A comparison of historic Moreau Lake water levels among USGS topographic maps shows a 10-ft variation. The 1:62,500 Schuylerville quadrangle map (U.S. Geological Survey, 1900), surveyed in 1898, indicated a lake water level of 341 ft above mean sea level and that all the adjacent embayments and ponds east and north of the lake held water (fig. 1). The 1:24,000 Gansevoort quadrangle map (U.S. Geological Survey, 1968), surveyed in 1964, indicated a lake water level of 331 ft above mean sea level, and all adjacent embayments and ponds are mapped as wetlands.

During the 1960s drought, which was the longest, driest period on record in the northeastern United States (Barksdale, 1968; Cook and Jacoby, 1977; Pederson and others, 2013), lake water levels at Moreau Lake dropped substantially, which prompted the Capital District State Park Commission of New York State to fund a study of lake water levels and possible remediation options by a local engineering consultant. A variety of data were collected from July 1968 through July 1969 (Rist-Frost Associates, 1969). That study included six subtasks, documented in appendixes: a seismic-refraction study, a water balance study, dye studies and temperature profiles, a ground-magnetic study, water surveillance charts, and a geologic report.

The surficial geology of the lake and surrounding area was mapped by Cadwell and Dineen (1980–88) and reproduced in Heisig (1994, pl. 1). Moreau Lake occupies a kettle (depression), the largest of several within a kame terrace. Numerous hydrogeologic studies of kettle lakes in eastern mainland Massachusetts and on Cape Cod have documented the hydrogeologic framework, groundwater flow systems, and useful techniques for tracing kettle lake water movement into local groundwater flow systems (Colman and Friesz, 2001; Friesz and Church, 2001).

The study was initiated in mid-2016 to collect hydrologic and hydrogeologic data in the park and surrounding area and to document seasonal and annual fluctuations in lake water and groundwater levels. The data collected from 2016 to 2021, combined with precipitation and groundwater use estimates, provided a sound basis for understanding the factors controlling lake water levels. The following tasks were undertaken:

The Moreau Lake area straddles two physiographic provinces: the Adirondack Mountains on the western side and the Hudson-Mohawk lowlands on the eastern side. The Adirondack bedrock of the Palmertown Range (specifically Grant Mountain) consists of resistant, intensely deformed metamorphic rocks (gneisses) of Precambrian (Proterozoic) age (Fisher and others, 1970; Fisher, 1984), which rise as much as 900 ft above the adjacent lowlands. The lowlands are primarily underlain by less resistant shale of late to middle Ordovician age and minor amounts of underlying carbonates, which are exposed at the base of the mountainside immediately southwest of Moreau Lake and in the subsurface about 1.5 mi north of the lake (figs. 6 and 7). The boundary between these two provinces marks the trace of the McGregor fault, a normal fault along which the older, more resistant Adirondack bedrock has moved upward at least 2,070 ft relative to the younger, less resistant shale of the lowlands (Geraghty and Isachsen, 1980; Tice, 1993).

Bedrock is at or near land surface over much of the uplands (Palmertown Range) but is blanketed by as much as 380 ft (well SA10287) of unconsolidated deposits in the lowlands. The elevation of the bedrock surface beneath the lowlands in the Moreau Lake area was estimated from well logs and from HVSR seismic measurements (fig. 8). The main feature of the bedrock surface is a 50- to 100-ft deep north-south trough immediately east of Moreau Lake, which is interrupted by some subsurface bedrock highs adjacent to Moreau Lake and thereafter trends southwest and deeper along the mountain front (figs. 7 and 8). The minimum estimated elevations in the trough are less than 50 ft above sea level. East of the trough, bedrock surface elevations are 150 to 200 ft above sea level. The trough is completely buried by unconsolidated deposits (figs. 7 and 8). The depths to bedrock in this interpretation are significantly greater than those interpreted from previous seismic work (Rist-Frost Associates, 1969), but are supported by more recent well-log data.

Unconsolidated deposits consist of glacial (Pleistocene-age) and recent (Holocene-age) deposits. Glacial deposits make up most of the unconsolidated material and were deposited in the lowlands largely during final northward retreat of the Hudson Valley Lobe of the ice sheet at the end of the ice age (a little younger than 14,800 years ago; Ridge, 2003; 2012). Recent deposits are primarily alluvium deposited in floodplains of the Hudson River and along the courses of the largest streams.

Surficial deposits of the area are depicted in figure 6. Till deposits in the form of drumlins record the direction of ice movement (northeast to southwest), showing the deposition and reshaping of till beneath the ice before ice retreat. The ice-contact sand and gravel deposit along the base of Grant Mountain contains Moreau Lake and numerous other kettle (depression) features in the area, water-filled or not. The landform of this unit is a high (kame) terrace, which was deposited in a gap between the mountainside and the ice tongue in the Hudson valley by outflow from an upgradient glacial lake (fig. 9; Chadwick, 1928; LaFleur, 1965). The meltwater exited farther south into glacial Lake Albany, which was in contact with the retreating ice. Detached blocks of ice in this gap were surrounded and at least partially buried by this sediment. These ice blocks eventually melted, and overlying and adjacent sediment collapsed, leaving depressions. The largest and deepest of these kettles are below the present-day water table, and thus contain lakes. Moreau Lake is the largest of these kettle lakes. This kame terrace feature is referred to as the Moreau Pond terrace by Chadwick (1928) or Moreau Pond kame terrace (LaFleur, 1965).

Continued ice retreat and meltwater discharge through the Hudson River gap at the northern end of Palmertown Mountain widened the gap between the ice and the terrace, which allowed deposition of kame delta or lacustrine deltaic deposits (Chadwick, 1928; Tice, 1993) and two levels of contiguous spit deposits (fig. 6). The high plain (400 to 380 ft elevation) of mostly sand (from well logs) north of Moreau Lake is also attributed to Hudson River discharge into this area, constrained by ice to the east (Chadwick, 1928).

With the continued northward retreat of the ice margin, open-water conditions prevailed in the area from Lake Albany I at about 420 ft elevation at Milton Delta (LaFleur, 1965) through Lake Albany II at about 350 to 360 ft elevation (DeSimone and LaFleur, 1985). Fine-grained lacustrine sediments (clay, silt, and fine sand) were deposited in the deep waters of the lakes, and fine to medium sand, in shallow areas (fig. 6).

