Brief History of the Principal Aquifers

Before the publication of the Ground Water Atlas of the United States, the USGS RASA program published summaries of 24 large regional aquifers in the United States (Miller, 1999), including the Virgin Islands and Puerto Rico. The Ground Water Atlas of the United States expanded on those studies and added aquifers that were regionally important for water supply but not necessarily widespread (U.S. Geological Survey, 2000b). Aquifers in the study were organized by rock/sediment type into six categories: unconsolidated sand and gravel aquifers, semiconsolidated sand aquifers, sandstone aquifers, basaltic and other volcanic-rock aquifers, carbonate-rock aquifers, and mixed sandstone and carbonate-rock aquifers. A total of 61 principal aquifers, plus the glacial deposit aquifers that overlie bedrock in many areas, are described in the Ground Water Atlas of the United States (U.S. Geological Survey, 2000b). Individual chapters (or “Segments”) were published between 1990 and 1999, and the final collection was published in 2000. A poster-sized map of the 62 principal aquifers was published in 2003 at a scale of 1:5,000,000 (U.S. Geological Survey, 2003).

The Ground Water Atlas of the United States describes suggested uses of the information presented, which was intended to be an overview of aquifer conditions useful as a reference work for planners, consultants, teachers, and hydrologists in other countries for learning about the hydrogeology of the United States. The list of uses for which the atlas is not intended is longer, pointing out that local conditions should not be interpreted from the maps but that finer scale data sources should be consulted instead (Miller, 1999). The metadata for the online datasets of the principal aquifers contains use constraints that the data not be used at scales larger than 1:2,500,000 (for example, see U.S. Geological Survey, 2000c). Although the recommendation that the principal aquifer boundaries published in these works not be used for regional-scale analyses, the principal aquifers have been used for several decades as useful constructs around which to describe and analyze the groundwater resources of the Nation (DeSimone and others, 2014; Qi and Harris, 2017; Stanton and others, 2017; Lovelace and others, 2020a, b).

Large regional studies completed by the USGS, such as the RGA studies, have generally included a remapping of the principal aquifers to incorporate more local-scale information and to refine the definitions of the principal aquifers themselves (Faunt, 2009; Qi, 2010; Burns and others, 2011; Thamke and others, 2014; Williams and Kuniansky, 2016). To date, however, these updates have only been done to a few of the principal aquifers and have not been collected into a single resource for convenient access. The desirability of using the principal aquifer delineations to interpret data at a regional or multistate scale (such as 1:100,000 to 1:2,000,000), and the proliferation of digital geographic analysis tools, indicates a need for more comprehensive finer scale revisions of principal aquifers that were not the subject of an RGA study. This need resulted in the current investigation into methods to revise the boundaries of the principal aquifers in a way that allows for analysis of related information within the context of these aquifers at a finer scale and the move towards nationally consistent finer scale spatial datasets.

Some challenges with reassessing the principal aquifers are related to the manner in which the segments or chapters were originally analyzed. Because of the scale of the original project, the collection of 13 segments has more than 15 authors, and each segment has 1 to 3 authors who were free to interpret the definition of “principal aquifer” and used whatever data sources were available to them in those States. Geologic data were not necessarily organized in the same manner for adjacent States, so the definitions of the principal aquifers sometimes have substantial state-line mapping discontinuities. As noted above, some of the principal aquifers that were investigated in an RGA study were given revised definitions, in part to rectify the state-line mapping discontinuities and to reflect better, local-scale information (for example, the revisions to the Lower Tertiary and Upper Cretaceous aquifers [Thamke and others, 2014]).

One characteristic of the principal aquifers in the Ground Water Atlas of the United States is that they were generally based on lithology, and sometimes also on time-stratigraphic units (U.S. Geological Survey, 2000b). Although the general categories of principal aquifers are based on lithology, each aquifer may have definitions that are based solely on lithology within the segment area, or they may also include time-stratigraphic definitions; for example, the Cambrian-Ordovician aquifer system is mostly sandstone (but not entirely) and also is defined by a specific time-stratigraphic sequence. The Silurian-Devonian aquifers (also time-stratigraphic) are all carbonate rocks. The upper carbonate aquifer only has a lithologic definition. Also, definitions of these aquifers were often heavily affected by state-level hydrogeologic publications and local usage. The lithology categories are often not necessarily “pure,” and mixtures of lithologies are common within some of the aquifer definitions, beyond the officially mixed category of sandstone and carbonate-rock aquifers.

