Russell A. Ambroziak, Ph. D.
U. S. Geological Survey
Office of Energy and Marine Geology
National Center, STOP 915, Reston, VA 22092
Tel: 703-648-6168, Fax: 703-648-5464
(Published on 1993 USGS Open File Report, CD: 93-231)
GIV allows geologists, other scientists, teachers, students and the general public to view, combine and analyze geologic data on an inexpensive PC with little or no training. GIV uses simple straight forward data formats and color theory to graph and display any type of geologic data. Map data tends to be the most pervasive for geologic use and most useful to non geologists so it gets the most attention. Maps, however, are only part of GIV and should not be considered to be what GIV is. GIV is unique because it is unconcerned with data format, uses color as a coordinate system, works well on an inexpensive PC, does just about anything, is cheap to build and run, and anyone can learn to do it.
GIV is based on the fact that writing your own programs to answer simple questions is not only not bad, but it can be faster than using a package, will allow you to do any analysis you understand, helps you understand what you do, and allows you to analyze a surprising amount of data. Even if they never plan to program, scientists should know how to program in order to know when to program, when to have someone else program for them, when to buy a program, and how to evaluate programs they buy.
GIV now consists of a small set of USGS programs that can quickly turn a paper map into a computer displayable map, display and manipulate a variety of image types, combine maps with images and text in a variety of ways, and integrate just about any type of data display. GIV is easy to use, easy to learn to use, completely owned by USGS, and will run on any PC that has a color display. A complete set of hardware to run the entire suite of software costs less that $2000 and total training time to run the total program set is less than a week. GIV is scale and projection independent and is moving towards platform independence.
GIV is not a system, but rather, it is an attempt to give control of science back to the scientist. GIV is in part a set of functions that allow a scientist to control data display and analysis. More importantly, GIV is a philosophy that it possible to start at the bottom with very simple tools and quickly build fairly complex data processing machines to work with any type of data. Moreover, it is possible to learn how to do this even if computer programming is not a strong skill possessed by the scientist.
Graphing and displaying data is one of the primary ways in which scientists understand the masses of data with which they work. Before data can be used they must be understood, and to do this a mental model of the data must be constructed. Many times this is achieved by simplifying, sorting and graphing the data in various ways. The current way of analyzing data involves getting the data out of the system where it resides, finding and procuring software (and sometimes hardware) to read and process the data, learning to use the software, then finally processing the data. Each step of the process can take weeks, months or even years and, if the choice of software is incorrect, results will be disappointing.
Scientists who have a PC on their desk and a grasp of just about any computer language can build a simple program set to answer the questions that need to be answered in a few days to a few weeks. If the programs that did the work are thrown away and never used again there can still be a large savings in time and money over finding, procuring and learning to use commercial software. Also, if the scientist writes the code to solve the problem, any test can be tried on the data rather than only those tests contained in the software.
This is not to say that one should always write one's own programs -- this would be just a limiting as the current "never write your own programs" philosophy. One should become familiar enough with programing to know how simple or complex a program must be to solve a problem. Only then can a knowledgeable decision be made as to how to proceed.
After FORTRAN the thought was that what was needed was higher level programing languages. GIS (geographic information system) packages and systems are a type of high level programing language. They allow the user to do large tasks and to assemble the results into complex geographic outputs which can be combined in various ways to answer questions. The problem with these systems is that they are complex to operate, require long learning curves, and create data format problems when more than one system is used. The most successful computer language of late is C, which is much lower level language than FORTRAN. It is possible that for many applications what is needed is lower level not higher level languages. C is a good box of crayons for drawing graphs on a computer. GIS is not very competitive with a C compiler in the hands someone who wants to use it.
We make CD-ROMs of USGS data and on each disc we put software to display the data contained on the disc. The software has no other purpose than to display specific data and to display it quickly and easily on any type of PC with a color monitor. Each program was written from scratch with no more than a week or two spent developing any program. After the software is written for a disc the aut hor reviews it and decides what else it should do,usually by saying "Gee, wouldn't it be nice if ..." and the '...' are added to the program as new features. As the number of specialized programs grew the functions they performed began to duplicate each other and libraries of functions became grouped into a few sets. We now can create programs using these libraries in a few hours and we can modify any part of a libraryto do any additional task we wish. The current state of GIV, which is placed on each CD-ROM as they are produced, is due to this process being used over and over.