Holocene-age deposits are primarily floodplain alluvium and alluvial terraces in the Hudson River floodplain (perhaps 10 to 20 ft thick) and less along the courses of the larger streams in the area. Since the glacial lakes drained, lacustrine sand deposits have also been subsequently partially reworked into dune fields in some areas of the lake plain (not mapped, but visible in fig. 6).

Well logs from the area provide information in the third dimension that is critical for delineation of the subsurface hydrogeologic framework and how it relates to Moreau Lake. However, many well logs in the area only specify a general sequence of deposits in the overburden and do not give specific depths where changes in sediment type occur. The limited number of well logs that do detail the sediment types and their depth intervals were used with the less descriptive logs to infer the framework in three cross sections—two west-east cross sections north and south of the lake and a north-south cross section that includes Moreau Lake (fig. 6). The cross sections focus on sediment descriptions because the deposits change with depth and because sand is a primary component beneath the two most important surficial units—lacustrine sand and ice-contact sand and gravel. The key difference in subsurface material is whether it is fine-grained or coarse-grained. Saturated sand and gravel generally constitute aquifer materials and fine-grained deposits may act as confining units, which impede groundwater movement. Surficial deposits from figure 6 are listed above the cross sections for reference.

The surficial material of the northern west-east cross section A–A′ is largely lacustrine sand with a narrow strip of ice-contact sand and gravel at the base of the bedrock mountainside at the western end (fig. 10). Well logs indicate that the sand is 80 to 126 ft thick and underlain by lacustrine fine sand, silt, and clay that is at least 152 ft thick over the bedrock trough and thins to about 80 to 100 ft along the edges of the trough. The elevation of the top of the fine-grained deposits is 270 to 300 ft. The log of a single deep well log north of the section (well SA2696; fig. 8) indicates sand and gravel beneath at least some of the fine-grained deposits.

The surficial material shown in the southern west-east cross section C–C′ includes a wide band of ice-contact sand and gravel eastward from the base of the mountain that transitions first to lacustrine delta sands and then to lacustrine fine sand and silt (fig. 11). Depth to bedrock is greatest beneath the ice-contact deposits and shallows to the east. The sediments beneath the mapped ice-contact area are varied but mostly sand: the logs for two wells (SA4132 and SA7489) indicate 40 to 50 ft of gravel over 40 to 60 ft of sand. Beneath that, about 50 ft of fine sand, silt, and clay is underlain by an unknown thickness of sand. The fine-grained units are not continuous; two other well logs indicate 290 ft of sand (well SA 162) and 101 ft of mostly medium to coarse sand with fine sand and silt layers between 30 and 39 ft (well SA4064), and a third well log indicates 380 ft thickness of sand and boulders over till (well SA10287).

The tops of the fine-grained units are at different elevations. West of the till, elevations are about 345 to 350 ft, whereas east of the till, the elevation is about 310 ft. The depth at which the till was penetrated is not noted in the well log, but higher elevation of fine-grained deposits to the west might be explained as local impoundments in part dammed by till—if so, the till would need to be thick, as outlined by the inferred contact in figure 11. Eastward in the cross section, the lacustrine delta deposit includes a well-sorted sand of unknown thickness. The lacustrine fine sand and silt surficial deposit is very thin and overlies 75 ft of clay at well SA7434, which overlies sand and gravel of unknown thickness.

The third cross section (D–D′) is oriented south to north across Moreau Lake and extends about 1 mi in each direction from the lake (fig. 12). Two-thirds of the cross section is within the ice-contact (kame terrace) feature and one-third is within the lacustrine plain to the north of the lake. Subsurface differences between the areas are evident. To the north, fine-grained sediments constitute a thick confining unit that is likely continuous. To the south, confining units are thinner and discontinuous, causing only local confinement, and masses of till on bedrock are more common. Beneath the lacustrine sand, as far south as well SA 408, there is about 110 ft of sand overlying about 150 ft of fine sand and silt or silt and clay. The top of the fine-grained sediments is at about 270 to 275 ft in elevation. The log for well SA2696 indicates some sand and gravel beneath the fine-grained sediments. South of well SA 408, there is an abrupt change—the well log at well SA8015 indicates 190 ft of sand, but no obvious fine-grained sediment, overlying 80 ft of till over bedrock. Perhaps this is the subsurface transition to the ice-contact deposits. There is no subsurface geologic information for the remaining distance to Moreau Lake where the surficial deposits transition to ice-contact deposits.

Beneath the ice-contact surficial deposits south of Moreau Lake, the subsurface is characterized by (1) apparently discontinuous and thin (approximately 50 ft) fine-grained units (top-of-unit elevation 345 to 350 ft), (2) areas of till beneath the sand, and (3) instances of thick sand over sand and gravel. In effect, there is not a consistent fine-grained unit (confining unit) above the deep sand or sand and gravel deposits, and the basal sand and gravel unit is generally absent at depth where till is present atop the bedrock surface.

The silt and clay unit just south of Moreau Lake (figs. 11 and 12) is about 50 ft thick with an upper surface elevation of about 350 ft. The known range of lake water level elevations at Moreau Lake is about 331 to 341 ft, so the bottom of the confining unit may extend 30 to 40 ft below the surface of the lake, which has a maximum depth of about 50 ft. This might indicate that subsurface outflow of lake water to the south is largely blocked by this fine-grained confining unit. It also implies thin saturated sand above the confining unit because it is above lake water level and because the thick unsaturated zone (about 80 ft) does not allow for rapid recharge to the water table. A thin, unproductive saturated zone above the confining unit is also indicated by well completions in the area—all wells draw from saturated sand or sand and gravel beneath the confining unit, rather than from above it. The confining unit does not appear to extend completely to the lake because there are no exposures of fine-grained material and no springs are evident on the slopes along the south side of the lake. There likely is a layer of collapsed sediment (stratified and glacial till) around and beneath the lake resulting from melt-out of the ice block that occupied the area. The confining unit, if present, is also not thick enough to completely underlie the lake. Historic seismic data from Rist-Frost Associates (1969) indicate till at the southern end of the lake as well; because the confining unit does not completely underlie the lake, the till would likely inhibit or block the southward outflow of lake water into the aquifer.

The sediments beneath and immediately adjacent to Moreau Lake (at and below the lake water level) are poorly defined because there are few wells in the park and there are no well logs for them. However, inferences can be made, given the environment in which Moreau Lake developed and from other data sources.

