Purpose and Scope

The purpose of this report is to describe the data sources and methods used to create digital hydrogeologic maps in the eight study areas in upstate New York. The scope of this effort included the interpretation and evaluation of the hydrogeologic framework derived from light detection and ranging (lidar) datasets, surficial geology, soil descriptions, aquifer maps, and lithologic well logs. The surficial geology; thickness of the surficial aquifers (primarily sand and gravel); and tops of the confining units (primarily lacustrine silt and clay), confined aquifers (primarily sand and gravel), and bedrock (primarily shale and sandstone) surfaces were interpolated and digitally mapped.

Hydrogeologic Settings of Valley-Fill Aquifers

During multiple continental glaciations beginning about 120,000 years ago, glacial ice deeply eroded existing stream and river valleys throughout the northeastern United States. Almost all of the unconsolidated sediments filling the valleys in the study areas were deposited during the most recent deglaciation, which occurred between 18,500 and 13,000 years ago (Cadwell, 1986). The uplands are typically covered by till with local bedrock outcrops. The till consists of unsorted silt, clay, sand, and clasts directly deposited by glacial ice as lodgement till and during deglaciation as ablation till. The valley fill as described by Randall (2001) primarily consists of three depositional facies that are locally successive but are transgressive over longer distances (tens of kilometers). The three depositional facies are the following: (1) early deglacial, proximal, or ice-contact facies deposited after the Wisconsin glacial maximum, composed of predominantly gravel, sand, and silt, varying widely in degree of sorting, and deposited close to active ice or amid abundant stagnant ice masses (referred to as kame deposits in this report, and kame moraines where deposits are contiguous); (2) mid-deglacial distal facies of lacustrine silt and clay deposited in moderately large bodies of water (proglacial lakes) at distance from the active glacial ice; and (3) late deglacial and postglacial facies of surficial gravel, sand, and silt that can be well sorted, deposited as alluvium, alluvial fans, and outwash, most commonly atop early and mid-deglacial facies (Randall, 2001, p. B12). All three depositional facies are present in many postglacial valleys; however, in some valleys, only one or two of the depositional facies are present. The lacustrine facies generally are the most extensive and thickest of the three facies. Figure 2 provides definitions of the geologic units that are referenced throughout the text and figures.

The hydrogeologic framework of the valley-fill aquifers resulting from this depositional sequence is depicted in figure 3. Where surficial alluvial and ice-contact deposits are saturated, they form an unconfined sand and gravel aquifer. The base of the unconfined sand and gravel aquifer is in contact with either the top of the lacustrine unit or bedrock. The lacustrine deposits of silt and clay form a confining to semiconfining unit for underlying ice-contact deposits of sand and gravel.

Raster Surface Interpolation and Evaluation for the Villages of Ellenville and Wurtsboro and Hamlets of Woodbourne and South Fallsburg Study Area

Digital surfaces were created by using an array of point and contour inputs and common interpolation tools including Topo to Raster, Natural Neighbors, Kriging, and Bayesian Kriging (Esri, 2019). Interpolations were calibrated and validated for each dataset by using a stepwise statistical-testing approach (Jahn, 2020). Interpolation tools tested include those listed above and incorporated multiple kriging (simple, universal, and ordinary) and model types (stable, circular, spherical, exponential, rational quadratic, Gaussian, K-Bessel, and J-Bessel) (Esri, 2019). Each kriging and model type produces a different interpolation of the data tested in this study.

The testing consisted of excluding 10 percent of all data points from the initial interpolation. A raster was then generated for the remaining 90 percent of points. Using this newly generated raster, the absolute difference was calculated at every data point. The models with the smallest amount of mean and median absolute difference were chosen on the basis of statistical results, visual inspection, and representation of local hydrogeology. In cases where the majority data type was contour lines (which was the result only for the bedrock surface in the Cortland study area) (fig. 1), the Topo to Raster interpolation method was used because both contour and point data types can be incorporated with this method.

