Abstract Lake Michigan is the world’s sixth largest freshwater lake and has many features in common with oceanic settings, albeit at a smaller scale. All of the constructional features typical of ocean coasts can be found along the shore of Lake Michigan, and it has a shelf-slope system where coastwise rectification of currents, coastal downwelling jets, Coriolis veering of lake currents, benthic nepheloid layer, and density currents have been observed. Unlike ocean coasts, however, the wave climate is predominantly mild, and only a very small lunar tide exists, although other (quasi) periodic water-level fluctuations such as seiches and edge waves do occur. Another significant difference is the occurrence of quasicyclical climatically induced lake-level fluctuations of as much as 2 m (6.6 ft) that greatly influence the way that coastal sediments accumulate. Lastly, the Lake Michigan coast during the late Wisconsin and Holocene experienced multiple noncyclic transgressive and regressive events. Lake levels have been as much as 18 m (60 ft) higher and 60 m (200 ft), or more, lower than present, and changes have commonly occurred at rates several magnitudes greater than the most rapid eustatic sea level changes. In this chapter, we will show how hydrodynamic processes, cyclic and noncyclic lake-level changes, and the way in which sediments are supplied to the lake have interacted to shape the architecture of sedimentary deposits along the coast and in the deep basins. We will summarize the results of our own work, but we are also indebted to many researchers whose work is included in this narrative.

Abstract “A major obstacle to performing reliable predictions of plume evolution and to the design of effective detection, monitoring and aquifer remediation strategies is (understanding) the natural heterogeneity of geological materials.” The difficulties associated with quantifying both the “nature and degree (of heterogeneity)... are immense” (Sudicky and Huyakom, 1991, p. 240).

Abstract In most glaciated basins, the arrangement of aquifers and confining units is defined by facies distributions that result directly from a variety of regional and local-scale controls on sedimentation patterns. Glacial terrain models utilize abundant subsurface data, particularly downhole samples and geophysical logs, to document the composition and spatial variability of glacigenic sequences and attendant landscapes. When combined with water-level data and other hydraulic information, they provide a useful sedimentologic framework for interpreting hydrogeologic settings at both local and regional scales.

Abstract Aquifers and confining units in glaciated basins commonly exhibit a remarkable array of geometries, internal discontinuities and changes in physical and chemical properties at a variety of scales. These heterogeneities commonly control the distributions, both locally and regionally, of productive zones in an aquifer system as well as the locations of preferred pathways for contaminant migration. Glacial terrain models emphasize the nature of variation within individual sediment bodies as well as between entire depositional sequences; thus, they provide a useful sedimentologic framework for interpreting hydrostratigraphic complexity at both local and regional scales. The usefulness of these models is illustrated by the Huntertown aquifer system of northern Allen County, which consists of several distinct proglacial, supraglacial and glaciolacustrine hydrogeologic facies tracts whose distribution and internal variability reflect the interaction of the Saginaw Lobe and its meltwaters with contrasting morphologic regions of the depositional surface. Aquifer system properties, such as hydraulic conductivity and transmissivity, and the character of internal and regional system-bounding confining units vary systematically across the depositional basin in response to changes in sedimentaiy architecture. The development of a glacial terrain model for the depositional systems of northern Allen County provides the essential regional context necessary for interpreting, depicting and forecasting the nature and hydraulic significance of heterogeneities that characterize different parts of this aquifer system and its confining units.

Abstract Glacial aquifer systems are sediment bodies and their bounding surfaces. While the sedimentologic and geometric character of these systems can be predicted from general depositional models, the actual configuration of sediment bodies and contacts in a given aquifer system is determined by the sequence of glacial geologic events at that location. Geologic analysis integrates depositional models with local data to define the (1) bedrock surface containing or underlying the aquifer system; (2) geometry of the sediment units comprising the aquifer system; (3) contacts between the units; and (4) texture, sorting and stratification within the units. The bedrock surface may be a fluvial valley unmodified by glacial erosion, a basin created entirely by glacial erosion or a glacially modified fluvial valley. The geometry of the sediment units may be tabular, basin-filling, beaded, ridged or blanketing; and they may be bounded by fluvial erosional, glacial erosional, onlapped, gradational, interfingered or deformed contacts. Texture, sorting and stratification within the sediment bodies may be nearly homogeneous and isotropic or heterogeneous and anisotropic.