The important point is not that we have the libraries, but that one person built them in two years while turning out programs that worked. These programs are published as CD-ROM discs. Almost anyone could duplicate this effort if they tried, but they are prevented from doing it by cliches such as 'you're reinventing the wheel' or 'GIS is a complex subject that is best left to the experts'. Neither of these cliches is correct. The invention of the wheel did not put wheelwrights out of business, rather it created their craft. Geographic information is relatively easy to use, but GIS's tend to complicate the process. Because GIS's are written to do everything for everybody, they can be difficult to use and require a long time to learn to use.
GIV, if correctly used, is unconcerned with data format because a simple program can usually be written in a matter of minutes to convert almost any data to an ASCII format that GIV can use. GIV uses ASCII flat files as one form of data input and output for all applications. ASCII flat files are also used to control the programs to minimize the need for software changes. The data and control files can be created, checked and edited with any text editor. All earth locations are in decimal latitude and longitude. Each file contains a small group of like information and there are many files with names that describe their contents. The entire program set is designed for the PC user who does not have a lot of time to learn to use complicated software but wants to do some mapping to solve a problem.
One final point about GIV that distinguishes it from GIS is the use of color. A color image is a 5-dimensional graph that is useful for displaying all types of information. The five dimensions are 2 locational (x and y) and 3 color (red, green and blue or intensity, hue and saturation). If you learn a little color theory it is quite easy to use color to graph data, and the current programs that make up GIV use color theory to a great extent. You do not need to understand color theory to use these programs but you will need to learn something about it if you wish to create your own. If you want to learn color theory contact the Rochester Institute of Technology where you will find the most knowledgeable color experts in the country.
Before the computer science was learned with the slide rule and graph paper. The reason for the graph paper was to simplify and visualize complex concepts. There were graphs in the books and on the chalk boards, but drawing your own graphs helped you to understand what is going on. It was not the graph itself but the construction of it that gave the insight. Now that we have replaced the slide rule and graph paper with the computer there seems to be an all pervasive philosophy of let others do your graphing for you and just look at the results. Nowhere is this philosophy more prevalent than in geographic information.
Getting back to the simple truths about learning and understanding science, we scientists need to draw graphs. What we had to work with in the past was basically a 2-dimensional graph of points on an x,y plane. Several lines could be drawn on the same graph but they were all 2-dimensional. Attempts to make 3-dimensional graphs where limited and consisted mainly of contouring. With color monitors on computers we now have three more dimensions to work with namely red, green and blue. If we use these dimensions correctly we can draw and analyze 5-dimensional graphs.
The thought that image analysis and geographic information analysis are complex subjects better left to the experts is not universally true. Drawing a map is as simple as connecting dots much as we did when we where 5 years old in the 'follow-the-dots' book, but we use a function called 'plot_line' in a computer rather than a crayon. OEMG has developed and implemented, with some success, a simple to use and understand program set of data visualization tools which is a bottom-up rather than a top-down approach. We started out with the expressed purpose of writing specific software to display a single set of data and ended up building a small program set that was data independent.
When we started writing software for CD-ROM publication we specifically said that we would not do data base management nor would we do any data analysis. As it worked out the process of displaying data does all of the management functions required to present a clean, accurate, readily accessible data bases. Moreover, it is not the computer that does the analysis, it is the scientist; by displaying these data in various ways to the scientist, we provided the tools necessary to do the analysis. The group of programs that resulted do a very limited number of operations, but they are the operations our scientists need and they work with the data we have.
You could take these programs and learn to use them to display data. This is not, however, the intent of this exercise. The intent of the exercise is to show the average scientists that mapping and image processing are not as difficult as they might think and that there are times when it is both quicker and cheaper to do the job yourself than to find a system to do it for you. Image processing and graphics consists only of placing any color dot any where on the screen and reading the screen to find out what color dot is there. Mapping is using graphics to connect latitude and longitude dots to make lines, and to fill in areas with colors and patterns to make maps. You can use our functions to do this if you know C or you can write your own. Either way will free you from the bonds of complex software and data format incompatibilities.
