Introduction |
Top Previous Next |
|
Welcome to Walls!
This program is freeware designed by cave surveyors intent on using it for their own projects. The need for specific and sometimes unusual features has guided its development over the years. As the original programmer, my first goal was to make it possible for surveyors to handle the basic data management tasks for several long-running projects in Texas, Montana, and Mexico, each encompassing hundreds of linked caves and karst features. Walls is now being used to manage survey data for some of the largest cave systems in this hemisphere, including both the longest and deepest.
Getting Started Since the program uses standard text as its native data format, you can't immediately begin entering survey data without spending a few minutes reviewing the script-like language used to specify measurement units, column order, and so forth. An interactive "getting started" tutorial would be nice, but this is just one of many desirable features the program could have. For now I suggest opening the small example project, Tutorial.WPJ, that Walls Setup installed. You'll see a tree diagram with a branch icon resembling a page. Double-click the page icon to examine a commented data file in the built-in editor. You may be able to modify it and start entering your data right away. The text editor in Walls is functionally similar to the basic editor that comes with Windows (Notepad). When the edit window opens, press F1 to see a description of the editor's features.
Be sure to probe the menus and dialogs as you run the program. Also be willing to reference the online help, which is quite extensive. If your habit is to explore without reading the documentation, and if you are sufficiently familiar with the Windows interface, you may be able to get by while missing some tricks. (Whenever an unfamiliar dialog window appears, press F1 for instructions.) On the other hand, if you're considering extensive use of Walls you should browse through the entire help file at least once to get a feel for what's available.
Design Goals The program was named Walls in anticipation of a feature that was implemented fairly recently in its history: the incorporation of complete map documents in the project database. Our mapping software should support the creation of detailed cave maps that automatically track an ever-evolving survey database -- something that commercial CAD and illustration programs can't manage at all. The unique problem we face is that estimated passage locations change as successive surveying trips are made into the cave. New traverses will be averaged into the data set, possibly exposing blunders in old data. Therefore, it would be nice if we could quickly produce up-to-date maps, both for publication and field use, that are more than just computer-drawn line plots or crude passage representations. Ideally, the maps will show realistic passage outlines and complete decoration. I believe this goal has largely been achieved due to the availability of Scalable Vector Graphics (SVG), a standard format that allows Walls to share the task of creating dynamic cave maps with a commercial drawing program.
Other features that go beyond basic data reduction relate to project management and data screening. Having once used certain statistical methods in Ellipse, a mainframe-based program, I've been wanting ever since to move them to a personal computer with a graphical interface. Also, it appears that blunder detection in networks -- surely an interesting and non-trivial problem -- has been given lower priority than it deserves by other program developers.
Cave surveyors are resigned to the fact that certain failures in technique are inevitable. The main categories of failure are bad instrument readings, mistakes in data recording, and various kinds of systematic error. My own experience has been that such problems completely dominate, or swamp out, all other influences on map accuracy, including the inherent imprecision of the instruments used. Almost irrelevant is the particular weighting scheme we choose to finally average the data that survives the screening process. By this I mean that even after we've done our best to filter out spurious data, the benefit of choosing one "loop closing" method over another is probably insignificant. (I admit I would like it to be otherwise.) Loops or no loops, the accuracy achievable with our instruments would be good enough if we could just avoid the mistakes.
Final map accuracy isn't the only consideration. There's also just the frustration and delay caused by mistakes that will eventually be noticed, like misnamed stations and reversed shots. Even here, unadjusted plots are rarely helpful when the cave geometry is complex. This problem continues to plague large mapping projects where just one person, usually, is trying to manage data collected by many different teams. Consequently, much of Walls has to do with network error analysis as opposed to simply network adjustment, the idea being that bad measurements can often be singled out immediately when loops are present.
While there's no need to have even the faintest appreciation of statistics to use the program as a tool, the basic idea can be stated very simply: By approximating the behavior of measurement data with a mathematical model, we can compute numbers that measure consistency in a data set. If by throwing out some measurement the consistency improves by more than we might reasonably expect given the model's assumptions, then we consider the measurement suspect. Walls uses this method to rate all traverses in a loop system, allowing us to confine attention to a relatively small part of the data. The program presents graphical views of an "outlier" traverse with the surrounding network undistorted by its presence. It also suggests specific measurement corrections that would bring the traverse into line.
About the Documentation Walls with its error checking is an approach to data reduction that's probably new to most cave surveyors -- perhaps drastically so -- and some may want a better explanation of the underlying theory even though it's not required. If you're in that category then I can recommend just about any introductory statistics text. The help file is definitely not a tutorial on linear models and least-squares, but it should be enough to get you started with the data screening tools. (The article cited in History and Acknowledgments describes in detail both the theory and the numerical methods used in Walls. Also, see Statistical Formulas for a terse summary of what the program actually computes.)
The help file also assumes that you're already familiar with the Windows interface -- moving and resizing windows, navigating menus and dialogs, and so forth. My hope is that you're at least willing to try the most popular method of software checkout: Open a sample file (a WPJ file in this case) and venture forth, mouse clicking everything in range.
You can visit the Texas Speleological Survey Web site to download the latest version of Walls along with some example project data sets. Try the program out and let me know what you would like to see (or not see) in the next build.
Thanks for your interest!
David McKenzie
|