Land Surface

The land-surface altitude was compiled from multiple New York State lidar dataset collections obtained from the New York State GIS Clearinghouse (New York State Geographic Information Systems Clearinghouse, undated). The lidar datasets were developed by multiple agencies (USGS, NYSDEC, and New York State Department of Environmental Protection) or by the county of interest. These collections all have a resolution of 1, 2, or 3 meters (m). A lidar-derived, land-surface altitude map for the EWWSF study area is shown in figure 4. The land-surface altitude coverage shown is a mosaic dataset of three individual lidar sources (Terrapoint USA, 2008; New York State Department of Environmental Preservation, 2009; U.S. Geological Survey, 2015). To match the resolution of the other interpolated geologic layers, the land-surface raster cell size was up-scaled to a 38.1 m (125-foot [ft]) discretization. The land-surface altitude for each grid cell was assigned the minimum lidar value within the cell. Cell altitudes were calculated in this fashion for groundwater-modeling purposes.

Surficial Geology

Surficial geology was determined from published 1:24,000-scale maps (from Anderson and others, 1982a; Reynolds, 2007) and supplemented with data layers from the SSURGO, lidar, topographic maps, and lithologic well logs where published surficial geology was limited. In study areas with no published surficial geology, the geology was approximated by using SSURGO data layers, lidar, topographic maps, and lithologic well logs. The extent of surficial sand and gravel aquifers in the valleys and their contact with the till-covered bedrock uplands was defined by the extent of contiguous valley-fill sediments of alluvium, alluvial-fan deposits, outwash, deltaic and lacustrine sand, and kame deposits (fig. 5). Peat and muck in contact with sand and gravel deposits in the valleys were included in the extent. Contiguous deposits constituting kame moraine were also included, despite these deposits often containing discontinuous till, silt, and clay beds. Till moraine was excluded even though these deposits contain scattered beds of sand and gravel. An example of the surficial geology and its relation with the subsurface geology near Wurtsboro in the EWWSF study area is presented in figures 6 and 7. The surficial geology shown in figure 6 is underlain by a hillshade base layer derived from lidar (Terrapoint USA, 2008). This hillshade effect highlights the steep topography prevalent throughout the study area.

Thickness of Surficial Sand and Gravel Unconfined Aquifer

The thickness of the surficial sand and gravel aquifer in the valleys was defined by subtracting the top of the confining lacustrine silt and clay, where present, from the land-surface altitude. Bedrock-surface altitude was subtracted from the land-surface altitude if the confining unit was not present. Lithologic well logs from NYSDEC, NYSDOT, and USGS were individually evaluated to determine the altitude of the top of the confining unit (if present). Hydrogeologic sections were also used from previous USGS reports for the EWWSF study area (Anderson and others, 1982a; Reynolds, 2007).

The thickness of the surficial sand and gravel aquifer for the EWWSF study area is shown in figure 8. Where lacustrine silt and clay is mapped at the surface, it is assumed that there is no aquifer present at that location (fig. 8B). In areas where a confining unit was not identified (fig. 9), it is assumed that the thickness of the surficial sand and gravel aquifer represents the entire thickness between land surface and bedrock surface.

Top of Lacustrine Silt and Clay Confining Unit

The top of the lacustrine silt and clay confining unit was defined by the contact with overlying surficial sand and gravel aquifer or land surface where the silt and clay were exposed at the surface. As stated above, the confining-unit extent and thickness were primarily obtained from lithologic well logs and published hydrogeologic sections. The top altitude of the confining unit was interpolated by using the Topo to Raster tool. For example, the spatial extent and altitude of the top of the confining unit in the EWWSF study area are shown in figure 9. The confined extent illustrates where a thickness of at least 1 m of laterally continuous lacustrine silt and clay was present. This extent was established from published reports and sections (Anderson and others, 1982a; Reynolds, 2007) and lithologic well logs.

