The Iowa Geological Survey Bureau (GSB) has been involved in a three year STATEMAP project in Linn County, Iowa to map the surficial geology of five 1:24,000 quadrangles and map county-wide bedrock geology and surficial geology at 1:100,000 scale ("STATEMAP" is a component of the USGS National Cooperative Geologic Mapping Program). The main goal of this project was to develop detailed geologic information for use by local officials, private businesses and the public in Linn County to aid them in natural resource decision making. While providing geological information for such purposes has always been a significant part of the mission of GSB, the development and construction of detailed 1:24,000 scale maps has not been a significant part of meeting it. Thus a secondary goal for this project was to develop expertise in field mapping techniques and construction of geologic databases. It was felt early on that developing traditional geologic maps alone would not meet the first goal of providing useful information to decision makers, such as county planners and engineers, solid waste authorities and county health officials, due to the lack of geological expertise found at their local level. More useful would be information in the form of interpreted "derivative maps", or geological data and other information presented in a thematic form, such as maps of aggregate resources or groundwater vulnerability. At the same time it was felt that providing the data as geographical information system (GIS) coverages could also enhance their usability in fledgling county and city GIS programs. At the start of the project, all of these ideas were major unknowns, and while subsequently have mostly proved to work, the acid test of really being used by those in the county still remains to be seen.
The Linn County STATEMAP project became an experiment to develop "paperless" geologic maps, wherein most if not all of the construction of the geologic map takes place on the computer. This does not mean that a computer "black-box" does all the geologic mapping nor does it mean that some non-geologist computer operator interprets geological contacts. Geologic interpretations were all done by the mapping geologists using a GIS and detailed digital coverages as a base for mapping. Through the use of GIS techniques, their ability to bring many different data layers together for viewing, interpreting and mapping was greatly enhanced. The mappers still used traditional data sources such as soil surveys, topographic maps, and aerial photos, but in non-traditional digital format. Field notes and drill samples were collected and recorded on paper field notebooks and had to be entered into databases and spreadsheets. Some structural surfaces and isopach maps were also done on paper first, then digitized. The biggest experiment was in mapping surficial geologic contacts, which were composed on-screen rather than on a paper 1:24,000 scale topographic map and later digitized on a tablet. One advantage to on-screen mapping was the ability to view background data and draw contacts at a larger scale (usually 1:12,000), rather than the 1:24,000 target scale. This makes locating outcrops easier and drawing smoother looking lines. The main advantage to this procedure was that the geologist making the interpretations also did the digitizing, which in the past would normally have been done by someone else, usually a graduate student. This removed the potential for the digitizing person to misinterpret the location of the line work and made the final GIS coverage reflect the mapper's original interpretation much more closely. A side benefit was getting some non-computer-literate geologists directly involved in the creation of their spatial database, getting them to see other uses for the technology and getting away from having GIS staff do the "computer mapping" part of the project. The steps used to create the main geologic coverages are outlined below. Digital GIS coverages are underlined.
Steps taken to create 100k bedrock topography coverage:
Steps taken to create 100k bedrock geology coverage:
Steps taken to create 1:24,000 surficial geology coverage:
The main disadvantage to these procedures is the amount of data (soils, orthophotos, etc.) that had to be assembled before the on-screen digitizing could begin. We were very fortunate that Linn County was a test project area for digital compilation of 1:24,000 line features for revising topographic maps, and many of the necessary layers were easily obtained. Also, Iowa has a statewide project to digitize all the existing soil surveys in the state, which is nearly complete. Digital raster graphics (DRGs) or scanned 1:24,000 topographic maps are coming into wide availability as another useful base map layer. On-screen digitizing was done with the Arcview 2 product from ESRI on an IBM-compatible desktop computer with a Pentium processor. Grid processing was done with the Arc/Info 7.0.1 GIS program on a UNIX workstation. It should be noted that both capabilities now exist in the Arcview 3 product with the Spatial Analyst extension for use on desktop PCs.
The three basic geologic coverages were then used to create various derivative maps. The bedrock geology, depth to bedrock and surficial geology coverages were intersected at the 1:24,000 quad level to create one combined coverage with attributes and polygons from all three. One of the nice features of this approach is the ability to turn on and off various combinations of bedrock units, surficial units and depths in order to try out different scenarios and models.