As the Moreau Lake ice block melted, sediments within and atop the ice melted out, leaving a residue of unsorted sediments (flow or ablation till) and possibly some stratified sediments in the depression left by the ice block. Till may have been deposited beneath the ice block before the melt out, resulting in a discontinuous unsorted mixture of dense sediments beneath and around Moreau Lake. Seismic-refraction measurements (from 1969) along much of the southern and all the eastern shores of the lake show seismic velocities (7,000 to 10,000 ft/s) for the material above bedrock that are indicative of till (Rist-Frost Associates, 1969, app. 1, fig. 1.4). In the present study, a shallow wellpoint could not be installed on the central-eastern shore of the lake because a dense till-like material was reached within 2 ft of land surface at the shoreline.

The seismic data and interpretations from Rist-Frost Associates (1969) for the Moreau Lake study area, however, differed from the HVSR seismic measurements from this study (fig. 8; Heisig and Sporleder, 2026) and recent well logs that define the bedrock surface outside the immediate lake area as deeper and till less pervasive along parts of the southern and eastern parts of the lake. If the presence of thick till is correct, this would limit groundwater flow in those directions. However, groundwater temperature and water isotope data (discussed in the “Movement of Water From Moreau Lake in the Local Groundwater Flow System” section of this report) from nearby wells immediately east (well SA 400) and southeast (well SA 404) clearly show lake-derived groundwater flowing in those directions.

Evidence for permeable material on the eastern side of Moreau Lake is particularly strong at well SA 404, an unused downgradient well (161 ft deep) on a knoll about 60 ft above the lake (fig. 12) that has about 100 ft of water in the casing below lake water level. Water temperature profiles in this well are nearly 10 degrees Fahrenheit (°F) warmer than that of well SA 402, a bedrock well on the western (upgradient) side of the lake, which indicates that lake water is moving eastward past the entire 100-ft saturated length of well SA 404 (Heisig, 2026b). If there was a confining unit at this location, one might expect to see discontinuities in temperature profiles within the casing, which were not observed. In summary, at the very least, low permeability till exists as discontinuous shallow lenses and deeper mounds on bedrock in the area beneath and around Moreau Lake, as depicted in figure 12, but with considerable uncertainty.

Aquifers in the lowland area around Moreau Lake include unconfined sand, locally or semiconfined sand or sand and gravel, and several confined aquifers, which include sand and gravel or fractured bedrock (figs. 13 and 14). The unconfined sand aquifer is the most widespread in the areas of lacustrine sand and ice-contact sand and gravel (fig. 6) where its overall thickness and saturated thickness can support year-round domestic water withdrawals. This aquifer is the most extensive and reliable in the terraced sand plain north and northeast of Moreau Lake. It occupies a narrow strip east of Moreau Lake and extends south and southwestward within the ice-contact deposit, although it is not consistently used for water supply there presumably because of insufficient saturated thicknesses (fig. 13).

The semiconfined aquifer lies beneath discontinuous confining units (lacustrine fine sand, silt, and clay or till) mostly south and southwest and immediately east of the lake (figs. 12 and 14). Confined aquifer areas (fig. 14) include (1) thick lacustrine fine sand, silt, and clay that overlies and confines sand and gravel or fractured bedrock at depths as great as 250 ft, and which is north and northeast of Moreau Lake within the bedrock trough (figs. 8 and 12); (2) a few instances of very deep (270 to 380 ft) sand and gravel or fractured bedrock confined by till; (3) an area of lacustrine clay over sand and gravel southeast of Moreau Lake; and (4) farther east of the aforementioned areas, fractured bedrock overlain by lacustrine clay or till confining units.

Before an accounting of potential effects of increases in groundwater withdrawals on lake water levels can be made, the distribution of wells and the aquifers they withdraw from must be documented and interpreted. Within an arbitrary 2-mi radius around Moreau Lake (fig. 15), there are 675 parcels served by private water sources (Saratoga County, 2016). Within that same area, 237 wells have been drilled for water supply since 2000 (New York State Department of Environmental Conservation, 2022), indicating growth of about 35 percent from 2000 to 2016.

Well logs for the new wells were examined to determine the type of aquifers being used and to map the distribution of water use within a 2-mi radius of Moreau Lake (fig. 15; table 2). The results indicated that the unconfined aquifer was most used north and east of Moreau Lake and that the semiconfined aquifer was most used south and southwest of Moreau Lake. Confined aquifers were minimally used near Moreau Lake. Pumping from unconfined aquifers releases groundwater by dewatering of pore space and causes only localized drawdown, whereas pumping from confined aquifers releases groundwater by depressurizing pore space and causes more areally extensive drawdown. Semiconfined aquifers are not continuously confined, which means that any recharge windows (local unconfined areas) that do occur will dampen or prevent declines in water levels from being transmitted any great distance. Thus, it appears unlikely that semiconfined aquifer withdrawals from the large number of new wells south of Moreau Lake have had an effect on lake water levels.

Nevertheless, for completeness, estimates of post-2000 (new) and total annual groundwater withdrawals from the semiconfined aquifer south of Moreau Lake are presented in this report and are compared with other water budget components in the “Water Budget Components Within the Moreau Lake Watershed and Comparison With Estimated Water Use” section of this report. The land use of the area of interest is essentially all single-family residential. The census block that covers the semiconfined aquifer area south of the lake (BG 360910601.013) indicates an average household size of 2.6 people (HomeTownLocator, 2020). Using a water usage estimate of 75 gallons per day per person for New York State (Dieter and others, 2018, table 6) and 101 households for the area (table 2), the estimated increase in water use from the aquifer during the past 20 years (since 2000) is about 7.2 million gallons per year (Mgal/yr). Residences in the area are served by onsite septic systems. Septic leachate is returned to the unconfined aquifer through a generally thick unsaturated zone (commonly greater than 40 ft). Only limited return to the confined aquifer is possible where confining units are absent. About 9 percent of domestic water use in New York State is consumptive (lost to the atmosphere) through activities such as lawn watering (Shaffer and Runkle, 2007, app. 1, table 1.1).

Before 2000, there were few wells in the area; therefore, groundwater use was low. Tax parcel data indicated that about 127 parcels in the aquifer area were supplied by private wells (Saratoga County, 2016). There were a total of 120 wells tapping semiconfined and unconfined aquifers since 2000, which indicated that there were seven preexisting wells in the area. Assuming six wells tapped the semiconfined aquifer, annual water usage before 2000 was estimated to be about 0.4 Mgal/yr. The total annual withdrawal in 2020 included 107 households and was about 7.7 Mgal/yr.

