QUOTE (ngunn @ Jun 29 2011, 08:42 PM)
Yep, that would do it. Pete may have produced the profile first, then fine-tuned the location for the reverse visibility plot a little further SW to a point about 5m higher than the end of the profile. That, plus the 1.5m camera height, would be enough.
Sorry, probably a lack of information on my part. I made the visibility map first - this 'backwards' method means that I have to offset every point of the DEM by 1.5 m to simulate the camera height of Opportunity. The visibility is then determined from a series of lines all over and around what I think is the highest point (it's there or thereabouts) of Cape York. I think (still trying to get into the guts of this procedure) the actual visibility is worked out from the vertices of those lines, so it might not exactly cross the highest point itself. This is a powerful technique used for, say, checking the visibility of a road on the Earth, but I'm still getting to grips with it for Mars.
So bearing that in mind, I'd say the visibility threshold should definitely be taken as a general guide, not absolute, but on the whole it does seem to be quite close to Cape York.
I then just took a quick profile to show why that is probably the case. I'm afraid that I didn't go to the exact highest point, just through part of Cape York, so it might have led to the confusion here. So here are the raw data for a new profile going from the Sol 2635 position through the highest region of Cape York - I thought it might be easier this way if you wanted to have a look at the effects of vertical exaggeration (it really is one of the flattest places I've made a DEM of!).Click to view attachment
QUOTE (serpens @ Jun 30 2011, 07:45 AM)
Great visual representation there - love your DEM profiles. Please excuse my ignorance but just what is the vertical and positional accuracy of your DEM? Given the minimal vertical change compared to the distance baseline surely there would have to be wide error bars in the 'visible from' location. Shouldn't we take this as an indicative marker rather than as an absolute?
Yep, worth repeating, probably best as a general, rather than asbolute, visibility marker. Plus, does that get me off the hook if it pops up much sooner than I've predicted?!
Although some DEMs have errors, especially when you're looking at things of the order of the resolution, the vertical precision of the underlying DEM in this case is pretty good. Here are the details in case you're interested. Spatially, it's at 1 m/px, made from image pair ESP_018701_1775 (26.9 cm/px) and ESP_018846_1775 (28.2 cm/px) - so about 4 image pixels to get a single DEM pixel. We can estimate the precision using the method of Chris Okubo (Icarus, 207, 2010): the stereo convergence angle between images is 14.7 degrees, so assuming 1/5 pixel correlations with 0.282 m/px in the more oblique image, the vertical precision is 0.282/5/tan(14.7) = 0.21 m.
So although it's small, we should still bear that in mind when we're dealing with these flat areas and camera heights of only 1.5 m above the ground.
Hope that helps,