Top of Sand and Gravel Confined Aquifer

The top surface of the sand and gravel confined aquifer was defined by the contact with the overlying confining unit. This surface was interpolated by using the Topo to Raster tool. The spatial extent and altitude of the top of the sand and gravel confined aquifer in the EWWSF study area are shown in figure 10. The confined unit was assumed to have a minimum thickness of 1 m in areas with no nearby sections or lithologic well logs. The bottom of this unit is inferred to be the top of bedrock, although till deposits may locally be present above the bedrock.

Bedrock Surface

The bedrock surface defines the bottom of unconsolidated deposits. Although the completed bedrock-surface maps are presented as a singular continuous surface (fig. 11), different interpolation methods were applied in the valleys and the uplands. Although the methods differed, the same interpolation tool (Topo to Raster) was used for both the valleys and uplands.

In the valleys, published hydrogeologic sections and lithologic well logs were used to create the bedrock surface. Glaciated valleys are commonly U-shaped; therefore, when extrapolating the bedrock surface from sections and logs, the following technique was applied to ensure that the interpolated bedrock surface approximated a U-shape. A series of points were generated 61 m (200 ft) apart down the center of the entire valley (fig. 12). These points, referred to as smoothing points, were each assigned a bedrock-surface altitude based on nearby lithologic well logs or published cross sections. In areas where lithology logs and sections were unavailable, the bedrock-surface altitude was linearly interpolated between two known values.

The process of adding points and assigning altitudes within the valley was also used along the valley walls. To best interpolate a U-shaped surface, the point inputs must reflect a continuous rise in bedrock altitude from the center of the valley to the valley walls. Along the valley walls, it was assumed the bedrock surface was approximately 3 m (10 ft) below land surface in areas that lacked hydrogeologic sections and lithologic well log data. A thickness of 3 m (10 ft) was used to accommodate the minimum thicknesses of the other layers in the hydrogeologic framework, which was done for groundwater-modeling purposes. The exception to this application was if there was a well nearby that indicated an appreciable depth (greater than 3 m) to bedrock. That depth to bedrock value was then assigned to the valley wall edge points near the well location.

In areas within the valley where little to no bedrock data were available, the altitude was obtained and extrapolated from the closest known values. In these cases, it was assumed that the bedrock-surface altitude was not constant. For example, valley point altitudes parallel to the valley (perpendicular to geologic section lines) were obtained and extrapolated from each geologic section (fig. 12). This process was necessary so that the Topo to Raster interpolation could generate the assumed U-shaped valley bottom.

In some local upland areas, especially where data were lacking, casing depths in wells were used to estimate depth to bedrock. In general, there is a consistent relation between casing depth and depth to bedrock. Drillers commonly set casing 3 m (10 ft) into bedrock. Thus, the assumption was made that in wells where the total well depth was at least 6 m (20 ft) greater than the casing depth, the unconsolidated thickness was estimated to be 3 m (10 ft) less than casing length.

In the upland area, published surficial geologic maps and SSURGO data layers were used with lithologic well logs to define the top of the bedrock surface. Instead of using bedrock-altitude data directly for the interpolations, the thickness of unconsolidated deposits was interpolated and then subtracted from the land-surface altitude to create the bedrock-surface altitude. The reason this method was used is that the steep and variable upland topography causes interpolated bedrock surfaces to be substantially above or below land surface in areas where data were sparse. The surficial-geology map units designated as “rock” (from Anderson and others, 1982a; Reynolds, 2007) and SSURGO map units described as rock outcrop and thin soils were assigned depths to bedrock of 3 m (10 ft) below land surface. Points were added within these surficial and SSURGO polygons in a gridded fashion, spaced approximately 100 m (300 ft) apart. The equidistant spacing of these points was altered depending on the amount of SSURGO bedrock representative polygons within the eight study areas.

Modifications for Cortland, Cincinnatus, and Greene Study Areas

Hydrogeologic mapping of the Cortland, Cincinnatus, and Greene study areas (fig. 1) was generally completed by using the same methods as used for the EWWSF study area. However, variations in available data or site-specific hydrogeologic conditions commonly required minor deviations from the methods described above. The modifications to the methods applied for each study area are described below.