One mapping issue that should be addressed at some time is the determination of the accuracy of the mapping. I believe the need for this is obvious, as in letting map users know what kind of confidence we geologists place in the interpretation. As to how to do this, I don't have any good ideas, only some old tests used in accuracy assessments of land cover derived from remote sensing data. This usually involves randomly selecting a number of locations from each mapping unit and physically visiting each site and determining whether it is correctly mapped. The documentation for the coverage or map might state something like
"8 out of 10 test points for this unit tested correctly." There are many questions as to the practicality of this, and how test points would make a statistically valid sample. Assessing bedrock elevations and geology are even more problematic. Typically, a plot of the well data points is used to give a qualitative indication of the amount of information available to the mapper, but does nothing to test the validity of the geologic interpretation. This does not even begin to address the question of validity of derived products, which may be based on the intersection of several geologic coverages of unknown quality.
Throughout this article I have used the terms "geologic map" and "derivative map" to describe something which is really more specifically geologic databases, made up of one or more GIS files or coverages. Rooted in our past experiences and training, it is more comfortable to speak in terms of maps than digital geologic data. It will take some effort to develop concepts that more fully describe our geologic models in digital form. Maps, while familiar, are objects limited to a particular time, scale, and concept of reality. Digital geologic models (in our case in the form of GIS coverages) are somewhat more flexible because they can be modified and updated more readily (if one has the time), can be viewed at differing scales, and can be combined with other digital coverages to create totally new models and applications. Maps are still a very useful and necessary means of conveying information, but our focus in the "mapping" field must be turned to creating and developing the geologic databases necessary for solving society's natural resource problems (a main justification for the existence of the USGS National Cooperative Geologic Mapping Program and the National Geologic Map Database).
This gets into the discussion of developing standards for geologic data models, not standards for producing maps. Again, standards for the cartographic representation of geologic data are important and necessary, but the really critical universal element is the underlying data, its structure and content. Our experience has been that making geologic coverages compatible between different quads in the same project is sometimes difficult (lack of discipline and attention to detail), and between projects in the same state even more difficult (different mappers with different backgrounds), and between different state surveys under current circumstances may be next to impossible (state-line "faults"). Imagine trying to do some kind of a regional industrial site suitability assessment across a state border when one state's geologic model includes certain geologic units and the other state used a different set of attributes for slightly different mapping units, mapped at a different scale. Each mapping project or mapping geologist seems to have their own geologic model which usually isn't 100% compatible with other projects. Perhaps what is needed is some minimal standards for simple geologic data models that could be used as a starting point for new mapping projects. Developing such data models could be the focus of an EDMAP (educational equivalent of STATEMAP) project or some joint STATEMAP projects between adjacent states (Quad Cities, Iowa and Illinois for example). Elements that need to be considered in developing standard geologic data models include what geologic attributes are to be mapped (time, lithology, biostratigraphy, hydrogeological properties, engineering properties, etc.), type of mapping (bedrock, surficial, stack unit, soils, geomorphology, geophysical, etc.), resolution or minimum mapable features (minimum size of an object, which is somewhat related to eventual display scale; mainly a function of an attribute's variability) and how to represent three dimensional data (as a series of 2-D layers in polygon form, a series of 2-D grids, a 3-D grid of "voxels," or as 3-D vector type objects). The surficial geologic databases constructed for Linn County are 2-D vector GIS coverages mapping mostly time units with a few other material attributes thrown in, to a maximum depth of 5 meters. Even then they don't really do a complete job of portraying the variability within that 5 meters. Availability of data is a big controlling factor in what gets mapped, at what resolution, to what depth and areal extent.
Finally, a very important consideration (perhaps the most?) is the customer (probably a non-earth scientist) who will try to use this geologic information. Will a geologic map of chronostratigraphic units meet their needs for land use planning, locating sources of aggregates and groundwater, seismic hazard mitigation? Can anyone in their office read and interpret a geologic map? What system (CAD package, vector GIS, or none) do they have for using our data with theirs? I think this is a real challenge for geologists to deal with, particularly in our state survey. We need to figure out what information local users need, in what form, and how we assemble the resources to meet those needs. We're not there yet.