Groundwater level measurements (2017–2020) from a network of 16 wells, together with historic groundwater level measurements and lake water level monitoring (July 2016–2021; fig. 3), were used to define the local groundwater flow system around Moreau Lake. Three years of discrete groundwater level measurements from wells in unconfined to semiconfined settings near Moreau Lake with those from Moreau Lake (well SA1305) indicated that Moreau Lake water levels were higher than water levels in all but two wells on the western (mountain) side of the lake (fig. 16). This indicates a groundwater level gradient and general groundwater flow from high-elevation areas in the west to low-elevation groundwater discharge areas in the east. In addition, groundwater level fluctuations during the period of record are largest near areas of highest groundwater recharge at the base of the mountainside and around Moreau Lake (5.7 to 6.9 ft) and progressively decrease downgradient to a low of 1.5 ft near an area of groundwater discharge (well SA 399).

A map showing contours of the water table surface in the greater Moreau Lake area was constructed using measurements from 15 wells, Moreau Lake, and a kettle pond south of Moreau Lake measured on June 11, 2019 (a time of high water levels), 15 historic water levels reported in driller’s logs, and from mapped elevations of streams (fig. 13). The historic water levels are given less weight for delineation of water level contours. They were measured at different times of year, but the median water level fluctuations from 13 recently measured wells with minimal pumping effects was about 5.5 ft, an indication that it is reasonable to use them with 10-ft elevation contours. All wells in the network were completed in what are interpreted as either unconfined or semiconfined aquifers. The map depicts the general decrease in groundwater levels from the mountainside eastward as well as subsequent bending of the water table contours and groundwater flow lines toward discharge areas. Discharge areas are springs and seepage areas that are the source of streamflow. Groundwater flow lines indicate the direction of groundwater flow. A notable feature is a large eastward bulge of high water levels north of Moreau Lake, indicated by the 330 and 340 ft contours, that correspond to the highest terrace elevation of the sand plain. The southern side of the bulge encompasses the kettle ponds immediately north of Moreau Lake. This implies that groundwater levels, and thus kettle pond levels, are higher than the level of Moreau Lake. The pond immediately north of Moreau Lake (northern embayment; fig. 1) is mostly separated from the lake by a narrow sandy strip of land but connected through a small channel under a footbridge. When the channel has been dammed, a higher water level is maintained in the pond than in Moreau Lake and erosion from groundwater seepage on the Moreau Lake side just east of the footbridge has also been observed, all of which is consistent with the water table map.

Water levels from confined aquifer settings (fig. 14) were not used to construct a potentiometric (confined water level) map because of uncertainty in the degree of hydraulic interconnection of the confined aquifers and the availability of only historic measurements. Wells completed in semiconfined settings were monitored and were considered most similar to adjacent unconfined settings. No wells in strictly confined settings were monitored during the study. It was considered unlikely that the deep confined aquifer in the bedrock trough was in hydraulic connection with the much shallower confined settings 1.5 to 2 mi from Moreau Lake. Historic water levels within the deep confined aquifer (wells SA10287, SA8015, SA2696, and SA 408) showed a general, but not consistent, decrease in water levels from south to north, but the differences were mostly within a 5-ft range, which was within the likely errors of the historic measurements (plus or minus 5 ft) and below the 10-ft contour interval that would have been used. All deep confined aquifer water levels east and north of the lake were lower than in the overlying unconfined aquifer, whereas in the single deep confined aquifer well south of the lake, water levels were higher than that in the unconfined aquifer.

An understanding of the hydrogeologic framework of the greater Moreau Lake area provides a basis for looking more specifically at the movement of water from Moreau Lake after it enters the local groundwater system. Interpretation of groundwater flow direction from water level data was supplemented with field measurements (water temperature and specific conductance) and laboratory analyses (oxygen and hydrogen isotope ratios of water) that serve as groundwater tracers. If lake water is being drawn towards areas of increased groundwater withdrawal (to the south), it would indicate the potential for negative effects on lake water levels. Alternatively, if lake water is moving through the groundwater system to natural discharge areas, then it can be inferred that increased groundwater withdrawals have no discernable effect on lake water levels.

An annual accounting of water in the Moreau Lake area can be estimated from regional maps of 1951–80 mean annual values of precipitation, runoff, and evapotranspiration from Randall (1996) as shown schematically in figure 23. Precipitation and runoff values were interpolated from National Oceanic and Atmospheric Administration weather stations and streamflow records at USGS gages, respectively, and evapotranspiration was estimated as the difference between those values.

The lack of precipitation data in southern Saratoga County and atop the Palmertown Range are limitations. Annual precipitation, the sole water input, is about 38 in., annual runoff is about 18 in., and annual evapotranspiration is about 20 in. Runoff consists of two components: rapid surface flow to streams and other waterbodies during intense storms or snowmelts and water that infiltrates land surface and either replenishes soil-moisture deficits or continues downward to the water table and recharges groundwater. At Moreau Lake, the two main tributaries generally lose all flow to infiltration along the streambed where the streams leave their mountainside bedrock channels and flow over sand and gravel adjacent to the lake.

Water budget components such as evapotranspiration and runoff vary in time during the course of a year because of the large seasonal return of water from the Earth’s surface back to the atmosphere during the warmer months (the growing season) through plant transpiration and evaporation. The monthly distribution of precipitation and estimates of evapotranspiration provide an indication of whether evapotranspiration may be limited by rainfall during the growing season as well as the amount of water available for runoff and recharge during the course of a year (fig. 24). About 78 years of data from the Floyd Bennett Memorial Airport (formerly the Warren County Airport) in Glens Falls were used to generate monthly mean air temperature and monthly mean, minimum, and maximum precipitation. Monthly means of precipitation (fig. 24) differed by 1.3 in. or less, but with higher totals in the summer months and lower in winter, particularly in February and January. The minimums and maximums show that a wide range of departures from the mean are possible in any given month. Monthly temperature means were used as input for the Thornthwaite method (Thornthwaite, 1948) for calculating potential evapotranspiration by month (fig. 24).

Potential evapotranspiration is the amount of evapotranspiration possible if water is not a limiting factor—essentially the upper limit of possible evapotranspiration. The annual total from this method was 24.2 in. Because this was considerably higher than the approximately 20-in. estimate from Randall (1996), the monthly distribution was prorated for a seasonal total of 20 in., which is considered more realistic (fig. 24). These estimates of monthly mean evapotranspiration exceed the precipitation means (thus producing a water deficit) by less than 1 in. during June, July, and August, which is a relatively small amount. For comparison, the Thornthwaite estimates extend deficit months from May to September, with a maximum deficit of about 1.75 in. from mean precipitation in July. These deficits are much smaller than the variations inherent in monthly precipitation, which indicates that conditions of water surplus or deficit relative to plant needs may be highly variable during the warmest summer months in the Moreau Lake area. This figure also indicates a water surplus most other months of the year, and this surplus becomes either overland or shallow subsurface runoff and groundwater recharge.

