I’ve used some contour maps recently for hiking and notice that they are pretty comprehensive. You don’t see areas marked “Elevation not available.” How do they collect such a massive amount of information, especially for large areas that are barely accessible on foot?
I assume they are based on satellite terrain data.
It’s done from aerial photography. A plane flies in a line and snaps photos that overlap about 60%. If you look at one photo with one eye and the other photo with the other eye, you can see the terrain in 3D.
Now, if you have targets on the ground that you can see in the photos and you have surveyed the targets so you have accurate 3D coordinates for them. You can tell your computer to do a least squares spatial adjustment. The photos are put in a machine called a stereo plotter and the results of the spatial adjustment are used to setup up the photos so that an operator, called a photogrammetrist, can look at the photos, one in each eye, and make accurate 3D measurements.
The photogrammetrist sees two dots and their eyes are focused on the ground. By making an adjustment on the stereo plotter, they can read the elevation and by moving around and keeping the dot on the ground, they can draw contours.
They used to make them from stereo photos using equipment that lets a human operator trace the contours, called a stereoplotter. Nowadays it’s mostly computerized but the principle is the same.
What about nautical maps, which show the contours of the bottom of your lake or river? They can be incredibly detailed, yet the only way I can imagine they’d be made is by dropping weights here and there to see how deep the water is at each spot.
I remember using them in the early 1970s, and even then they were decades old.
For a shallow lake or river, they probably do drop weights on the end of a line. For deeper water, there are techniques based on sonar.
Also, especially for challenging areas - the seafloor, particularly deep seafloors, for example - there are decently large error bars, resolution limits, and some interpolation at play.
The company I work for has done a lot of high resolution multibeam sonar bathymetry mapping. At a large scale, it looks very detailed and high resolution (and it is by historic standards!) but if you dig into the data at a finer scale, you clearly see the resolution limits and place where measurements don’t always agree exactly (especially clear at intersection points of any orthogonal sail lines on the grid used for the mapping).
Hydrographic charts of, say, the entrance to a harbour, were compiled with hundreds or thousands of individual depth readings. They show safe passage rather than the nuance of seabed shape, but that can be done now with a range of remote sensing techniques using either direct or indirect indicators of depth.
For land survey, yes its all done by satellites and computers now, but in various field based sciences part of your training used to be to produce a credible contour map with minimal gear and variable data. Once your head is primed to work in 3D its almost nonsensically easy.
Just as an aside contours aren’t the only means of conveying relative height and terrain info. Far more beautiful and harder to produce are hachures [but so satisfying when you’ve done it right].
There’s a whole slew of reefs reported in the Pacific that don’t seem to exist. One, Filippo Reef, was reported to be 0.6 to 0.9 meters under the ocean. More recent measurements show the ocean depth there to be 5.3 km.
Here’s another dubious reef with links to several others. And an infamous phantom island.
The knowledge of the fairly shallow parts of the oceans is not that great. Nevermind the deep parts.
In addition to reefs on maps that aren’t really there, we occasionally have rather expensive problems with undersea mountains not on maps that really are there: USS San Francisco (SSN-711) - Wikipedia
About 20 years ago the US spent quite a lot of the taxpayers’ money on this project which produced a lot of very nicely detailed terrain elevation maps of places we would have had a very hard time sending survey teams to, if you get my drift.
The Shuttle radar mapping data is often the go to dataset for a lot of work in remote places. Mining, oil, and gas exploration especially. Indeed it can often be the only dataset.
How do they compensate for heavy foliage, like 30-foot trees, when the ground is not visible from the air?
Modern systems can also use LIDAR for scanning the terrain. Fundamentally the same process, flying back and forth with a plane, but scanning in a different manner.
Air photography has two seasons: spring and fall during snow-free, leaf-free conditions. There can still be problems.
ETA. At least that’s what they do in my temperate climate. Don’t know what they do in the tropics.
And that’s what they use in tropical regions nowadays. I expect that previously they just had to live with high error bar data. Lidar has also been used to find lost cities in heavily forested terrain. Here’s the most recent example:
I believe the contour line concept was developed accidentally. I saw a film years ago of how it was stumbled upon. But basically tracing the projection. Pretty interesting. Sorry no more details.
I recall reading about a dam they built in the Amazon many many years ago (late 60’s?). It was particularly in the news for the rescue efforts for wildlife stranded in submerging islands as the water level rose. there was an off-hand mention that they had to stop filling the dam to take corrrective measures because they had misjudged the terrain since it was all hidden under the forest canopy and water was bypassing the dam in assorted hidden valleys. So in the days before LIDAR and satellite measurements, the stereoscopic aerial photos produced educated guesses when the forest canopy was too dense.
I took a computer cartography course once, and one algorithm was for producing a contour map from a grid of elevation points, by interpolating the grade lines from connecting the dots between two points of elevation. If A is 50 feet and B is 120 feet, then the 100-foot contour passes between them at point C, simplifying by assuming a constant slope… “close enough”.
This is exactly how it is still done today, for small construction and engineering projects. Land surveyors walk over the area taking measurements, and then a computer connects the dots.
I would think they are using LIDAR now. It’s crazy accurate.
Back when I was an ink on mylar draftsman, a surveyor would bring plotted points and the elevation of them. Soooo… as the draftsman, I did my best to draw the contour lines between all the points. This was on small properties an acre or two. But it involved a shit ton of guesstimation.
Accuracy is a weird thing in some cases. “How many angels can dance on the head of a pin” stuff.