The end result I would like is a map showing how many new electricity poles are visible from various locations. Most of the map will show 0, some will show 1, or 2, 3, etc. I would probably cap this at "5 or more".

My proposed method:

Create a surface when I go away for Christmas, because the area I am talking about spans from one side of New Zealand to the other. I realise some of you may have worked with bigger areas but I have Manifold running on a PC (all be it a fast one) not a server.

I intend to do this by: combining all my z values into 1 drawing, copying them when I go home on 23rd Dec, then coming in briefly on 24th Dec to past them as a surface then letting it run for nearly 2 weeks.

Then, in January, create a viewshed for each pole. I realise this will take some time. There are lots of them.

Then use a yet to be determined predefined function to combine the resulting drawings into one with a field specifying whether 0, 1, 2, etc drawings intersect.

Then play around with the map and layout to get them looking good. This is the easy part.

Does anybody have an suggestions on how I can do this better, or am I on the right track? Am I right in thinking there is no way to select data from multiple drawings in Manifold like there is in ArcView? (Maybe ArcView has some redeeming features).

Hi Kaliana

What part of NZ are you looking at?

I have already generated a surface for some areas, that may save you the two weeks processing.

Contact me at nglover@xtra.co.nz if you want.

In another context Klaus has promulgated a saying, "Don't eat the elephant all in one bite". That would be good advice here as well.

From how far away can you make out the any one power pole, even given the clearest weather, a flat Earth, and a following wind?

What I'm saying is, you'd be better to break the task up into workable chunks. There's certainly a way to do that.

From how far away can you make out the any one power pole, even given the clearest weather, a flat Earth, and a following wind?

Yes, I would look at a maximum radius of 5km (although, with a following wind...). It may be necessary to create the whole Surface first, though, and then copy and paste sections of it for each sub-section of your study area to avoid edge effects caused by the interpolation method and to create continuous viewsheds... For example, what do you do when a pylon is near the edge of the Surface (especially if it's in a highly visible area)?

As for the multi-viewshed that you want to create, you can try using the spatial overlay methods (we had a somewhat inconclusive discussion on something similar at http://forum.manifold.net/forum/t64602.10). I suspect things will get fairly tedious if you have a couple of hundred pylons, and you'd probably want a script. Or am I way too negative on this topic? I will go back to the script in the link above and see if there's a way to get this automated...

It may be necessary to create the whole Surface first, though, and then copy and paste sections of it for each sub-section of your study area to avoid edge effects caused by the interpolation method and to create continuous viewsheds... For example, what do you do when a pylon is near the edge of the Surface (especially if it's in a highly visible area)?

Assuming a radius of 5km (for the practical limit of visibility), each pole or pylon needs to stand within a continuous patch of surface measuring 10 by 10km. (Obvious enough.)

For this to be true for every pole, I think each tile should measure 20 by 20km, with an overlap of at least 10km between adjacent tiles. (So that, if a given pole is closer than 5km to the edge of tile A, it will always be less than 5km from the edge of tile B.) [Corrected radius and overlap, from 15 and 5km to 20 and 10km, after drawing a diagram.]

Do you agree with that Henry?

Those sizes assume that the edge effects from interpolation have been taken care of by another means, such as blending adjacent tiles, with or without an extra overlap. How much was necessary would depend on the source data and the interpolation method.

If, say, LINZ contour and height data were interpolated in Manifold using DEST, then I'd expect 20km tiles, with a 50% (10km) overlap, to produce a good result. After interpolation, I'd blend each surface tile with its neighbours, using a weighted average dependent, for each pixel, on its distance from the centre of the tile (or on the square of the distance).

Tim,

I can't disagree if you've drawn the diagram. Surely?

It would be interesting to get some more info on the project from Kaliana. What type of power line? Distances involved (total and between pylons)? etc. My introduction to visual impact assessments was through large power line projects here in South Africa. I use TNTmips to generate the viewsheds and cumulative viewsheds for pylons, and then I do everything else in Manifold using the imported surfaces.

"From how far away can you make out the any one power pole, even given the clearest weather, a flat Earth, and a following wind?"

Are we talking pylons or poles?

The angular resolution of the human eye being what it is, I doubt the eye could resolve a typical 0.3-0.4m diameter pole from distance of more than 2Km distance on the clearest of days.

For an observer at a height above sea level sea-level, the maximum distance of the visible horizon is given by the expressions presented here:

http://en.wikipedia.org/wiki/Horizon#Exact_formula

Some follow up to what i stated above.

The angular resolution of the human eye is 1–2′ (about 0.02°–0.03°), which corresponds to 30–60 cm at a range of 1Km.

See:

http://en.wikipedia.org/wiki/Naked_eye

So if you are performing a viewshed analysis to determine whether or not an object will be visible, you need to take into account the dimensions of the subject being viewed. The bigger the subject the greater the distance it will be visible from (up to the limit of the visible horizon). There is therefore no purpose in performing a viewshed analysis to see whether a small object like a utility pole will be visible at a range of more than a few kilometres, because it wont be, unless the observer is using a telescope or is superman!

Of cours the above does not apply to viewshed analysis performed for telecoms purposes.

ah! that's why I loose connection on my celphone standing behind a 30cm pole ...

Hello everybody. I'm sorry I haven't replied sooner. I forgot about this post so I wasn't checking it.

Are spoedniek and tjhb suggesting performing the analysis on small surfaces, not feeding in the whole thing? The area is also quite hilly, but as there is a wind farm involved there could be poles on ridgelines. There is also a harbour and a river, so there could be underwater cables too. This is going to be a pole line not towers, and it will be a 220 kV line.

The total area of the whole study area is aproximately 1000 km², in an L shape, currently with lots of alternatives although there will gradually get less and less and the area will get smaller. The final line could be 80 km long, which might not be that big by international standards but it is fairly big for us.

I will also get approval from my manager to email Rakau about getting the surface from him.