The fate of precipitation over the course of an annual cycle in regional (upstate New York) conceptualizations such as the schematic presented by Heath (1965) is considerably different from that of the Moreau Lake area (fig. 25). In upstate New York, precipitation is depicted as relatively constant from month to month, but evaporation and transpiration are major components during the warm months (growing season) and cease or decrease substantially during the cold months (nongrowing season). Runoff, divided into overland runoff and groundwater recharge, behaves in the opposite way—it is highest during the nongrowing season and lowest or largely absent during the growing season. Overland runoff to streams and evaporation and transpiration to the atmosphere are shown as losses from the system, but there is an additional loss: groundwater outflow. Groundwater systems are dynamic—groundwater recharge is an addition of water to the groundwater system, but groundwater is constantly flowing toward discharge areas to more or less balance the inputs. If not in balance, groundwater storage may increase or decrease (a change in water levels). In many areas, recharge may be a minor component, but in areas of sand and gravel aquifers, it can be significant.

Moreau Lake and its groundwater contributing area (fig. 25B) have a somewhat unusual hydrogeologic setting (fig. 13) and have water-budget components that depart from those of the general schematic for upstate New York (fig. 25A; Heath, 1965). Much of the contributing area (0.68 square mile [mi²]) to the lake (0.20 mi²) consists of bedrock mountainside (78 percent) separated from the lake by a fringe of sand and gravel (22 percent). Groundwater flow from the mountainside, through and beneath the lake, and exiting to the east and southeast from the lake (fig. 19) is indicated by higher-than-lake groundwater levels in the bedrock and sand and gravel on the western side of the lake and lower-than-lake groundwater levels to the south and east (fig. 16). This means that the contributing area of groundwater to the lake is strictly upgradient (west) from the lake, as shown in figure 13. The surface or topographic watershed of the lake might suggest water contributions from steep sand and gravel slopes on the eastern, southern, and western sides of the lake. High infiltration rates of sand and gravel indicate that only intense summer thunderstorms with rainfall rates that exceed the infiltration capacity of the soil or rain or snowmelt on frozen ground in winter could result in surface runoff to the lake. This represents a small area of infrequent contribution and is therefore not considered further.

Each water-budget component of the local hydrogeologic system has its own variation (fig. 25B) compared with those in upstate New York (fig. 25A). All precipitation falling directly on Moreau Lake is direct recharge (no overland runoff), although some water losses from plant transpiration along the shoreline are expected. Evaporation during ice-free periods is likely larger than plant transpiration. Groundwater loss through flow or discharge to the east (primarily) and southeast is the major loss from the lake and underlying sediments.

The water budget of bedrock uplands most closely follows that of upstate New York (fig. 25A) except that overland runoff is limited to two small tributary streams (fig. 1), and much of the water input infiltrates and moves through the shallow subsurface of broken-up bedrock (regolith) and through fractures (fig. 26) in the upper 100 ft or so of bedrock (as recharge) toward the sand and gravel at the base of the mountainside.

The sand and gravel that separates the bedrock mountainside from the lake is an important link in the system because of its high permeability. Water from the mountainside recharges either groundwater in the subsurface where bedrock fractures discharge to the sand and gravel or sand and gravel where mountainside streams cross the sand and gravel and largely infiltrate into the streambed before reaching the lake (fig. 27). Only the two largest tributaries regularly have flow on bedrock before reaching the sand and gravel, but it is only during periods of high-intensity or long-duration rainfall, major snowmelt periods, or spring wet periods that tributary flow directly replenishes Moreau Lake (fig. 27). The period with the most flow to the lake from the south tributary is in late spring when high water inputs from the mountainside raise groundwater levels in the sand and gravel enough that tributary flow is maintained all the way to the lake. The net result (fig. 25B) is that overland runoff is a minor component of the water budget and that recharge is a major component. The end date of plant transpiration is also shifted a little later in the year (fig. 25B) compared with upstate New York (fig. 25A), on the basis of water level responses at well SA1305.

Recharge during the growing season is limited, but more likely in the Moreau Lake area (fig. 25B) than in upstate New York (fig. 25A) because fractured bedrock near or at land surface on the mountainside and the sandy soils around Moreau Lake can accept water from occasional large, long duration rainfalls. Thick unsaturated zones in sandy soils effectively delay recharge water reaching the water table (offset in overland runoff and recharge in fig. 25B). Groundwater recharge is more rapid where stormflows in streams cross dry sand and gravel streambeds.

The variation in seasonal monthly precipitation departures from median values (fig. 24), especially during the nongrowing season, can be a large percentage of the seasonal monthly precipitation medians. Figure 28 shows the largest variations in seasonal precipitation from the respective medians during a recent 5-year period (2015–19) in terms of volume. The span in the difference between extremes during the nongrowing season is 25.5 million cubic feet (Mft³) of water, about 75 percent of the seasonal median, while during the growing season, the difference is 16.5 Mft³ of water, about 40 percent of the seasonal median. This variability in precipitation affects lake water levels at the seasonal scale, which is addressed in the “Comparison of Continuous Lake Water Level Data (2016–21) With Precipitation Data” section of this report.

Water use within the Moreau Lake groundwater contributing area is limited to the primary production well (SA 401) on the southwestern side of the lake; a second production well (SA 400) is downgradient from the lake and was considered in the discussion of the semiconfined aquifer. Well SA 401 is a shallow (30 ft deep) well completed in the unconfined sand and gravel aquifer. The well serves most of the campground and the beach facilities at the park. An estimate of annual withdrawals from typical summer-season pumpage (J. Burnham, park supervisor, New York State Department of Parks, Recreation and Historic Preservation, oral commun., February 2, 2022) is approximately 1.25 Mgal/yr. The well is not used during the off-season (winter). The well is close to the shoreline of the lake and initially was assumed to induce water from the lake into the sand and gravel aquifer when pumping, but this is not the case. The well is close to the south tributary, which follows and drains a large linear feature on the mountainside. Most flow from the tributary infiltrates into the stream bed (fig. 27).