Abstract Sedimentary facies, reflecting original depositional environment, define the trend, dimensions, connectivity and internal heterogeneity of transmissive zones within clastic aquifer systems. Three heterogeneity styles—layer cake, jigsaw puzzle and labyrinth—reflect increasing degree of complexity. Style is directly determined by the depositional origin of the aquifer. Heterogeneity occurs over a broad range of scales. Megascopic heterogeneity is determined by the external dimensions, trends and degree of interconnection of permeable units. Macroscopic heterogeneity, which occurs at the depositional facies scale, includes (1) compartmentalization due to flow barriers between specific facies within larger permeable units, (2) vertical and lateral permeability gradients created by patterns of grain size and sorting, and (3) stratification and low-permeability mud baffles, which create anisotropy. Mesoscopic heterogeneity reflects lithofacies, sedimentary structure and lamina-scale variability. All three scales of heterogeneity structure can be efficiently described, quantified, interpolated and predicted within the context of a well understood depositional system framework.

Abstract Fluvial depositional systems constitute major aquifers in closed and semiclosed terrestrial basins, alluvial valleys and intracontinental sags containing axial river systems, and coastal plains containing fluvial, deltaic shore-zone and sometimes fan and fan-delta systems. Fluvial systems can be grouped into a spectrum defined by trunk-stream bed-load, mixed-load and suspended-load channel types. Each channel type produces a predictable range of external aquifer geometries and internal heterogeneities. Because of the strong contrasts in fluvial systems of permeable, transmissive channel fill facies and confining flood-basin facies, flow patterns are controlled by channel connectivity, which correlates to fluvial system type and overall sand percentage. Bed-load (generally braided) and mixed-load (generally meandering) fluvial systems typically deposit “jigsaw-puzzle” aquifers systems. Heterogeneity increases with decreasing scale of facies units and increasing shale (or mud) in the system. Small meandering rivers and stable mud-rich, low-sinuosity channel systems create “labyrinthine” aquifer systems. In mixed- and suspended-load channel fills, vertical and along-channel textural changes and lithofacies partitioning and the dipping shale baffles in the upper point bar commonly create separate permeability units within a single point-bar sand body. Low-sinuosity suspended-load channels typically consist of a lower active fill and an upper abandonment phase fill—creating two units of differing aquifer properties. Sand-body width-to-thickness ratios are greatest for braided (largely bed-load) channels and are least for low-sinuosity, stable (largely suspended-load) channels and delta distributaries. Facies dimensions control both lateral extent of permeable units in labyrinthine systems and continuity of the more permeable units within jigsaw-puzzle systems.

Abstract Stochastic models have been used extensively to represent uncertainty in the spatial distribution of aquifer properties and its impact on prediction of groundwater flow and transport behavior. Because natural porous media are often strongly heterogeneous and exhibit complex spatial structure, it is not possible for any model to completely characterize the spatial distribution of aquifer properties. Several models have been proposed, each of which relies on specific assumptions regarding the character of natural spatial structure. In this study we have performed a direct comparison of the performance of a number of stochastic simulation methods, using a geologically realistic synthetic dataset as the basis for such a comparison. The methods are evaluated in a Monte Carlo sense through comparison of the predicted distributions of a variety of measures of flow and transport behavior. The results indicate that classical (second-order) stochastic models can be expected to provide biased and nonconservative predictions of many practical measures of flow and transport behavior and to underestimate the uncertainty in those predictions. These empirical results are supported by theoretical arguments based on statistical entropy. Models that preserve the spatial continuity of geologic facies to a greater degree provide better predictions of flow and transport behavior.

Abstract Large-scale (< 1 m) variability in hydraulic conductivity usually is the main influence on field-scale groundwater flow patterns and dispersive transport. Incorporating realistic hydraulic conductivity heterogeneity into flow and transport models is paramount to accurate simulations, particularly for contaminant migration. Sediment lithologic descriptions and geophysical logs typically offer finer spatial resolution, and therefore more potential information about site-scale heterogeneity, than other site characterization data. In this study, a technique for generating a heterogeneous, three-dimensional hydraulic conductivity field from sediment lithologic descriptions is presented. The approach involves creating a three-dimensional, fine-scale representation of mud (silt + clay) percentage using a "stratified" interpolation algorithm. Mud percentage then is translated into horizontal and vertical conductivity using direct correlations derived from measured data and inverse groundwater flow modeling. Lastly, the fine-scale conductivity fields are averaged to create a coarser grid for use in groundwater flow and transport modeling. The approach is demonstrated using a finite-element groundwater flow model of a Savannah River Site solid radioactive and hazardous waste burial ground. Hydrostratigraphic units in the area consist of fluvial, deltaic and shallow marine sand, mud and calcareous sediments that exhibit abrupt facies changes over short distances. For this application, the technique improves estimates of large-scale flow patterns and dispersive transport. The conductivity fields mimic actual lithologic data, providing a more realistic picture of subsurface heterogeneity. Field-observed preferential pathways for contaminant migration are replicated in the simulations without the need to artificially create zones of high conductivity.