Problem Statement

The current two-dimensional delineations of many of the principal aquifers are not sufficiently accurate for regional and local-scale hydrologic analysis and do not meet the needs of the 2022 USGS IWP science program priorities (Miller and others, 2022). Therefore, the primary goal of this project is developing a method for the reanalysis or revising of the principal aquifers to a finer resolution and applying that method to an initial set of principal aquifers to test the workflow. A new generation of geospatial data with the new principal aquifer delineations also is needed, and this work includes a stand-alone geographic information system dataset of the revised principal aquifer polygons as an accompanying publication (Nielsen, 2023). A secondary goal of a new workflow for improving the principal aquifer delineations is identifying and fixing inconsistencies with the original definitions as they are applied across State boundaries and applying modern workflows to meet current needs for quantitative assessment, possibly consolidating or dividing some of the units if warranted. Although it is recognized that all the principal aquifers have a three-dimensional component, evaluating the three-dimensional boundaries is beyond the scope of the current project.

Since the original work to publish the principal aquifers was done, first with the Ground Water Atlas of the United States (U.S. Geological Survey, 2000b) and then the poster-sized map derived from the atlas, several of these aquifer systems were more fully investigated by the Ground-Water Resources Program RGA studies. The aquifers that had RGA studies have generally much more detailed spatial boundaries defined, and most of the studies developed three-dimensional surfaces to use in an aquifer-wide groundwater flow model. These datasets would be considered first for the upgrade to a more refined principal aquifer dataset, but most of the 62 aquifers identified as principal aquifers have not had this level of study. Of the 62 original principal aquifers, 18 have had further in-depth study through RGA studies, 5 were partially studied in an RGA study, 38 have had no nationally coordinated further study (although could be part of local-scale studies), and 1 was started but not finished.

Purpose and Scope

This report documents a pilot workflow to create a set of finer scale principal aquifer boundaries using data from the U.S. Geological Survey State Geologic Map Compilation (SGMC) geodatabase (Horton, 2017) and decision points to evaluate if a redefinition of the principal aquifers themselves would be warranted in a version 2 dataset. The workflow was applied to four principal aquifers in the upper Midwest. The report presents the source information, analysis of the original definitions of the principal aquifers, and resulting refined definitions and boundaries of these aquifers, as well as considerations for the further application of the workflow to more principal aquifers in the future. Maps of the original and refined principal aquifers are given, and a prototype geospatial dataset to make those data accessible and usable is described in the report. A description of each of the attributes in the geospatial data also is included.

Resources Used

Each analysis began with the national-level principal aquifers dataset (U.S. Geological Survey, 2003), which includes the surface extent of each aquifer. The Ground Water Atlas of the United States segments (U.S. Geological Survey, 2000b) include maps for many of the principal aquifers that depict the extent of the aquifer at depth as well (these were not included in the national-scale map compilation), many of which are available as geospatial datasets. Each of the aquifers in this group has a geodatabase available online (U.S. Geological Survey, 2000a, c, d, e). If further work was done for RGA studies, those datasets also were used. For the current pilot work, only the Lake Michigan Basin groundwater model (Lampe, 2009; Feinstein and others, 2010) was part of an RGA study, which included part of the Cambrian-Ordovician aquifer and part of the Silurian-Devonian aquifer. For the Cambrian-Ordovician and Silurian-Devonian aquifers, the Ground Water Atlas of the United States (U.S. Geological Survey, 2000b) used earlier RASA studies, which included surface elevation maps (such as Mandle and Kontis [1992] for the Cambrian-Ordovician aquifer) to show contours of the aquifer surfaces and which are included in the geodatabase on the USGS Water Mission Area National Spatial Data Infrastructure Node (U.S. Geological Survey, 2015a, b).

The principal source used here for the finer scale spatial boundaries was the SGMC geodatabase (Horton, 2017), which is a national-level spatial database of 48 State geologic maps that range in scale from 1:50,000 to 1:1,000,000. An important feature of the SGMC geodatabase is that each rock unit contains lithology information (primary and secondary) as well as age (minimum and maximum age ranges, in terms of the major divisions on the geologic time scale) These data were critical to applying the definitions of the original principal aquifers to these finer scale units.