Together with local contributions from the mountainside, this is an area of focused recharge. Groundwater from the mountain is low in mineral content—the south tributary specific conductance range is 32 to 74 µS/cm—and groundwater from wells and springs not influenced by Moreau Lake and associated ponds is cold, 43 to 49 °F, as discussed in the “Water Temperature” section of this report. Lake water, in contrast, shows higher specific conductance (115–133 µS/cm, from August and September 2018 lake-center vertical temperature and specific conductance profiles) and wide seasonal temperature fluctuations. For example, lake center vertical temperature profiles taken during winter (February 19, 2020) and summer (August 16 and September 12, 2018) ranged from 33 to 41 °F and 50 to 82 °F, respectively (Heisig, 2026b). When well SA 401 was sampled on September 3, 2018, (Labor Day weekend) after pumping at least 2,500 gallons, the specific conductance was 43 µS/cm and the temperature was 45 °F, a clear indication that at the end of the summer camping season (the maximum possible stress), the well water was derived from the mountain and not from the lake. A water isotope sample from this well also conclusively shows that, unlike lake water, the well water was not subject to evaporation (figs. 21 and 22).

In effect, the focused recharge of mountainside water into the sand and gravel near well SA 401 results in a gradient toward the lake, most obvious during the nongrowing and recharge season (fig. 4), that prevents lake water from reaching well SA 401. Thus, some groundwater that would discharge to the lake is intercepted, but lake water is not being induced into the aquifer. This withdrawal is not a new stress on the system and should not be a factor on any recent lake water level behavior because the park has been in operation since 1961.

Withdrawals from wells that are completed in the semiconfined sand and gravel or bedrock aquifers were evaluated given the possibility that lake water levels might be affected. However, groundwater flow directions derived from groundwater level elevations and from physical and chemical parameters showed that there was no significant groundwater flow to the south, where most wells drawing from confined aquifers were. The water use estimate (about 1.0 million cubic feet per year of water) shown in figure 28 is added for completeness and to show that its magnitude (a worst case) is small compared with that of seasonal precipitation variability.

Summary.—Increased groundwater withdrawals since 2000 do not contribute to observed lake water level fluctuations. Investigation of the hydrogeology, groundwater use, and water budget of the Moreau Lake area has provided multiple lines of evidence that indicate that increased groundwater pumpage since 2000 has no discernable effect on Moreau Lake water levels. To summarize the evidence:

The amount of precipitation available for groundwater recharge varies during the annual hydrologic cycle (fig. 25). During the growing season, much of the precipitation is taken up by vegetation and transpired back to the atmosphere or evaporated from the land surface; groundwater recharge is sporadic and limited. Vegetation is largely dormant during the nongrowing season, which means more water is available for surface runoff or groundwater recharge from precipitation or during a winter thaw when there is snowmelt. Given this seasonal shift in the water budget, dividing annual precipitation data into growing season (May through October) and nongrowing season (November through April) periods was selected for comparisons with changes in lake water levels.

Seasonal precipitation departures from median values during the past 78 years (1944–2021) indicated several periods largely below or above the median within each season (fig. 29). The most striking period was the 1960s drought: both growing and nongrowing seasons were deficient for 11 years (1961–71), except for the nongrowing season of 1968–69. Reconstructed precipitation records from tree-ring analysis (Pederson and others, 2013) indicated that this drought was the most extreme dry period in the Hudson Valley area since at least 1531 (about 490 years). Other periods that stand out included the consistently wet (both seasons) mid-1970s, wet nongrowing seasons from 1948–49 to 1954–55, wet growing seasons from 2003 to 2013, and dry nongrowing seasons from 2008–09 to 2017–18. Long-term changes included generally wet growing seasons and dry nongrowing seasons since about 2000. The record of precipitation data used covered from 1944 to 2021; however, the growing season record has data for 75 years, and the nongrowing season, for 76 years. The missing data are for nonconsecutive years in the early part of the record.

The low water levels at Moreau Lake in 2015 and 2016 fall within the extended nongrowing season dry period from 2008–09 to 2017–18 and within the most extreme dry period of 2011–12 to 2015–16. This 5-year period is the driest nongrowing season period on record, even exceeding any consecutive 5-year period during the 1960s drought. The 2011–12 to 2015–16 average departure from the nongrowing season median was −3.9 in., whereas the 1962–63 to 1966–67 average departure was −3.2 in. This 2011–12 to 2015–16 period, when lake water levels were at their lowest, was essentially a nongrowing season drought. Recovery from dry nongrowing seasons was limited from 2012–13 to 2017–18 by growing season precipitation totals that were above the median, but not by more than 1.7 in., except in 2016, which was −3.3 in. below the median.

Ranked bar charts of growing and nongrowing season precipitation data provide some perspective for the 2014–15 to 2015–16 period as well as long-term changes in seasonal precipitation since 2000 (fig. 30). The 2014–15 nongrowing season ranks as the driest within the period of record (76 years). This is significant because most replenishment of the local hydrogeologic system is from nongrowing season precipitation. The 2016 growing season was the driest since 1993 (28 years).

Since 2000, which represents only 28 percent of the precipitation record, 82 percent of growing season precipitation totals have been greater than the median value, an overwhelmingly wet signal. This also includes 50 percent of the top 10 wettest seasons and 75 percent of the top 4 wettest seasons. None of the top 20 driest growing seasons have occurred since 2000.

Nongrowing season precipitation values since 2000 have been below the respective median value 73 percent of the time, but the dry conditions have been punctuated by extremely wet seasons. Along with the 2014–15 nongrowing season being the driest on record, 50 percent of the top 10 driest nongrowing seasons have been since 2000. Two of the top four wettest nongrowing seasons have also been since 2000.

These changes maybe be decadal in nature and not long-term trends. However, regional changes in climatic characteristics have been documented that appear consistent with the findings herein:•

A 55 percent increase in the heaviest precipitation was documented between 1958 and 2016 in the northeastern United States (Easterling and others, 2017, fig. 7.4), which is consistent with more frequent organized clusters of summer thunderstorms and an increase in summer precipitation. Pederson and others (2013) also noted an increase in summer precipitation starting in 2003 in the lower Hudson Valley region.

Water level data from well SA1305, a surrogate for lake water levels, collected from early July 2016 to the end of 2021 provide a basis for evaluating seasonal precipitation as a driver of lake water level. A comparison of seasonal water level changes at well SA1305 with the seasonal precipitation percent differences from the seasonal medians provides a qualitative picture that makes conceptual sense (fig. 32). Deficits in nongrowing season precipitation correspond to small seasonal lake water level rises, whereas nongrowing season precipitation in excess of the median corresponds to the largest increase in lake water level during the study. During the growing season, a precipitation deficit corresponds to the largest seasonal lake water level decline. Conceptually, water levels in the lake and local aquifer should reflect the balance between water inputs and outputs in the local hydrogeologic system.