Abstract Predicting flow and contaminant transport in sedimentary aquifers requires three-dimensional, quantitative characterizations of facies distributions, including the proportions, mean lengths and spatial correlation of the facies. Sedimentary facies analysis focuses on these same characteristics and, when coupled with geostatistical methodologies, can provide the necessary quantification of facies distributions as well as ideas about the origin and typical geometry of facies, what we call quantitative facies models. Because they incorporate an understanding of depositional processes, quantitative facies models must be based upon facies distributions that are genetically related. In aquifer systems composed of facies having sharp permeability contrasts, and in which indicator geostatistics can be utilized, genetically related facies assemblages are the appropriate statistical populations. We illustrate these points by considering a portion of the Miami Valley aquifer system in southwestern Ohio, where three facies assemblages can be defined. These assemblages, which differ in mean facies proportions, mean facies thicknesses and lateral facies dimensions are attributed to different combinations of depositional processes associated with Pleistocene glaciation.

Abstract In many hydrogeological investigations of groundwater flow and contaminant transport, considerable uncertainty arises from unknown physical heterogeneity of the aquifer system materials. Indicator geostatistics may offer tools for stochastic simulation of heterogeneity, however, existing methods were not designed with hydrogeological problems in mind. Typically, hydrogeological data are vastly undersampled in lateral directions. Subsequently, geostatistical analyses must rely, in part, on geologic interpretation to yield geologically plausible results. Compared to traditional variogram-based approaches, the transition probability/Markov approach introduced herein more rigorously considers spatial cross-correlations (juxtapositional tendencies), yet is more conducive to integration of geologic interpretation. Three-dimensional (3-D) Markov chains are introduced as a conceptually simple yet theoretically powerful model of spatial variability, supported in theory and practice by numerous prior 1-D geologic applications to vertical sedimentary successions. The geostatistical conditional simulation algorithms of sequential indicator simulation and simulated quenching (zero-temperature annealing) are modified to include consideration for all spatial cross-correlations. “Nonstationarities” related to anisotropy directions, proportions of depositional units and commingling depositional systems also can be considered. Example applications are given for alluvial fan and fluvial depositional systems in California in portions of the Livermore Valley, Kings River alluvial fan and Salinas Valley.

Abstract Two synthetic examples and one field example demonstrate how improper definition of the spatial distribution of geohydrologic units in conceptual models of groundwater flow leads to erroneous parameter estimations, resulting in poor predictions of flow system behavior. Efficient and objective parameter estimation is obtained through inverse modeling techniques. The examples demonstrate the sensitivity of the estimated geohydrologic parameter values within the geohydrologic units to variations in spatial distribution of those units. In all three examples, geological information in the form of stratigraphic opinion or geophysical interpretation regarding unit distributions, as a well as geologic rules ranking hydraulic conductivity of units, improves conceptual model development.

Abstract The Pleistocene and Holocene alluvial sediments filling incised valleys beyond glacial margins share a number of characteristics, despite variations in magnitude and timing of events in their respective drainage basins. In a gross sense, the alluvial fills are composed of a two-part stratigraphy consisting of a lower, coarse-grained interval deposited in response to glaciation of the drainage basin and an upper fine-grained interval deposited by Holocene rivers. In detail, however, both lower and upper intervals commonly consist of multiple sequences deposited in response to variations in discharge and imposed sediment load. Sequences were deposited during periods of aggradation, and sequence boundaries formed during erosional episodes. The resultant architecture is characterized by inset stratigraphic relationships and abrupt lateral facies changes. Investigations in the Ohio, White, Whitewater, and Wabash Rivers in Indiana reveal that as much as 50 meters of alluvium fill bedrock valleys. The valleys normally are steep-sided with broad, flat floors incised by a narrow, deep trough (called the Deep Stage). These valleys were eroded in step-wise fashion during late Tertiary time and earliest Quaternary time. The basal alluvium in these valleys consists of consolidated gravel and sand deposited during pre-Wisconsin ice advances, probably by braided stream systems. These earliest outwash sediments are capped by a boulder-cobble lag or by channel and overbank deposits that accumulated during interstadials. Pre-Wisconsin sediments were eroded in response to increased flow during the initial stages of the Wisconsin glaciation, and valley floor aggradation occurred when outwash arrived somewhat later. Stacked sequences, that accumulated in response to multiple glacial events in their drainage basins, can be recognized in Wisconsin fluvial deposits in these valleys. Sequences, in turn, are composed of stacked depositional units that accumulated in response to the annual meltwater cycle and the passage of storms. Diurnal variations in the glacial meltwater regime are not reflected in these deposits because their large drainage basins served to buffer short-term events. Braided-stream deposits may be interbedded at valley margins with lacustrine deposits where rapid aggradation of valley floors dammed tributary valleys and formed small lakes. Other associated deposits include windblown silt and sand that was stripped from exposed channels during low-flow stages and deposited at valley margins. Rivers readjusted to new conditions of sediment and water discharge toward the end of Wisconsin time and the beginning of Holocene time by incising their valley floors and changing their channel patterns. The Holocene rivers are meandering, single-channel streams. Migration of channels across valley bottoms produces parasequences, consisting of sandy channel deposits and thick, fine-grained floodplain deposits, separated by erosional surfaces produced during rapid changes in channel position. Tributary streams draining large basins adjusted to these new conditions by eroding new channels across these alluviated plains and often rejoining trunk streams at different places than their precursor channels. Smaller tributary streams deposited fine-grained alluvial fans at valley margins. In places, these are large enough to modify migration patterns of trunk streams. The sedimentary response in the trunk valley to these readjustments resulted in complex architecture consisting of multiple facies within single-cycle fluvial deposits.