One reason for the slight variation in methods among the eight study areas is the range of data available from previous reports for the study areas. Different data types were available for each study area, which required slightly different methods for each study area to achieve the required hydrogeologic framework. For example, published bedrock contours were available in the Cortland and Green study areas but not in the EWWSF or Cincinnatus study areas. Subsequently, where bedrock contour lines were available, these lines were used as the basis for the bedrock surface raster (Cortland and Greene study areas). In contrast, the major inputs for bedrock surface raster interpolation for EWWSF and Cincinnatus groundwater-flow model areas were lithologic well log data, geologic sections from published reports, SSURGO data layers, and implementation of smoothing points. The large variation in data required different interpolation methods to best fit the data for each study area. The interpolation methods used for these study areas differed from the EWWSF study area, where all surfaces were interpolated using the Topo to Raster tool. Metadata in Woda and others (2022) details the interpolation method for each digital layer of the hydrogeologic framework for each study area.

The Cortland hydrogeologic framework has been extensively studied (Miller and Brooks, 1981; Miller and others, 1998). The published data were adequate to allow for creation of all the necessary digital layers of the hydrogeologic framework without intensive analysis of lithologic well logs. However, the abundance of data within this study area required different methods from those used in other study areas when interpolating the altitude of the hydrogeologic surfaces. For example, the bottom of the surficial sand and gravel aquifer was based on published contour lines of the potentiometric-surface altitude and the unconfined aquifer thickness (Miller and Brooks, 1981, available in Woda and others, 2022). These contour lines were then interpolated to create two separate raster layers. The resulting unconfined aquifer thickness raster was subtracted from the potentiometric-surface altitude raster, and the resultant raster represents the bottom of the surficial sand and gravel aquifer, or the top of the lacustrine silt and clay confining unit (where lacustrine silt and clay are present).

Another difference in the Cortland study area, as compared to the other study areas, is that the use of smoothing points was unnecessary in the main valley because a U-shaped valley bottom was already depicted in published reports (modified from Miller and others, 1998). Slight adjustments to the published contours were made to reflect newly available data. In addition, the confining-unit bottom altitude (and thickness) was derived primarily by using previously published cross sections (Miller and Brooks, 1981).

In contrast to the Cortland study area, no published report data were available for the Cincinnatus study area. Consequently, all layers in the hydrogeologic framework were interpolated by using lithologic well logs, lidar, SSURGO data layers, and geologic interpretation. Where data were unavailable, smoothing points were implemented, as described above for the EWWSF study area.

Data available for the Greene study area (fig. 1) varied, as published report data were available in some parts of the area (Hetcher and others, 2003; Hetcher-Aguila and Miller, 2005), whereas no data were available in other parts. Data from multiple reports were used in the analysis, and in some areas the data conflicted, such as conflicting water-table altitudes or aquifer thicknesses. Where data from the reports conflicted, contours were modified to reflect multiple datasets, with all available data in those parts of the study area taken into consideration.

Smoothing points and valley-edge points were used in areas without existing data. In these cases, smoothing points were given an assumed altitude that best reflected the known characteristics of the valley-fill aquifer.

Modifications for Fishkill-Wappinger Falls, Jamestown, Olean, and Ellicottville Study Areas

Hydrogeologic mapping of the Fishkill-Wappinger Falls, Jamestown, Olean, and Ellicottville study areas was generally completed by using the same methods applied in the EWWSF study area. However, unlike the other four study areas, the hydrogeologic maps developed for these study areas were not applied to the groundwater-flow models. Consequently, methods were modified and changes were applied to both the hydrogeologic framework and its presentation to make the data usable by a wider audience. Notably, data for all four study areas were not mapped to a specific model grid developed for these four study areas. Instead, hydrogeologic surfaces were produced as raster datasets for general use and application.

The interpolated hydrogeologic surfaces for the study areas without models included the altitude of the bedrock surface and the altitude of the top and bottom of the confining unit. These three surfaces, along with the land-surface altitude, allow for the calculation of the thickness of the surficial sand and gravel aquifer (land-surface altitude minus either the altitude of the bedrock surface or the altitude of the top of the confining unit). A calculation of the thickness of the sand and gravel confined aquifer, if present, can also be performed (altitude of the bottom of the confining unit minus the bedrock-surface altitude). Method variations and notable observations by study area are described below.

