Using Modern GIS Techniques to Interpret Old Data

The Case of the Map of Kahola Lake

by Jeremy Aber for ES551 - Computer Mapping

Modern computer based GIS techniques have proved themselves many times over to be valuable tools for many different trades. They allow for fast and easy manipulation and creation of graphical map data that can be used in a variety of ways. But the data that go into a GIS must come from somewhere, and as is the case with many GIS's, that data is adapted from or based on older paper-based data. How can something like this be accomplished? This project shows one way that older map data can be adapted within the context of a computer based GIS, and it also shows some of the benefits of this change.

The Genesis of the Project

This project started as a map. Well, more of an engineering diagram really, but at least it was a starting point. The map in question was a hand-drawn diagram of Kahola Lake, the individual plots of land surrounding the lake, and the outer boundaries of the lake property. The map was enormous, and for most practical purposes, had fairly limited cartographic uses. The sheer size prevented anyone from actually being able to use the map effectively, and there were many other problems as well. First and foremost, the map was drawn using magnetic bearings, but there was no date anywhere on the map to indicate how the magnetic bearings should be adjusted for accuracy in modern times. The map, while being relatively clean to the eye, was nowhere near an accurate projection of what the lake is actually shaped like. Both the size and shape of the lake are incorrect, although the geometry is relatively close. Also, it was clear from aerial photos that the boundaries shown on the map were not exactly correct anymore (a creek feeding the lake had even shifted substantially from where the map indicated it was). And this was the best representation of the lake that had been available to Kahola for many years.

So, would it possible to salvage this map using modern GIS technologies? Obviously, it would be an option to redraw the map by hand, making changes and updating the data, but that would be pointlessly difficult and time consuming. So then how could this map be made usable again? Through modern computer GIS and GPS technologies of course!

Collecting and Manipulating Data

Data collection took several different steps. First, a copy of the map was needed. After much searching, a GIS Technician at the County Appraiser's office, Becky Samuelson, offered a solution. As the map was so large, using a standard scanner to copy the map was a possibility, but one that would require hours of scanning work. Becky offered to take the map to the City of Emporia's Planning and Zoning department and scan it using its equipment, as neither the university, nor my place of employment at the county had scanning equipment big enough to copy the map quickly. Becky returned a few days later with a full size copy of the map, as well as several smaller copies and, most importantly, a .TIFF file of the scanned map.

Now that I had copies of the original map, the second step to making the map usable again was to collect GPS data that would be used to resample the map. I collected this data on site with a hand held GPS unit. Instead of picking points along the boundaries of the lake property, points that were more easily recognizable were used. Road intersections, public buildings, and the edges of the dam bridge were used as points for data collection. This was done for a couple of reasons, the most important being that the lake's boundaries are not clearly defined all the way around. At many points, the boundaries are marked with fences or roads, but, at some points, the boundaries are not marked at all. After the GPS data was collected (in the form of UTM coordinates for 16 different points around the lake), I returned to a computer to apply the data to the map.

Before the UTM data could be applied to the scan of the map, it needed to be cleaned up. This was accomplished with the help of Adobe Photoshop. The .TIFF file of the map was just as huge as the map itself, and the dimensions were more than halved with a negligible loss of resolution. After that, the tone of the paper that showed up in the scan was removed, leaving a majority of the map's background as a clean white, as opposed to grey blurs. After the map was prepared, it was imported into Idrisi Kilimanjaro and saved as an Idrisi raster file so that it could be manipulated. The map was then manipulated through the RESAMPLE function of Idrisi. This took the 16 UTM coordinates (ten of which were used in the final resampling to insure the highest possibly accuracy) and applied them to the surface of the map. This warped the map so that it was correctly projected in the UTM-14N projection. However, it is not entirely perfect. The GPS unit is not perfectly accurate (at its most accurate, it is within 3 meters), and it shows up in the resampling. This is why some points were removed, because they caused the RMS error to be unacceptably high.

Now that the map was projected properly, it was suitable for use in a GIS. Images in the next section show how this all worked out.

Results of the Map Manipulation

Here are images of the map as it went through different stages in the process. Click for a larger image.

TThis first map is the original scan of the Kahola map after being cleaned and resized in Photoshop.

And this is the original Kahola map that has been resampled in Idrisi with the recorded GPS coordinates.

Here is a direct comparison between the unsampled and sampled versions of the same Kahola map. The unsampled map is represented in black, and the sampled version in red. From this, it is clear that the differences between the two are quite large.

This is an aerial photo of the area that was pieced together using imagery from Terraserver. Unfortunately, I was unable to resample this image, as Idrisi encountered file format errors and would not process it.

Further Results and a Few Words with the Industry

Although completing the tasks required for resampling the map took a fair ammount of time, that ammount pales in comparison to the time and resources that would be required to redraw the map by hand. Even if the map were redrawn by hand, it would still not be nearly as accurate as the new GPS defined map. The resampled map has much potential as a resource to the lake's caretakers. It has a clear diagram of the roads around the lake, and clear markings on where the individual lots are that follow the shores like the original did. Combined with the aerial photos gathered for free from Terraserver, this map could be a strong asset to the lake. In its newly projected form, it is much more accurate, and could be integrated into a larger GIS that could be used to help manage the lake.

Of course, the new map is not perfect by any means. The GPS unit is not perfectly accurate (the reason for only using 10 of the 16 points recorded was to minimize error), and inputting the GPS coordinates into the map's raster grid is not perfectly accurate either. The map still shows the errors that it had originally, such as the currently incorrect placement of the creek feeding the lake (it probably was correct at the time of the map's creation); the main difference now is the higher level of geographic accuracy the projected map has over the old hand-drawn map.

As another part of this project, I interviewed Becky Samuelson on her experiences with GIS technology for this project via email. She was kind enough to forward my questions to other people working with GIS in both governmental positions as well as private sector ones. Many interesting things were said, and some were quite relevant to the project. Rhonda Robidou of Kansas Gas Service had this to say about GIS tech:

"When I began my job 23 years ago we drew sketches of our projects on onionskin paper and had to go to maps that were in drawers to retrieve information. Now we are able to retrieve information for the field immediately using GIS and we are able to print detailed sketches for the field. The program also gives us very accurate information .... that aids in designing projects. GIS has increased both productivity and accuracy."

Her thoughts echo the major conclusion of this project: that modern GIS technology greatly simplifies the process of making maps while saving time. Gary Bacon, GIS/Business System Analyst with Oneok summed his feelings about GIS up succinctly:

"[GIS is] mandatory. I do not know how we operated without GIS before."

The full text of all the interviews is also avalaible.

Sources

Aerial photos were taken from Terraserver - http://terraserver.microsoft.com

Special thanks to Becky Samuelson for her help in making this project possible.

And thanks to everyone who responded to the interview.

All other data collection and manipulation by Jeremy Aber.