Plots of seasonal lake water level change as a function of seasonal precipitation departure, in percent, from respective seasonal medians mostly show strong linear-regression relations and conceptual consistency (fig. 33). Precipitation during the nongrowing season always resulted in a proportional positive net water level change at Moreau Lake, although the 2021 season approached no net gain. The relation between the nongrowing season precipitation and the lake water level change shows a strong positive linear relation (with a coefficient of determination [R²] equal to 0.95) from 2017–18 to 2020–21 and the 1968–69 nongrowing season data:Δh_NGS = 0.071 × PDSPM + 2.10,(5)where

Lake water level elevations measured from July 1968 to July 1969 by Rist-Frost Associates (1969) encompassed the full 1969 nongrowing season. The nongrowing season relation is likely as strong as it is because evapotranspiration is very low during this season, such that most precipitation in the contributing area (mostly the mountainside) eventually ends up in the lake or underlying sand and gravel aquifer either from groundwater flow or surface runoff from rainfall events or periods of snowmelt. This relation also indicates that nongrowing season precipitation inputs typically exceed downgradient groundwater losses from the lake and aquifer system.

The relation of lake water level change as a function of growing season precipitation has two parts (fig. 33): (1) a strong positive linear relation when seasonal precipitation is less than or equal to about the +8 percent precipitation departure from the seasonal median and (2) an approximate zero-slope plateau (variable data, but perhaps weakly positive) response at greater than +8 percent departures. During the growing season, water losses increase with warmer temperatures, and groundwater outflows from the lake and aquifer are supplemented by water losses from transpiration from vegetation and increases in evaporation (fig. 25). The seasonal increase in water loss always exceeded precipitation input, resulting in net growing season declines in lake water level during the study period.

The regression equation (R² = 1.0) for seasons with precipitation less than +8 percent departure from the median is as follows:Δh_GS = 0.1537 × PDSPM − 2.12,(6)whereΔh_GS

is the growing season lake water level change, in feet, and

The scatter is large and data points are too few to generate a meaningful regression. Instead, the water level responses of the three points (years 2017, 2019, and 2021) were averaged (−0.80 ft), and the zero slope (flat) line was added as an approximation of the uncertain but limited response to further increases in growing season precipitation. This response indicates growing season lake-level response is relatively insensitive to precipitation variation from about +8 up to +45 percent departures.

The most likely reason for the minimal or nonresponse of lake levels during growing seasons with precipitation greater than or equal to +8 percent from the seasonal median is the distribution of precipitation within the growing season. Months of deficient rainfall (less than the median) during the months of highest evapotranspiration (May through September; fig. 24) would maximize the negative effect of evapotranspiration on lake level, similar to deficient growing-season precipitation in figure 33. When monthly precipitation of wet growing season years (2017, 2019, and 2021) are plotted over median precipitation values and estimates of potential evapotranspiration (fig. 24), it is evident that monthly rainfall was below median precipitation and below potential evapotranspiration during August and September in 2017 and during July, August, and September in 2019. In 2021, monthly precipitation was above median precipitation and potential evapotranspiration for every month except May, and it was below the higher of the two potential evapotranspiration estimates in August. Thus, the 2021 growing season point on figure 33 is probably the most representative of a typical wet growing season lake level response of the three seasons. The least representative is the 2019 growing season, which was deficient in rainfall during July through September but above median precipitation during May, June, and October. This is consistent with the 2019 growing season point plotting lowest (least positive lake level response) of the three wet growing seasons depicted in figure 33.

Two other factors may cause departures from ideal seasonal conditions: hotter than average growing seasons and disparity between the calendar-defined growing and nongrowing seasons and the start and end of groundwater level recession and recharge hydrologic regimes. Increases in evapotranspiration (transpiration and evaporation) are expected with hotter-than-average temperatures and thus might limit groundwater-level recovery or increase groundwater level declines. Total annual GDD50 is one measure of how hot (or long) a growing season was (fig. 31). However, this effect is subtle or not evident among the wettest growing seasons (2019 and 2021). 2019 was the coolest growing season (lowest GDD50) and 2021 was the second hottest. A more likely factor in the scatter of points about the regression lines (fig. 33) are annual variations in the start and end of hydrologic regimes (periods of groundwater or lake water level recession and recharge) relative to the calendar start and end of the growing season as defined in this report. These periods are evident at well SA1305 at Moreau Lake (fig. 34) and in many groundwater level records where depths to water are not large (less than about 20 ft to water); for local examples, refer to wells SA1285 (U.S. Geological Survey, 2023a) and WR 80 (U.S. Geological Survey, 2023c).

The annual period of groundwater recession generally corresponds to the growing season and is characterized by generally linear declines in water level, which are typically punctuated by minor spikes during rainfalls that, if large enough, result in an upward shift in the recession line that is parallel to the previous one. Exceptions are major summertime rainfall events (organized clusters of thunderstorms), such as those in July and August 2021, that likely resulted in significant runoff from the mountainside, causing sustained rises in water levels (for about 1 week each) before recession resumed.

The annual period of recharge generally corresponds to the nongrowing season and is characterized by prolonged, convex upward groundwater level rises followed by much slower recession rates; refer to, for example, 2018 and 2019 (fig. 34C and D). The mountainside adjacent to the lake means that there is a large slope of poorly permeable bedrock that will continue to provide surface runoff and shallow groundwater downslope to the lake or aquifer for an extended time after a rainfall or snowmelt event, which results in a hydrograph peak in the form of a step-up and gradual decline rather than a well-defined peak with steep upward and downward limbs.

