The uncertainty of data

"My goodness Toto, we don't seem to be in Kansas any more"

It all seemed so easy: I had markers giving latitude and longitude of all the places where things should be positioned, and I had height data giving the contours of the land. All I had to do was follow the markers, laying track and raising or lowering the terrain beneath it to form the cuttings and embankments. At the southwest, Shepherdswell mainline station and Lydden tunnel seemed to be about right, but going towards Golgotha tunnel there was no sign of the spur of high ground where the high siding ran to meet the track at the level crossing. The line dropped satisfactorily down to the portal of Golgotha tunnel, emerged at the far end in another cutting, and ran down to Eythorne. But at Elvington, I found a mountain just to the north of the platform which didn't agree with any of the few photos I had seen.

At Eastry I couldn't get the cutting where the loop started and also get the twin embankments curving off to Wingham and running straight down to Poison Cross. I could lower the track and run into Eastry through the cutting, but then the lines to Wingham and Richborough were flat on the ground. If I lifted the track level so that the two embankments appeared, there was no cutting at the other end of Eastry platform. I did of course have the problem that I didn't have an accurate gradient profile for the route, but was relying on one in the Locomotion Papers book, I had taken several datum points where I would assume that the line was at the same level as the ground. One of these datum points was the pair of level crossings at Poison Cross, another was the level crossing at Eythorne (subsequently discovered to be a poor choice), yet another the level crossing at Knowlton. I was forced to work out from each of these places, setting out trackbeds in TSTools2 to the gradient I had established ran between the two datum points, then cutting the trackbed and going back into the Route Editor to lay the pieces and see if it looked reasonable. Far from being an exact exercise in using latitude, longitude, and shuttle-obtained hieght data, I found myself guessing just as if I had raised the hills and lowered the valleys using the built-in tools in the editor directly.

In an attempt to reaasure myself that I did have the SRTM files for the map I was using, I climbed to the top of the slight mound on which Richborough Castle was sited, knowing that it was the highest spot for miles around, and discovered to my consternation another mound, even higher, half a mile to the East. I flew across to this new mound, and saw that it had a depression in it that looked very like the amphitheatre shown on the map near the trig point beside Richborough Castle. But if the data was indeed half a mile out with respect to the map, what was the first mound? There was no other place half a mile to the west to correspond to it. Nevertheless, I tried re-aligning the map and data so that the new mound was the site of Richborough Castle, but found that the rest of the line no longer made sense. Lydden Tunnel and Golgotha were now wrong, Eythorne station was on top of a hill, Elvington station likewise, and the line from Knowlton to Eastry South was on the wrong side of the valley. My original alignment made better sense, but still gave me doubts.

I went off around the web learning a little bit more about SRTM data. This is what I found

• SRTM data at 90 metre resolution will not pick up fine detail like the North Bank at Shepherdswell.
  • This means it won't show the cuttings or embankments of a railway line, or a small rise in the land (such as at the approach to Eastry station), where the rise and fall occurs within 90 metres.
• It can be confused by clouds, or by rapidly changing landscape.
• It can also be confused by a body of water.
  • That got my attention, and I plotted a few more marker points. The unexpected hill I had found to the east of Richborough Castle corresponded exactly with the position of the Stonar lake.

So I had an answer for the missing north bank at Shepherdswell and the lack of a cutting at Eastry, and, more importantly, I could trust the position of the mound at Richborough Castle, but what about the mountains at Elvington? I realised that there was only so much I could glean from the maps and Google Earth. I would have to go and see for myself.

And so the story progresses to Ground work