As the site analysis continued, Landsat data were added to update the land use classification for some themes or data elements. Using predetermined decision criteria, a new set of simplified maps (with all but one decision theme blanked out) indicate areas of acceptability or unacceptability. By combining these factors with proper weightings, a BestSite map was generated, showing the one location that has the highest or optimum suitability.
The PP&L data sets were somewhat dated, having been acquired over a period of years. One of the main contributions from Landsat was to provide an updated land-use map. From this Multispectral Scanner False-color composite, a "cookie-cut" section appears, which represents the service area for locating the site:
ERRSAC ran this June 1977 scene through a Supervised Classification, in which we selected training sites from other data bases and field observations. This map resulted
15-17: Compare the Landsat map with the PP&L map on the previous page. What are the main similarities; the main differences? What factor(s) are largely responsible for the differences? ANSWER
Drawing upon the PP&L data elements, and the updated classification, we then applied a site-selection model, developed particularly for this study, to the data base. We chose six of the 43 elements as the primary criteria for establishing suitability: (1) Landforms; (2) Groundwater Supply; (3) Soil porosity; (4) Ease of Excavation; (5) Foundation Stability; (6) Distance to Surface Water (limits: 15 cells maximum linear distance). We added other factors, including several socio-economic ones, as secondary determinants.
The next illustration uses single colors (against a yellow-gray background with the river in blue as a location reference) to mark areas derived from the model analysis that were acceptable or unacceptable under the set conditions.
Panel A highlights areas (red) of high acceptability under all conditions. In B, the yellow area denotes unfavorable groundwater. In C, the few blue dots point to areas subject to unsatisfactory porosity infill rates. The purplish-red area in D refer to an unacceptable area because of foundation instability. The green in the E panel associates with landforms, in which slope steepness rules out any use for a site. F contains large areas in light-blue that relate to difficult excavation.
We combined these elements, along with others, to produce a Best Site map, made by creating a binary mask (0s and 1s for reject or accept) for each data plane and then adding them to form this composite image:
Note that only one small area (the white square, with enlarged details in the inset) broadly meets all these conditions and the constraint of proximity to a large surface water supply as intake for steam production (also an avenue for discharge of waste water). This area is a lowlands (in red) between ridges, just south of the junction of the Juniata River with the Susquehanna that is high enough to avoid most floods. It lies between the frontal Blue Mountain (south) and Mahony Ridge (north). Although organizations use much of the surrounding area along Route 11-15, the immediate area selected for the site has only two small towns, Duncannon and Perdix, on either side and still has open lands amenable to development. A good supply of labor is nearby, including Harrisburg, whose northern suburbs are about 10 km (6 mi) away.
Admittedly, the above example is a rather simplistic exercise. But it does illustrate the manner in which GIS supports various modes of decision-making. It also demonstrates the role of timely space imagery in the process.
15-18 Critique this Best Site final product. Is it convincing? ANSWER
Meanwhile, if you are new to GIS, you now should know enough to begin practicing on your own. There is a web site, sponsored by the Research Program for Environmental Planning and Geographic Information Systems (REGIS), a group of geographers at the University of California at Berkeley, that has data sets (maps, aerial photos, and space imagery) of the San Francisco Bay Area, which you can display, combine, and use to output new products. This is the BAGIS program, done with a version of GRASS, called GRASSlinks, developed at UCB by Dr. Susan Huse and her colleagues. Access it here REGIS. Links at this site also guide you to other projects by this group.
GIS has now been used in tens of thousands of studies. But the Internet has a dearth of good ones that go into enough detail to appraise you of what really can be done. Still, a search (try entering "GIS + Case Studies") on the Web will lead you to some examples that at the least hint at the power of this approach. Perhaps the best way to learn to use GIS is to purchase (it's expensive) the IDRISI program
, which guides you through most of the essentials and illustrates these with examples. To entice you to check these out, here is (a somewhat degraded) reproduction of one of those images, in this case several image windows on the IDRISI screen that are input/output to a suitability map.
ESRI, designers and vendors of ArcInfo - the most widely used GIS software systems, have placed on the Internet a free training program that assists the user in learning how to manipulate maps and perform some of the basic functions of GIS. They do this partly because of altruistic educational purposes but also as a means of introducing their products. It can be downloaded (12.8 megabytes - takes some time) by clicking on ESRI ArcExplorer; this puts up all the products from ESRI and includes a link labelled ArcExplorer listed under the category of Free Viewers. Version 2 works for most operating systems; version 4 is tailored to Java Users. Before you decide whether to obtain ArcExplorer, and at the least do the basic Tutorial exercise as a learning project, bring up the Adobe Acrobat .pdf set of instructions. Once online, look at page 6 (of the 81 pages) which summarizes what can be done with ArcExplorer; from the list you can decide whether to download the full package.
The writer (NMS) has received several items from the ESRI Education officer that give some examples of their products and the types of analytical outcomes these can develop. One of these is chosen to feature on this page. Put together by Dean Oman and Robert Kellogg of the Natural Resources Conservation Service of the U.S. Dept. of Agriculture, Beltsville, MD, its goal is to identify Priority Croplands in the 48 contiguous United States that have the greatest potential for soil loss, soil quality degradation, and nutrient loss. Various models were devised. A wide range of input maps were created by ESRI's ArcGIS Desktop 9 processing system. A number of representative maps are shown below (these can be accessed at higher resolution on this NRI site); the label on top of each indicates its theme: 
Several "bottom line" output maps are generated using these and other data sets. The summary map below shows the Priority Croplands (top 15%) experiencing the greatest soil loss, nutrient loss, and soil degradation. These are in most need to short term attention:
15-19 In this priority map, which part of the USA appears to have the greatest problem (in terms of priority levels); which nutrient is experiencing the largest loss; what other maps were particularly helpful in making the assessment of highest potential for detrimental effects; looking at the map below this question, where is the critical problem area there?ANSWER
The areas needing attention can be narrowed down to the top 5%:
Most of the input maps in this croplands examples depend on data obtainable mainly from ground observations and measurements. But some of the other thematic maps do benefit from remote sensing data that are invaluable in that they are updated and current. In general, GIS utilizes mostly stand-alone ground truth but in many instances relies also on space and air imagery capable of classifying appropriate themes.
For a fuller understanding of the interactions between GIS and remote sensing, you are reminded of the book cited in the Overview, entitled "Remote Sensing for GIS Managers", by Stan Aranoff, and published by ESRI.
In the last two decades, GIS has burgeoned into a major international industry. It has become the tool of choice for many applications of map-based, spatial data to problem solving in the Information Age. Its emergence has largely paralleled that of remote sensing and the two aid each other symbiotically. In the GIS realm, remote sensing contributes a subset of input data, whose value is mainly in updating land-cover types. As a measure of the import now attached to GIS, it has grown into a staple of the geography curricula in our colleges and is probably more widely taught than remote sensing.
We urge you now to proceed to the next two pages, the first of which presents a third, and very detailed and intriguing, case study that was received from an outside contributor. The second case study was found on the Internet; its write-up here is the work of the present writer (NMS) using Net inputs.
Primary Author: Nicholas M. Short, Sr.