An additional ancillary source of information was the secondary hydrogeologic regions (SHRs; Belitz and others, 2018). The SHRs categorized the spaces between the mapped principal aquifers. The SHRs may be geologic units that can transmit small amounts of water, or they may act as confining units between aquifers. The SHRs also are based on a database of the Geologic Map of North America (Garrity and Soller, 2009), and the SGMC units and represent the hydrogeologic units between principal aquifers in terms of lithology and whether they overlie a principal aquifer or lie underneath a surficial aquifer.

Several other local resources were brought to bear on the principal aquifer extent analysis, including publications about the extent, age, and makeup of specific geologic units that compose these aquifers and some additional groundwater model references for areas within each of these principal aquifers.

General Analysis Methodology for Geospatial Units

To be consistent about the analysis of the potential new principal aquifer boundaries, some rules to evaluate the mapping strategy and aquifer definitions were developed:

For each principal aquifer investigated, the first step was to read and review the segment of the Ground Water Atlas of the United States (U.S. Geological Survey, 2000b) associated with the principal aquifer and look at the rationale behind how the principal aquifer was defined lithologically and stratigraphically. If inconsistencies between the rationale and what was delineated were detected, that principal aquifer was investigated more carefully. In some cases, the distinction between two principal aquifers seemed somewhat arbitrary, which justified a reevaluation of the definition. Rethinking the definition was done in consultation with other geologic publications to determine if the approach was justified. Some revisions of the aquifers after the publication of the Ground Water Atlas were made in preparing the printed map (U.S. Geological Survey, 2003), and are documented in the metadata for the 1:5,000,000-scale data (available at https://water.usgs.gov/GIS/metadata/usgswrd/XML/aquifers_us.xml) for the printed map (U.S. Geological Survey, 2003).

Once the background analysis was completed, the geospatial units from the principal aquifer map and the belowground areas were put on a map. The SGMC data were overlain on the principal aquifer outline. An ipython (version 3.6; Pérez and Granger, 2007) jupyter notebook (Granger and Pérez, 2021) using the geopandas package (Jordahl and others, 2020) was developed to use the general definition of the principal aquifer (based on the States covered, lithology, and age of the units) to select polygons from the SGMC dataset. The resulting selection of geologic units was overlaid on the original principal aquifer using ArcMap (version 10.8). The principal aquifer definitions in the Ground Water Atlas of the United States (U.S. Geological Survey, 2000b) generally include the formations and formation groups, which were cross-checked against the SGMC units selected using the lithology, stratigraphic charts in the Ground Water Atlas, and age data. The resulting candidate selection of SGMC polygons was used to create a large shapefile of the SGMC units tagged with the principal aquifer attributes.

Once the candidate polygons from the SGMC were selected and compared to the original principal aquifer extent, they were hand-edited, making refinements based on the “Unit_Name” field, which principal aquifer was mapped at that location or nearby, and after researching the tectonic structural setting that applies to the geologic units composing the aquifer. The tectonic structure setting was useful for understanding whether or not the unit mapped at the surface is buried by younger units at their edge or if the unit mapped at the surface is at an erosional edge. Because most of the aquifers in this study area are sedimentary, this consideration was important for making a judgement of whether a small, mapped polygon was likely to be part of a wider, regionally continuous aquifer or aquifer system or was an isolated erosional remnant with older rocks underneath that are not part of the aquifer system. Small isolated erosional remnant polygons were generally removed from the set for a given aquifer.

In some cases, decisions were made about individual formations in specific States as to whether they should be included or excluded from the SGMC selection. There were some cases where the state geologic maps differed somewhat in how they mapped rocks that were assigned to a particular time period or range in lithologies such that one interpretation would cut the aquifer boundary at the state line. If possible, this was avoided to keep the updated aquifer boundary as close to the originally mapped boundary as possible. Another important consideration in the analysis was the definition of what units separate adjacent principal aquifers from one another, and whether these decisions were consistently applied across the extent of each principal aquifer.

A graphical depiction of these steps outlined previously is shown in an abbreviated format as a flow chart in figure 2. Part A includes the steps to evaluate how the original principal aquifer was defined, determine whether any adjustments may be warranted, and make the initial selection of proposed geologic units from the SGMC database (Horton, 2017) that should be included in the update. Part B includes the refining of the selection of geologic units based on the adjacent relations of other principal aquifers and the tectonic/structural setting of the candidate units to finalizing the units to be included in the updated set of units. Part C on the flow chart is not expanded but includes collecting the appropriate metadata for inclusion in a final geographic information system dataset. The application of these principles and steps is illustrated in the following sections on the analysis of the aquifer boundaries in the selected aquifers in this pilot study area.

