ABSTRACT The Mitchell Plateau of south-central Indiana is one of the iconic karst landscapes of the United States. The sinkhole-dimpled forests, fields, and farms; the extensive cave systems; and the deep windows into the groundwater system have fostered curiosity, exploration, and publication since the mid-1800s. This paper is designed to complement a field excursion to the classic features of this landscape. Included are literature reviews focused on three karst basins of the Mitchell Plateau: Mill Creek–Mosquito Creek, Bluespring Caverns, and Lost River. Geomorphic, hydrologic, and geochemical data are synthesized in the modern context of our understanding of epigenetic karst. Revealed are three styles of karst basin: (1) small, shallow karst aquifers strongly controlled by meteoric recharge and epikarst percolation; (2) intermediate-size karst aquifers with significant base flow and surface-water–groundwater interaction; and (3) regional aquifer systems with outcrop belt recharge, downdip transport into confinement with long water-rock interaction times, and artesian flow or entrainment of mineralized waters through fractures into springs or surface waters. Quaternary glaciation has greatly influenced the vertical position of base level through river incision and sediment aggradation; conduit development is controlled by proximity to the major rivers and the stratigraphic position of conduits.

Two trends are evident in the recent developments in engineering hydrology, one toward integrating surface-water and ground-water concepts, and the other toward integrating stochastic and deterministic techniques. This paper provides a review of these trends with the specific aim of clarifying the role of ground water in the current methodology. Among the interrelationships discussed are (1) the linear-reservoir representation of the instantaneous unit hydrograph and the concepts of ground-water storage; (2) the existence of autocorrelated stream-flow sequences and their dependence on baseflow recessions; (3) the conceptual prediction of flood exceedance probabilities and the mechanisms of streamflow generation; and (4) the nature of time-space tradeoffs in assessing the worth of data for rainfall-runoff models. The process of streamflow generation is controlled in large part by the near-surface saturated-unsaturated hydrogeological regimes that exist on the hillslopes within a watershed. Much remains to be done before the lumped calibration parameters that arise in time-series analysis, frequency analysis, and unit-hydrograph analysis of streamflows can be interpreted in terms of the spatially distributed, physically based parameters that control the runoff-generating processes on hillslopes.