Offsets between the start and end of annual recession and recharge hydrologic regimes relative to the growing and nongrowing seasons are evident in figure 34. The start of the recession regime, as observed from 2017 to 2021, was only delayed relative to the assumed May 1 start of the growing season because transpiration from vegetation is a prerequisite for a sustained recession condition. However, recession could potentially be initiated earlier by a lack of precipitation or by a warm spring and early leaf out (early onset of transpiration; U.S. Environmental Protection Agency, undated b). Recession was initiated within a week after the start of the growing season (May 1) in 2018, 2020, and 2021. Delays in recession, however, can be as late as mid-July (as in 2017). Delays are observed in years where regular rainfall during May, June, or even July continues to exceed losses from transpiration and extends the recharge regime. The recharge regime can only start when transpiration has largely ceased and when a rainfall event produces a sustained water-level rise rather than a brief spike. If the recharge regime is offset from the start of the nongrowing season (November 1), it is typically a delay in the start—if there is no significant rainfall or snowmelt, recession conditions will continue. The recharge regime started at the same time as the nongrowing season in 2018 and 2019, but it was delayed by about 1 month in 2016 and 2020; in 2017, it was delayed by more than 2 months into early January 2018. The cause of an early onset of the recharge regime in 2021 was likely related to extremely wet conditions from June to October. As transpiration was ramping down in October, the wet conditions apparently exceeded vegetation needs and allowed recharge to initiate early. First frost was not until late October, so that was not a factor in early onset of the recharge period.

The growing season and nongrowing season precipitation and lake water level relations used here provide a basis for estimating seasonal lake water level changes across the record of precipitation, recognizing that the greatest confidence is within the range of lake water level changes measured during this study. The relations show conceptual consistency (fig. 33). At zero departure from median precipitation (median seasonal precipitation), predicted lake water level changes for the nongrowing (+2.10 ft) and growing season (−2.12 ft) result in no net lake water level change during a full cycle of seasons.

The relations between seasonal precipitation and seasonal lake water level changes established in the “Seasonal Comparison of Continuous Lake Water Level Data (2016–21) With Precipitation Data” section of this report provide an opportunity to hindcast seasonal lake water level changes from the 78-year precipitation record. This allows a look at 2015–16 changes that were of concern as well as earlier lake water level behavior. The nongrowing season data used in the regression cover a wide range of lake water level change (0.46 to 5.09 ft), which encompasses most of the estimates, so there is a reasonably high degree of confidence within that range. The growing season relation has two components: a linear negative component, defined between about −1 and −3 ft of lake water level change within about −10 to +8 percent departures from the median precipitation values, and a general stable seasonal decline of about −0.8 ft for precipitation in excess of +8 percent departure from the median. Those relations were used for hindcasting growing season lake water level changes within the stated precipitation ranges. A lower limit for growing season water level declines was needed because applying the linear relation to the precipitation record resulted in seasonal declines that were excessive (7 to 12 ft during 1963–65 in particular) and that were not supported by surveyed lake water levels in 1964, 1968, and 1969.

Measurement and analysis of linear or near-linear lake water level recession rates from 2016 to 2020 were used to constrain the maximum growing season water level declines for hindcast estimates (fig. 35). Recession rates were calculated for 38 periods ranging from 2 to 27 days in length, and the date for each data point is the start date of the respective recession period (Heisig, 2026b). Recession is the result of (1) lake water and groundwater flow toward discharge areas to the east-southeast, (2) evaporation from the lake surface, and (3) transpiration by vegetation from low-elevation areas around the lake shore. Recession rates were compared with the number of days of recession measurement, the starting water level elevation of recession measurement, and the numerical calendar day. The lake water level recession rate as a function of numerical calendar day (fig. 36) shows maximum recession rates increasing early in the season toward a peak rate of 0.03 foot per day in the second half of July, followed by slowly decreasing rates to the end of the season. This pattern matches that of maximum potential evapotranspiration (fig. 24).

The upper bound of recession rates during the growing season was used to calculate a maximum recession amount. Linear segments of maximum recession rates were identified on the well SA1305 hydrograph (fig. 3), and each was averaged to obtain a representative maximum rate that was then multiplied by the number of days in the segment to determine the water level decline. The calculated water level declines were then summed for a seasonal maximum decline total, which is 4.76 ft. This estimate was considered to be a maximum because it assumes no precipitation during the season. This value was applied to the growing season hindcast as the maximum expected growing season decline for any seasonal rainfall amount that resulted in a water level decline estimate greater than 4.76 ft.

The resulting hindcast of seasonal lake water level change during the past 78 years is depicted in figure 37. Nongrowing season lake water level responses generally paralleled nongrowing season precipitation because a single regression relation defined the lake water level response and the data points it was based upon covered a wide range of lake water level fluctuation (fig. 33). The resulting nongrowing season hindcasts were nearly always a rise in lake water level (93 percent). It is notable that the 2014–15 hindcast was the most negative (−0.4 ft) for a nongrowing season during the 76-year period of record.

The growing season lake water level response (fig. 33), however, departs from growing season precipitation totals because the upper boundary condition (plateau or minor positive response) and steep regression relation (lower boundary) result in lake water level responses that are insensitive (do not proportionally rise) to precipitation inputs +8 percent or greater above the median, but that are very sensitive to deficient seasonal precipitation. As a result, growing season lake water level responses are frequently constrained (censored) at the upper bound (flat response) and at lower bound (too large a response).

The upper bound (minimum) lake water level decline during the growing season plateaus at about −0.8 ft at a low threshold of +8 percent or greater precipitation departure from the seasonal median (fig. 33). From 1946 to 1980, this upper bound was reached 9 percent of the time (three times; fig. 37). From 1981 to 2021, the upper bound was reached 58 percent of the time (23 times). This change indicates generally wetter growing season conditions during the past 40 years than the early part of the period of record.

During dry conditions, the steep slope of the growing season regression relation indicates a greater sensitivity in lake water level response to deficient precipitation than during the nongrowing season (fig. 33). There is a striking difference in the occurrence of dry growing season maximum lake water level declines during the period of record (fig. 37). From 1956 to 1971, when dry growing season conditions prevailed, the maximum seasonal lake water level decline (−4.76 ft) was reached 81 percent of the time. In contrast, this lower bound has not been reached in the past 27 growing seasons (since 1993).

In summary, hindcasts provide estimates of lake water level fluctuations for Moreau Lake during the past 78 years. Nongrowing season lake water level responses basically mirror precipitation inputs, whereas growing season lake water level responses differ, in that (1) lake water levels are more sensitive to dry conditions (decline faster) than during the nongrowing season (these declines are constrained to a maximum seasonal decline [−4.76 ft] in light of historic lake water level data), and (2) lake water levels are insensitive to any inputs wetter than slightly above median precipitation and, specifically, are constrained by a minimum seasonal decline of no higher than −0.80 ft. As a result, the upper bound is frequently reached during wet growing seasons (mostly since 1980). The lower bound is frequently reached during dry growing seasons (from about 1950 to 1971). Last, the nongrowing season hindcast for 2014–15, the period of most recent concern at the lake, indicated a −0.4 ft seasonal lake water level decline, the most negative of the few years with negative net change during the 78-year period of record.