intercepting water from rising water tables. Local occurrences of large seasonal changes in water table altitudes are near pumping wells and are the result of seasonal changes in groundwater withdrawals. Changes can be made to hydrologic stresses and the resultant heads and flows after a simulation period of 30 years can be compared to the final heads and flows from these baseline models to assess the effect of those changes on the hydrologic system, in the absence of other effects.

Drought

The average seasonal baseline model was modified to represent the effects of a 5-year drought similar to the 1960s drought, often referred to as the “drought of record.” The drought is represented as starting in the 5th year of the simulation and has a duration of 5 years. The simulation of drought conditions was made by decreasing average seasonal recharge by 20 percent and increasing average public-supply and non-public-supply pumping by 20 and 10 percent, respectively. The comparison of water table changes for in-season conditions after the 5-year drought simulation, or the 10th year of the simulation, showed water-level declines ranging from 0 to 18 ft from the coast to the inland tops of the groundwater mounds (fig. 53B). Simulated water-level declines were small away from the coast, in the vicinity of the large streams that drain to the southern shore (fig. 53B). The in-season streamflows decreased by about 32 percent in response to the 5-year drought, while water-level changes in the vicinity of the streams were relatively small (fig. 53B).

The small change in water levels in these areas was because of the hydrologic control of the streams on nearby water levels. The streams on Long Island are groundwater fed; therefore, changes in pumping and recharge in the vicinity of these streams result in large changes in streamflows rather than in water levels. However, once the streams go dry during a drought or from nearby large-scale pumping, it is expected that water levels in these areas would continue to decline similar to those in areas away from streams.

Sea-Level Rise

The annual average baseline model was modified to represent a sea level position of 6 ft above NAVD88 as an alternative to the baseline sea level of 0.34 ft above NAVD88 in 2019. The altitude of the new sea level was used to identify model cells that are in emergent areas under current [2019] sea levels that would be submerged under the new sea level. Newly submerged land areas were assigned a head-dependent flux boundary, by use of the GHB package (McDonald and Harbaugh, 1988). Likewise, freshwater streams that are below the new sea level were converted from drain boundaries (using the DRN package) to GHB boundaries (fig. 54). Average annual recharge and pumping from 2010 to 2019 were simulated in the models. The simulations were discretized into 30 annual stress periods.

Changes in sea level affect the base level of the aquifer and, therefore, water levels and streamflows (fig. 53C). The maximum change in water table altitudes, as determined by comparing the water table at the end of the 80-year simulation for current [2019] and future sea levels, generally is limited to the increase in sea level because only the base level is changed, and the recharge and pumping are unchanged. The maximum change in water table altitudes, assuming a 6-ft sea-level rise, generally is in areas adjacent to the coast in areas with no surface-water drainages. These areas are generally in the northeastern coast of Suffolk County and on the North and South Forks. The smallest changes in water table altitudes are along the southern shore of Nassau and Suffolk Countries (fig. 53C). These coastal areas are characterized by a dense network of streams. Streams are fixed boundaries that constrain heads by passively removing water from the system. This results in increased streamflows and the potential for local up coning of the shallow freshwater/saltwater interface beneath the streams, similar to what has been reported in analogous hydrologic settings on Cape Cod (Walter and others, 2016).