The U.S. Geological Survey is conducting a study of the Cambrian and Ordovician aquifer system of the northern Midwest as part of a national series of Regional Aquifer-Systems Analysis (RASA). An integral part of this study will be a simulation of the ground-water flow regime using the Geological Survey's three-dimensional finite-difference model. The first step in the modeling effort is the design and development of a systematic set of processes to facilitate the collection, evaluation, manipulation, and use of large quantities of information. A computerized data-base system to accomplish these goals has been completed for the northern Midwest RASA.

The input to the data-base system consists of either point values or contour maps of the hydrogeologic data required by the model. Digital samples of contoured surfaces are obtained by machine digitization. Uniformly spaced model-node values are then computed from the discrete data by two-dimensional interpolation. The interpolator uses the regional (long wavelength) characteristics of the data to compute a surface with minimum total curvature. Local (short wavelength) data are merged with the regional surface to produce the model-node values at the desired spacing. Nonuniformly spaced nodal values may be obtained by fitting two-dimensional polynomials (bicubic splines) to surfaces formed by uniformly spaced points and solving for the surface values at the node locations. The uniformly or nonuniformly spaced node values constitute the output of the data-base system and in turn are the input arrays of data values for the ground-water flow model.

Management of the data base is facilitated by forming a data file for each model input parameter and for other hydrogeologic data used in the study. The data files are subdivided into elements consisting of individual model-layer arrays. In this form the data base can be readily accessed and edited. In addition to the interpolation programs, software components of the system include programs: (1) to transform geographic coordinates on a Lambert conformal conic projection to cartesian coordinates and vice versa, (2) to machine contour gridded data, (3) to expedite manipulation and editing of data, and (4) to perform various interarray computations.

The data-base system has the following attributes: (1) manual handling of data is minimized and machine handling of data is maximized, (2) given unequally spaced point data over the extent of a study area, model input arrays for any reasonable uniform node spacing can be rapidly computed, (3) accuracy of computed node values are generally compatible with accuracy and spatial distribution of point data, (4) a relatively large class of nonuniformly spaced node configurations can be computed, (5) data within data-base files can be easily accessed and readily edited, and (6) the occurrence of data processing errors at various stages of data-base generation is monitored by machine contouring the computed grids.

Functioning of the data-base system is illustrated by the sequence of steps required to produce a model-input array of transmissivity, given raw geologic and hydraulic conductivity data.