Cambrian-Ordovician Aquifer System

The Cambrian-Ordovician aquifer system (fig. 1) is the most heavily used aquifer system in the pilot study area, having 921.3 million gallons per day (Mgal/d) of withdrawals (Lovelace and others, 2020b). It is described primarily in terms of time-stratigraphic units, with a minor focus on lithology. Segment 9 of the Ground Water Atlas of the United States (Olcott, 1992) provides an extensive definition of the aquifer system and covers most of the extent, although a small area of the South is covered in Segment 10 of the Ground Water Atlas of the United States (Lloyd and Lyke, 1995). According to Olcott (1992), the aquifer system is composed of two primary sandstone aquifers (the Mount Simon and Ironton-Galesville aquifers) and a sandstone-dolomite aquifer (the St. Peter-Prairie du Chien-Jordan aquifer) and has some intervening confining units. The top of the aquifer system (where it is covered by other units) is the top of the regional shale-rich Maquoketa Formation, which is at the top of the Ordovician system in the midwestern States. Previous study of the Lake Michigan Basin (Lampe, 2009; Feinstein and others, 2010) included the eastern part of this aquifer system, but refining the aquifer boundaries was not within the scope of that study. The Lake Michigan Basin study modeled the individual aquifers within the Cambrian-Ordovician aquifer system. Geospatial data from that study, including the subsurface extent for the Cambrian-Ordovician aquifer system (U.S. Geological Survey, 2015b) are available at https://water.usgs.gov/GIS/metadata/styles/landingPage/ds33camord_a10.xml.

Some inconsistencies in the application of the stated definition of the Cambrian-Ordovician aquifer system could be removed with the refinement of the aquifer boundaries. The definition in Olcott (1992) references earlier work for the RASA program, in particular, Young (1992) and Mandle and Kontis (1992). The earlier definitions discuss an underlying unit, the Hinckley Sandstone, which is a Precambrian aquifer in Minnesota that is adjacent to the Cambrian Mount Simon aquifer. Young (1992) does not include the Hinckley Sandstone in the Cambrian-Ordovician aquifer system, while Olcott (1992) does. Olcott (1992) also extended the definition to include another Precambrian sandstone aquifer in northern Wisconsin, the Bayfield Group, without much of an explanation. Although in southern Minnesota, the Hinckley Sandstone is in hydraulic connection with the overlying Mount Simon aquifer, an argument could be made that the Hinckley Sandstone does not belong in the Cambrian-Ordovician aquifer system because it is much older and part of a completely different depositional system, the Midcontinent Rift (Ojakangas and others, 2001). Wilson (2012) has removed both the Hinkley Sandstone and Bayfield Group from analysis of the Cambrian-Ordovician aquifer. Ojakangas and others (2001) show the depositional and lithologic relation between the Hinckley Sandstone, Fond du Lac Formation (another sandstone in northern Minnesota lumped in the Cambrian-Ordovician aquifer system), Bayfield Group, and Jacobsville Sandstone in Michigan.

Along the northern edge of the Cambrian-Ordovician aquifer system in Wisconsin (fig. 3), the SGMC units that are sandstones do not extend as far north in some areas, and extend farther in others, portraying a more accurate boundary between the sedimentary aquifer and crystalline bedrock compared to the Ground Water Atlas of the United States (U.S. Geological Survey, 2000b). In Minnesota, the finer detail of the SGMC geodatabase added new areas to the Cambrian-Ordovician aquifer system and removed others, based on the lithology and rock types that the Ground Water Atlas intended to include. Additionally, in the process of selecting units from the SGMC geodatabase that should be included in the finer scale Cambrian-Ordovician aquifer system and the necessity to correlate the units across state lines, discontinuities in the location of geologic units resulting from different approaches to mapping units along the border between Iowa, Wisconsin, and Illinois became evident.

