U.S. Geological Survey
National Center, MS 926A
12201 Sunrise Valley Drive
Reston, VA 20192
Telephone: (703) 648-6349
Fax: (703) 648-6953
e-mail: mmcrae@usgs.gov
Digital datasets facilitate many map-oriented activities such as updating and reprinting existing maps, rescaling data, recombining map units based on common attributes, and overlaying geologic data with other geographic information. Of course, these technological advances have altered neither our understanding of geologic information nor its role in decision-making. Since geologic maps have proven their ability to effectively communicate knowledge of geologic environment, database design practices have focused on translating the geologic map model into the digital arena so that individual paper map elements (i.e., lines, polygons, and symbols) become the geometric building blocks of their corresponding digital database. However, a digital database is not a map. Although applicable to the same problems, the publication media and methods of presenting, exploring, visualizing, and analyzing digital data are significantly different. These differences directly impact how the user perceives and applies the information. Consequently, a digital database whose geometry adheres strictly to the conventions of a paper map is less effective at communicating information than its analog counterpart. This paper attempts to characterize these differences and to suggest alternative models for database design.
This combination of individual, 2-dimensional graphic elements forms a single, cohesive product. In order to correctly apply geologic map information, one must understand that the geographic relationships of geologic units are not fixed, but rather change with depth below or height above the Earth's surface. The user must understand how to reconstruct the 3-dimensional framework from these components. Since this interpretation is largely visual, it is the author's responsibility to maximize this understanding by controlling the selection of graphic elements, their layout, and symbolization.
Current database development practices focus mainly on the map and legend components of the paper map. Typically, an existing paper map is scanned or digitized, separated into thematic layers, and attributed according to the map legend information. Due to the importance of positional accuracy, a great deal of effort is expended in 'quality control', i.e., ensuring that the source map's lines and polygons are accurately reproduced and attributed consistently. Although the cross-sections and other diagrams are often included as graphics files, they receive less attention. Consequently, the finished product accurately reproduces the map geometry and descriptive content, but with less emphasis on the interpretive information and geologic relationships.
This section outlines two conceptual issues that need to be addressed in order to improve geologic knowledge representation in digital databases: thematic separation of data layers and the geometric representation of geologic objects. (Note: For the purposes of this discussion, the terms 'feature' and 'object' have distinct meanings. An object generally refers to an entity that is identifiable by particular physical characteristics, relationships, and behaviors, while the term 'feature' generally refers to the geometric element used to represent that object.) Each issue is discussed separately, although in practice, they are interrelated and difficult to isolate. The context of this discussion is conceptual rather than practical; however, two recent publications (McRae, 1999; and Cannon, McRae, and Nicholson, 1999) provide some examples of how existing GIS tools and data structures can be implemented to address the issues presented here.
On a geologic map, the author controls the physical layout of individual components in order to facilitate the visual interpretation of the geologic relationships. Current database design practices require the user to reassemble individual components in some meaningful way. Recent policies adopted by the USGS have attempted to overcome this problem by recommending that a print quality graphic file of the geologic data be included with each dataset. This provides the database user with the opportunity to view the data as the author intended. Although this is a valuable visual reference, the issue of how to encode the author's interpretation within the database structure still needs to be addressed.
A negative consequence of this practice is that what the geologist considers "singular" can become "compound" due to either the limitations of its analog geometry or the digitization process. For example, consider the case of a fault that has been offset by another. The geologist views the crosscutting fault as a singular object and the offset fault as a compound object consisting of two line segments. However, some GIS software packages place nodes at all line intersections. Consequently, both faults will be divided into multiple line segments, effectively creating two compound objects. Without some mechanism to 'reassemble' the crosscutting fault's segments back into a single feature, the geologic interpretation is obscured. Similar problems occur with polygonal data. Paper map constraints force geologic units to appear mutually exclusive, so that their cartographic representation reflects only that portion of the unit not covered by another. On a geologic map, a volcanic unit that underlies a sedimentary unit may appear as multiple, disjointed polygons where the sedimentary unit has eroded to expose it. Common symbolization, annotation, and accompanying cross-sections help inform the map user that the unit is contiguous at depth. Current database production practices typically digitize and attribute each polygon individually. Again, this fragmentation obscures the geologic interpretation that the individual exposures are really part of a single, underlying unit.
In some cases, an object's cartographic representation may serve as the foundation for its digital geometry, if combined with the appropriate data structure. The behavior of the crosscutting fault, for example, can be modeled by using network geometry to aggregate the individual line segments into a single feature. However, many cartographic representations fail to reflect the real geographic extent of the objects being modeled. This is particularly true for geologic units. For example, the aggregation of the volcanic unit's individual polygons would still misrepresent the geologist's knowledge of its distribution.
On a geologic map, any knowledge of the distribution or understanding of how one geologic unit relates to another will be based on an individual's ability to interpret the 3-dimensional distribution from the 2-dimensional representation. A GIS can interpret the distribution of an object only through coordinate information and geometric properties. Hence, the 3-dimensional framework must be encoded in a way interpretable by GIS software. A key to accomplishing that is to ensure that the geometry of an object fully reflects the geologist's knowledge of its distribution. In many cases, that will involve a geometry not constrained by an object's cartographic representation.
Cannon, W.F., McRae, M.E., Nicholson, S.W., 1999, Geology and Mineral Deposits of the Keweenaw Peninsula, Michigan: U.S. Geological Survey Open-File Report 99-149, https://pubs.usgs.gov/openfile/of99-149/.
Johnson, B.R., Brodaric, B., Raines, G.L., Hastings, J.T., and Wahl, R., 1999, Digital Geologic Map Data Model: AASG/USGS Geologic Map Data Model Working Group Report, p. 13, http://geology.usgs.gov/dm/model/Model43a.pdf.
Levinsohn, A.G., 2000, Use Spatial Data to Model Geographic Knowledge: GEOWorld. v.13, no.4, p. 28.
McRae, M.E., 1999, Using Regions and Route Systems to Model Compound Objects in Arc/Info, in D.R. Soller, ed., Digital Mapping Techniques '99 -- Workshop Proceedings: U.S. Geological Survey Open-File Report 99-386, p.195-196, https://pubs.usgs.gov/openfile/of99-386/mcrae.html.