No published aquifer maps or other hydrogeologic studies were available for the Ellicottville study area (fig. 1). The available lithologic well logs in this area indicated that the valley fill consists primarily of sand and gravel. Relatively thin beds of silt and clay (less than 5 m thick) are discontinuously present at intermediate depths, so no continuous confining unit was recognized, and therefore only the top of the bedrock surface was produced. The thickness of the surficial sand and gravel aquifer was calculated as the difference between the bedrock and land-surface altitudes.

The valley-fill sediments in the Jamestown study area (fig. 1) were extremely thick. The bedrock at its deepest recorded location was about 300 m below land surface. The top of bedrock in these thick valley-fill areas was mapped on the basis of lithologic well logs from oil and gas wells in the Empire State Organized Geologic Information System (ESOGIS) (New York State Museum, 2021). The ESOGIS wells facilitated an update of bedrock surface contour lines mapped by Crain (1960, 1966) and provided an opportunity to interpolate the bedrock altitude in previously unmapped areas. In addition, thick lacustrine units were prevalent throughout the Jamestown study area and in some areas were exposed at land surface, as indicated by the surficial geology. Available data in Anderson and others (1982b) and in lithologic well logs were used to map the upper confining unit. However, thick intervals of the valley fill (in some areas hundreds of meters of thickness) and a lack of data limited any additional hydrogeologic mapping in this area. Therefore, the hydrogeologic framework provided for the Jamestown study area defined the surficial sand and gravel aquifer thickness, the top and bottom of the upper confining unit, and top of the upper sand and gravel confined aquifer. Additional lacustrine confining units and deeper confined aquifers may be present below what was defined.

The Olean study area (fig. 1) featured an abundance of published report data (see Woda and others, 2022), including numerous geologic sections in the main valley area. These sections were used with lithologic well logs to map the extent and thickness of the primary lacustrine confining unit. Additionally, confining units were excluded from areas where kame deposits were mapped at land surface, as was the case for all study areas. The confining-unit extent produced for this study area may not include all areas where a confining unit is present, especially where data were lacking. In addition, data from multiple reports, geologic sections, and lithologic well logs highlighted multiple confining units near the center of the study area. The abundance of available data allowed the extent and altitude of a second confining unit to be mapped. The Olean study area was unique in that the southern part of the area was not covered by the most recent glaciation (Zarriello and Reynolds, 1987). The published surficial geology in this study area (from Zarriello and Reynolds, 1987) was further refined by using the SSURGO mapping dataset.

The Fishkill-Wappinger Falls study area (fig. 1) was distinctive because the hydrogeologic framework differed appreciably from the typical glacial valley-fill system found in most of upstate New York. This difference required an adjustment from the methods used in the other study areas. Notably, the process of smoothing lines and setting the valley edge points to assume a minimum thickness to bedrock was omitted for the Fishkill-Wappinger Falls study area because the bedrock surface did not clearly define a U-shaped valley bordered by uplands. Valley edge points were used to smooth the edges of the interpolation between the valley and upland areas. The bedrock altitude of the valley edge points was determined by using the valley bedrock interpolation, and this interpolation was used as input for the upland bedrock interpolation. In addition, the complex shape of the valley coupled with a lack of data made it difficult to map lacustrine confining units. This difficulty also was described in Reynolds and Calef (2010). Consequently, the lacustrine confining unit was only mapped in one small section in the southern part of the study area on the basis of lithologic well log data. The production wells associated with this study area were near this section; therefore, accurately depicting the confining unit here was a priority. The raster for the bottom of the confining unit was generated by using a constant thickness derived from only one lithologic well log. In areas where the constant thickness indicated that the bottom of the confining unit was below the top of the bedrock, the bottom of the confining unit was assumed to be 1 m higher than the top of bedrock surface.