One set of geologic units that could benefit from clarification in the definition of the Cambrian-Ordovician aquifer system is the Galena Dolomite and Platteville Limestone, two carbonate-rich units that sit on the top of the mostly sandstone aquifer system. At the level of detail included in the SGMC geodatabase, in most areas these units are contiguous with other rocks that participate in the Cambrian-Ordovician aquifer system, but in some areas where they occur, they crop out far from other parts of that aquifer. In this pilot analysis of refining the principal aquifer boundaries, the decision was made to include the Galena and (or) Platteville units in the Cambrian-Ordovician aquifer system when they were determined to be contiguous with the rest of that aquifer system, but if they were not contiguous with sandstones in the Cambrian-Ordovician aquifer system, these dolomites were lumped with contiguous carbonates in an adjacent principal aquifer (see discussion on the upper carbonate and Silurian-Devonian system, later in this section). Grouping the Galena Dolomite with the upper carbonate aquifer in southeastern Minnesota is consistent with stratigraphic distinctions in Olcott (1992).

The results of refining the Cambrian-Ordovician aquifer system boundary, using the SGMC geodatabase of geologic units and the logic outlined previously and in figure 2, are shown in figures 3 and 4. Only the surface map of the aquifer boundary was reanalyzed. The specific additions and subtractions from the original boundary are indicated in figure 3, including the removal of the Precambrian Hinckley Sandstone and Bayfield Group, which is discussed in more detail later in this section. A simplified version of the new finer scale extent of the aquifer boundary compared to the old version is shown in figure 4. An important note for both figures 3 and 4 is that the Cambrian-Ordovician aquifer system does extend at depth below the Silurian-Devonian aquifers to the east (under Lake Michigan) and to the southwest (under Iowa and Illinois).

The Cambrian-Ordovician aquifer system is thin and not particularly useful for water supply in some places, and the geologic units can be heterogenous in terms of whether they are locally aquifers or confining units (Ken Bradbury, Wisconsin Geologic and Natural History Survey, oral commun., 2021). The original published information about the Cambrian-Ordovician aquifer system (Olcott, 1992) does not indicate where the individual aquifer units contained in the aquifer system are thin or contain mixed lithologies, which could limit their usefulness to potential users of the data. The approach described in this report can better represent details about heterogeneities such as these, particularly in the metadata describing aquifer characteristics (see next subsection).

Jacobsville Aquifer

The second aquifer in this pilot study for analysis was the Jacobsville aquifer. This aquifer (fig. 1), composed of a Precambrian sandstone in the Upper Peninsula of Michigan, was designated a major aquifer because it is the only source of groundwater where it is present (Olcott, 1992). This unit is part of a series of structurally and diagenetically correlated sandstones in the Precambrian Midcontinent Rift (Ojakangas and others, 2001), discussed in the earlier section on the Cambrian-Ordovician aquifer system, many of which also are locally important sources of groundwater. Although the Midcontinent Rift units have structural, lithologic, and time-stratigraphic coherence, they were not grouped together into one aquifer system in the original principal aquifer designations. For example, the Bayfield Group in northern Wisconsin, which is central to the Midcontinent Rift units, was recognized as an important aquifer but was lumped with a different aquifer system, the Cambrian-Ordovician aquifer system (even though it is not Cambrian-Ordovician in age).

The position of the Jacobsville aquifer in context with the other Midcontinent Rift sandstones is shown in figure 5. For the current analysis, it made more sense in this context to redefine the Jacobsville aquifer and other correlative sandstones as the “Midcontinent Rift sandstone aquifers” (fig. 6), including the Oronto Group, which is a finer grained sedimentary unit in the Midcontinent Rift along the shore of Lake Michigan in the western section of the Upper Peninsula of Michigan but is locally also used as a minor aquifer.

Silurian-Devonian and Upper Carbonate Aquifers

The Silurian-Devonian aquifer and adjacent upper carbonate aquifer (fig. 1) cover a large part of the upper Midwest and accounted for a total of 486.3 Mgal/day of withdrawals in 2015 (Lovelace and others, 2020b). The upper carbonate aquifer is described in Olcott (1992) as part of Segment 9 in the Ground Water Atlas of the United States. About one-half of the Silurian-Devonian aquifer is in Segment 9 of the Ground Water Atlas of the United States (Olcott, 1992), and the other half is in Segment 10 (Lloyd and Lyke, 1995). The upper carbonate aquifer, which exists principally in southern Minnesota and northern Iowa, is defined by lithology alone, but the Silurian-Devonian aquifer is defined primarily by time-stratigraphic units (although the principal lithology is dolomite and limestone). The Silurian-Devonian aquifer is not a wholly contiguous unit but has two major segments: an eastern part surrounding the lower peninsula of Michigan and the Michigan Basin (outcropping in eastern Wisconsin, northeastern Illinois, northern Indiana and Ohio, and northern Michigan) and a western part (outcropping primarily in southwestern Wisconsin, eastern Iowa, and northwestern Illinois; fig. 1). The Silurian-Devonian sediments thin and are eroded away between these two segments. Small erosional remnants also occur on the border between Illinois and Missouri and in central Kentucky.