Nearly 40 years ago, Bretz inferred that glacial Lake Chicago stood at the 189-m (620-ft) or Calumet level twice, first before the Two Creeks low-water phase and then again after the Two Creeks interval. Although Bretz argued for a double Calumet stage on theoretical grounds, he clearly attributed formation of the Calumet shoreline to the post-Two Creeks stage only. Willman, on the other hand, believed that Calumet shoreline features were formed during both Calumet stages. Eschman and associates argued that the Calumet phase was pre-Twocreekan in age, and that Lake Chicago did not return to the Calumet level following the Two Creeks interval. On the basis of available radiocarbon age control, we attribute formation of Calumet shoreline features and deposition of associated sediments in the type Calumet area at the south end of the Lake Michigan basin to a post-Two Creeks lake phase. We have examined two areas of the Calumet shoreline in detail—the Rose Hill spit at Evanston, Illinois, and two sites on the Calumet beach at Liverpool, Indiana. Radiocarbon dates on nine wood samples from beneath and within the Rose Hill spit deposits range from 11,870 to 11,000 yr B.P. At the Liverpool East site, beach deposits contain driftwood dated at 12,400 and 11,740 yr B.P.; the older wood is possibly redeposited Glenwood material, and the younger is certainly no older than Two Creeks. Five additional dates, on wood and peat from an overlying thick organic layer and younger dune sand, range from 11,290 to 9,080 yr B.P., including a date of 9,920 yr B.P. on a tree trunk in growth position. At the Liverpool West site, the oldest date (11,815 yr B.P.) is Two Creeks, and two dates on material higher in the section fall within the range of dates from the Liverpool East site. Although the radiocarbon evidence does not preclude the possibility of a pre-Two Creeks Calumet phase, the general lack of pre-Two Creeks dates from the Calumet beach indicates either that the pre-Two Creeks Calumet phase (Calumet I) was so brief that no prominent shoreline features were formed or that these landforms were obscured (destroyed?) during the post-Two Creeks Calumet phase (Calumet II). In any event, type Calumet shoreline features were not formed until the Lake Michigan Lobe readvanced to the Two Rivers Moraine in post-Two Creeks time.

The types and spatial distribution of subsurface sedimentary deposits in the Calumet and Toleston Beaches of ancestral Lake Michigan were studied to better understand the evolution of the southeastern shore of Lake Michigan. Deposits of eight depositional environments were recognized: (1) dune, (2) foreshore, (3) upper shoreface, (4) lower shoreface, (5) offshore, (6) back-barrier lacustrine, (7) paludal, and (8) glacigenic. The Calumet Beach formed at the end of a rise in lake level following the Two Creeks phase, a time period of low lake level in the Lake Michigan basin, to the Calumet level. This trasgressive event was primarily erosional and produced a ravinement throughout the study area. Locally, however, relief on the underlying till of the Lake Border Moraine was instrumental in the preservation of nearshore sediments. Progradation of the Calumet shoreline produced a vertical stacking of shallow-water coastal sediments over deeper water deposits. Lakeward translation of the shoreline occurred for an unknown period of time, until the altitude of the lake dropped to the level of the Chippewa phase of ancestral Lake Michigan. Unlike the transgression from the Two Creeks level to the Calumet level, the post-Chippewa transgression to the Nipissing I level was dominantly depositional. This transgressive event is recorded in an ascending sequence of back-barrier lacustrine, dune, and foreshore deposits in the western part of the study area and by the onlap of the toe of the Calumet dune and beach complex by back-barrier lacustrine, palustrine, and dune sediments.