Claire Heald’s BBC News article “The map gap” is an interesting read (and name-checks Ogle Earth — thanks!) but at the risk of betraying my non-GIS background, it seems to me that some of the quoted cartographers are not all that acquainted with virtual globes. First, there is this:
On a world scale, the pictures on Google Earth, and programs like it, look accurate. The view is less skewed than with some other projections, such as cartographic models used to set out the world.
[…] And [Google Earth] still does not solve the age-old problem. On a world scale, all’s well, “but when you zoom in you still get that distortion, because it’s on a flat screen”, says Mr Lennox.
I would argue that 2D representations on computer screens of 3D virtual globes are not skewed or distorted at all. They show exactly what a one-eyed astronaut or one eyed airline passenger would see. But even if I was wrong on that point (and I don’t concede it), wouldn’t zooming in on Earth reduce distortion, as you’re getting less and less curvature to deal with? The analog with maps is that on the most local scales, the Earth really is flat, so distortion disappears.
My one other criticism concerns this quote, by Steve Chilton, chair of the Society of Cartographers:
“Google Earth is just the satellite image. It doesn’t show us land use, slope, precipitation. So the need for cartographers still exists. The paper map hasn’t died.”
But in fact Google Earth, NASA World Wind and the upcoming ESRI ArcGIS Explorer can show us these kinds of data — exported from the same datasets with the same professional tools that are also the backbone of paper maps. Whether the operators of these tools call themselves cartographers or GIS specialists is a matter of semantics. The advantage of paper maps is that they don’t run out of batteries. The advantage of virtual globes is that they can show live data — interactively, and without distortion.
No, the paper map hasn’t died, but it is suffering the same fate as paper editions of newspapers.
Yes, a 3D virtual Earth can be perfectly accurate and rendered so as to be indistinguishable from what an astronaut would see from space. In fact, the imagery is often so far away from the virtual POV that stereoscopic pairs aren’t usually needed until you get close to the ground.
The only significant 3D distortions come from a) treating the Earth as a perfect sphere vs. an ellipsoid; b) having your eye position not match what the rendering assumes, e.g., off axis, too close or far from the screen; c) errors in orthorectification; and d) resolution of imagery or terrain.
Maybe they were confusing 2D map distortions from 3D projections? I don’t know. Paper maps usually have worse distortion from projection, point of view, and dynamic resolution, but the same issues with orthorectification of the sources.
Item (b) is interesting because it means that to avoid distortion, you should sit at a distance from your monitor that matches your real field of view to the virtual one. If you do, it’s as close to the image you’d get through a window as possible, ignoring rendering quality. Once computers start eye-tracking, they can match your real position vs. the other way around so that won’t be as much of a factor.
Stefan:
My take is that there really isn’t an apples-to-apples comparison between traditional 2-D paper maps and the emerging virtual globes technologies. To me the interesting question is whether 3-D-ish interfaces are a pre-cursor to ubiquitous true 3-D, holographic displays, or whatever else a Minority Report-addled imagination can come up with. And if they do become ubiquitous, will the human skill of making sense of 2-D representations of the spatial world–thousands of years in the making–decay precipitously? Lest you think I’m overstating the case, consider how in the world’s wealthiest societies, the ability to memorize large quantities of information has declined rapidly in just a few generations…and nobody is mourning the loss.
Brian
Just found out that the horizontal field of view of Google Earth’s screen is 60 degrees. I take this to mean that if you open one eye and place it in front of the screen so that it forms an equilateral triangle with the edges, you get the exact view as intended.
Stefan,
I was invited to give a talk at Society of Cartographers of which Steve is president this summer, I think there has been a translation issue between the BBC interviewing the cartographers and the final version. You are quite right that a virtual globe does not suffer the orange peel problem but I think the point in the artical was trying to make was that the representation on screen of a virtual globe is like a map: You can use a piece of string on a real globe to measure the distance between London and New York but you cannot use a piece of string to measure that distance on a *screen* that shows a vitual globe – you have to use the software function instead. It wasn’t well explained in the article.
As to Steve’s comment that Google Earth is just satellite data, I showed them lots of examples in my talk and Steve had gone to the lengths of releasing the location of the venue as a kml file. He is well aware that you can add layers on top of GE. I suspect he has been mis quoted in some way. My point in the talk was that the need for Cartographers has not dissappeared with the appearance of virtual globes because the design of a lot of kml files I see is terrible.
Richard Treves
You’re absolutely right about the production values of most KML. A lot of KML needs to have an Edward R. Tufte book thrown at it:-) Paper maps still look a lot nicer.
Stefan, yes, exactly. The 3D graphics could, if the actual eye position was known, account for distance or offset too (vertical and horizontal). FOV is the simplest adjustment to the math we can make.
The best way to think of the math is as if someone placed a piece of photographic glass in the virtual world that faithfully captures the light that hits it. As long as that piece of glass matches the shape and orientation of the 3D window on your monitor, you’re getting the ideal view.
The old axiom is, “All maps lie because all maps lie flat”.
And while GE doesn’t exactly lie flat, it is a 3D rendering on a flat screen. It is a question of nadir being the only 100% accurate portion of a flat map or map on a screen. Now, a holographic representation would be nifty, but we just don’t have those yet, as far as I know.
Yes the local detail in Google Earth have positional errors and bias. On review of a decade of spatial imagery, there are two source IMHO. Image registration… and projection bias. Lastly, depending on how you create geoEXIF, they may also be some rounding.
In general I am finding that Google Maps to have far fewer systematic error.
MidNight Mapper
I am trying to line up a Google Earth Pro image with a regular Google Earth image of the exact same area. Our CAD person says that the image from the free version is correct but that the Pro version has some horizontal distortion. Is this a bug in the new version of Google Earth Pro? I am using version 5.