There are several shocks when we begin a project with a visiting scientist who has not worked with us before. Data format is totally irrelevant, scale and projection have little or no meaning, just about anything can be done, and most importantly, the computer screen is not a piece of paper. We write our own code to do simple tasks and make things appear on the screen. The hard part is figuring out what you want on the screen -- getting it there is much easier. We frequently say that we ar long on answers but short on questions.
Because we are writing our own code we will transform whatever format we get to one we need in a matter of hours at most. If we need a new format we will make it while keeping to the primary rules of GIV -- keep in simple and keep it small. Flat ASCII files of one thing or closely related things with names that tell what's in them are best. We can convert these to binary files which can be quickly processed by our programs and convert the binary back to ASCII as we need to. People read ASCII, and computers read binary. You need both and they should be easily transferred from one state to the other.
A map scale and a map projection only have meaning for a paper map. If all data are in decimal latitude and longitude there is no reason to limit the scale or projection. All GIV programs handle a number of projections and users can switch between them as they see fit. Scale is important to some extent to tell the user the relative accuracy of the input data, which usually came from a paper map. However, the input scale of the data is usually obvious to the user as one zooms in and the lines become jagged with widely spaced points connected by straight lines.
Paper maps are static while computer programs are dynamic. In an atlas each page is fixed in size and content so maps tend to be divided into sections. In a computer you can zoom into any region and quickly create a screen sized map. The borders of sub maps become irrelevant to the user looking for a feature on the larger map. Also overlays such a roads, streams, place names, etc. can be added as needed in any combination the user wishes. Colors can be changed, flashed or areas measured at will. Any map on the screen can be output to some standard file format for printing at any scale desired. Scales and projections can be changed as desired and other data can be added from data files, paper maps or drawn right on the screen. Icons can be placed on the maps to mark objects and the icons can be used to locate text files, graphs or images. By choosing an icon the user can read information about it, see pictures of the location, or view graphs of data. All of this is quite easy to do with GIV, usually taking less than a day to make any application work.
Because GIV is not a system there is currently little documentation and it changes quite often. We are keeping it backwards compatible, but some programs become obsolete as other programs take over their functions. We do not guarantee that it works perfectly since we only use it for limited operations. We do, however, constantly fix things which we find to be broken and the latest version cost just as much as the first -- its free. We place every piece of software we use to make the disc on the disc in all forms needed to repeat the production process. Source code, object code and executable files are included as are all support and data files.
GIV is designed for the amateur who wants to do something quickly with a gentle learning curve yet not be limited by software if complex analysis is needed. What the average PC programer needs are simple to understand data in small clearly identified files. Large complex data sets which contain interrelated data of many types are difficult to work with and frequently impossible to decode. A scientist should not be forced to study cryptography in order to do their job. Very often I describe my position as the OEMG cryptographer because much of what I do is to get simple data out of complex formats that are frequently undocumented, or incorrectly documented.
Much of what we do is concerned with mapping. Mapping is not necessarily the primary science involved, but the data need to be displayed, at least in part, on a map of some sort. To create a map that contains all of the information necessary to answer any of the questions a map can answer you need only a set of vectors (lines with attributes) and, if the vectors make polygons, you need to have at least one point inside of each polygon with the polygon's attribute. Other files which mark places, things, etc. can be added after the base map is done. The GIV package allows you to do all of this quickly and easily on a simple PC.
The GIV imaging capability now includes conversion of 24-bit color images to 8-bit images which look like 24-bit images, mapping vectors and point data over images, creation of arbitrary image planes into color images, eigen vector rotation of up to 10 image planes, 3-dimensional colorsurface displays, and 5-dimensional color vector data displays.
Other GIV functions include time series display and analysis and seismic segment display. The time series display application is of current meter data. The programs allow graphing of temperature and current vectors as well as spectral analysis of velocity. Maps are used to locate available stations and all other parameters can be used to limit the data sets you wish to work with.
The seismic display can show complete segments of any size on the screen and allows zooming into the segment to show detail at any level needed. The traces are linked to surface position by maps on the screen and trace numbers which can be located in latitude and longitude position files. All that is needed to input data to the programs is a seg-y data file and a paper (or computer generated vector) map of the shot points.