The upper carbonate aquifer is defined in Minnesota and Iowa only (fig. 1) and fills the northern part of the Hollandale Embayment. It is primarily karstic, with some shaly layers, and is very productive (U.S. Geological Survey, 2000b). Because it is defined by lithology, not stratigraphy, the treatment of geologic units between this principal aquifer and adjacent correlative units is not consistent and may lead to some confusion; for example, the top of the Cambrian-Ordovician aquifer in Wisconsin and points east is within the Maquoketa confining unit (Olcott, 1992), but in Minnesota and Iowa, the Ordovician rocks (some of which are water bearing) extend above and into the upper carbonate aquifer. The Ordovician-age Galena Dolomite and Platteville Limestone are within the series of units within the upper carbonate aquifer in Iowa and Minnesota, even though farther to the east (in Wisconsin and Illinois), they are part of the top of the Cambrian-Ordovician aquifer.

Except for the upper Ordovician rocks, the upper carbonate aquifer is mostly Devonian, as the Silurian is missing from the section in Minnesota. Although the Cedar Valley Limestone is a major part of the upper carbonate aquifer in Minnesota, it is part of the Silurian-Devonian aquifer in Iowa. Thus, the dividing line between what composes the upper carbonate aquifer and what is included in the Silurian-Devonian aquifer is rather arbitrary and could be confusing. Because both aquifers are primarily composed of carbonate rocks that are mostly Silurian and Devonian in age, and the distinction between them is not entirely clear where they are adjacent to one another, it makes some sense to combine them into one larger aquifer system. This could be especially helpful for a regional-scale study that uses a finer resolution geologic dataset, in which deciding where the stratigraphic boundary between two adjacent aquifers or aquifer systems could be a difficult exercise.

The changes as described to the Silurian-Devonian aquifer system, including the addition of the upper carbonate aquifer to the overall system, are shown in figures 7 and 8. Details of additions and subtractions of units using the SGMC geodatabase (Horton, 2017) are shown in figure 7, and the overall changes are illustrated more simply in figure 8.

Units removed from the SGMC selection covering the original Silurian-Devonian aquifer include units that are older than Silurian in age (meaning that the younger Silurian-Devonian rocks have been eroded off the top), units that are primarily shales, and some units that are thin, disconnected pieces/erosional remnants that are not large or widespread enough to participate in a regional aquifer. Units added to the Silurian-Devonian aquifer included the Galena Dolomite and Platteville Limestone in Iowa or Illinois, where they were contiguous with the Silurian-Devonian aquifer but disconnected by a confining unit from the Cambrian-Ordovician aquifer system (a dominantly sandstone aquifer in which they were originally included). This is an example of examining how units at the stratigraphic and geographic edges of the principal aquifer delineations were treated, favoring including units that are contiguous, likely to be hydraulically connected, and lithologically similar and differentiating them from other nearby aquifers if they are not likely to be hydraulically connected with the other aquifer.

Combined Pilot Principal Aquifers—Fine Scale

Using the SGMC data (Horton, 2017) and workflows described previously, the extent of the updated (version 2) finer scale principal aquifers for the upper Midwest (Cambrian-Ordovician aquifer system, Midcontinent Rift sandstone aquifers, Silurian-Devonian-upper carbonate aquifers) are illustrated in figure 9. The Midcontinent Rift sandstone aquifers and Silurian-Devonian-upper carbonate aquifers are newly defined in this report.

The workflow prototyped in this study could be applied to areas recently the target of USGS regional models. Other principal aquifers involved in RGA studies that produced finer scale aquifer boundaries, useful for redefining principal aquifer extents at scales comparable to this pilot project, include the Upper Cretaceous aquifers (Thamke and others, 2014), High Plains aquifer (Qi, 2010), Columbia Plateau basaltic-rock and Columbia Plateau basin-fill aquifers (Burns and others, 2011), Central Valley aquifer system (Faunt, 2009), and the Floridan aquifer system (Williams and Kuniansky, 2016), among others (fig. 10). In future analyses, the refined spatial extents of these aquifer systems would be appropriate to add to the prototype analysis included in this pilot study as additions to the version 2 principal aquifer dataset.