Subscribe to this thread
Home - General / All posts - Smoothing scalloped lines - stream buffers
Forest
594 post(s)
#24-Jul-19 07:31

In some jurisdictions, vegetation is protected for a given distance from a stream centreline. In the clip below, the distance is 40 m. However, when a winding stream is buffered, it makes smooth rounded convex curves facing away from the stream and sharp concave points pointing toward the stream. From a land management point of view, this is not a desirable boundary. I wonder if there an algorithm for removing these sharp points. In the old days of manual drafting, one would just slide a circular object of the correct size up to to the buffer until it touches the buffer line at two points, then trace around the edge of the object. That is sort of the solution that I am thinking off but haven't worked out how to do it in a GIS yet.

Attachments:
StreamBuffers.jpg

tjhb

8,805 post(s)
#24-Jul-19 08:37

Can I please be really dumb, and ask where is the stream centreline that has been buffered in the example?

I can't see any stream centreline (except by eye from the photograph).

There are two straight area edges and two rounded area edges. There is not an obvious relation between them. Some overlaps (I think) are shaded in red, and some differences (I think) are shaded in green.

But no centreline.

(I am happy to be embarrassed by having read it wrongly, with or without colourblindness.)

Dimitri


5,491 post(s)
#24-Jul-19 08:38

vegetation is protected for a given distance from a stream centreline

and

From a land management point of view, this is not a desirable boundary.

Agree 100%.

I'm looking at exactly that situation myself, where some parcels I'm dealing with have pointed intrusions into them that mess up land management. Where did those pointed intrusions come from that stick into what otherwise would be a nice, regular parcel?

They are buffers around seasonal small streams, not even visible today except as slight ravines in the forest, that many years ago got recorded into the cadastral system and thereafter took a bite out of parcels that later on were surveyed. Here are two examples:

The field example is particularly annoying, in that it (in theory) prevents the farmer from plowing straight furrows right across a slight depression that is barely noticeable in the field:

But... how to deal with that depends on how much effect the law has, and whether you have to follow the law or are free to do something different.

If the law says "vegetation is protected in a region up to 40 meters from a stream centerline" that is going to create sharp angles in some cases where stream centerlines are bendy.

Consider the following image:

That shows a 40 meter buffer (in preview color) built on the stream centerline (black). If the law says "40 meters" you end up with acute angles in the change of direction of the buffer boundary.

But, if the law gives you some slack, in Release 8 you can convert the boundary area to a line, and then use the Spline transform toolbar command to smooth out acute angles.

Attachments:
spikes_field.png
spikes_forest.png
stream_buffer.png

volker

1,035 post(s)
#24-Jul-19 08:44

Have a look at this post:

http://georeference.org/forum/t98588.21


http://www.thegisservicesector.de

tjhb

8,805 post(s)
#24-Jul-19 09:03

But Dimitri, your examples (like Forest's) don't seem to show the centreline of a stream. The yellow lines look like parcel boundaries. That is a completely different question (legally as well as in geometry).

I'm not reading it right?

Dimitri


5,491 post(s)
#24-Jul-19 09:27

The photo/green line examples show a case where legal requirements caused boundaries that are inconvenient for land management. No streams are visible so you have to guess where they were, starting somewhere within the parcel and then flowing out.

It's an analogous situation, if you take the problem to be one where legal requirements result in inconvenient boundaries that contain acute angles instead of smooth, spline-like, paths.

The buffer in preview color is the one with a stream centerline.

tjhb

8,805 post(s)
#24-Jul-19 09:49

Yes.

If the current legal requirement is to do (or not to do) something within distance X of a stream centre line, then the first order of business is to find the stream centreline--wherever it currently is. That may be tricky. It might need a surveyor or might not--and at the next flood, the centreline might move. (The whole stream might move.)

If the legal requirement is to do something (or not) within distance X of a legal boundary, then that is different. Then the current or historical position of a stream are not directly relevant...

...Except if, as in common law, property boundaries follow stream or river centrelines. In New Zealand that is usually the case, except for navigable rivers which have different rules.

Nevertheless in New Zealand, property boundaries are drawn as if they excluded historic stream beds entirely--there appears to be something in between riparian properties called a "hydro parcel". These appear in the cadastre. But they are a fiction, existing only so that the surveyor could avoid getting his feet wet (or drowning). The legal boundary is normally wherever the centreline of the river happens to be today (with exceptions).

I'm only mentioning this because it is often misunderstood, and may also be the case in other jurisdictions. The shape of the buffer might be the least tricky problem. Finding out what the rule really means matters more.

oeaulong

203 post(s)
#25-Jul-19 03:23

Forest,

A couple of other approaches. assuming a 40m is reasonable.

  1. Take your centerline, Buffer (40m). Then take the result and run it through a Convex Hull. Would this satisfy a similar border?
  2. Take your centerline, Buffer (80m). Then Inner Buffer (40m). This seems to remove the sharp acute areas while keeping a wavy boundary. Might this work?

In your example photo, near the bottom the width seems greater than 80m wide "corridor", could this have been done, not with a centerline, but with a more precise water channel boundary area taking into account widening and narrowing of the stream channel?

Forest
594 post(s)
#25-Jul-19 05:33

Thanks oeaulong,

I will try idea number 2. I thought about using a convex null but that would not work if the river has a larger curve as well as smaller meanders which is common.

Regarding the legal side of things, the results will go out to public consultation and landholder consultation and after the consultation period, things will get locked in, so the actual shape of the boundary is negotiable to some extent. Having a convoluted boundary could be a problem if tree clearing is banned on one side or if cleared areas need revegetation. I think it is in every ones interest to have smoother boundaries. The work I am doing is in relation to a town planning scheme, so is not as defined as State legislation, which would be rigid about a 40 m buffer. The stream is also in an alluvial area and the stream channels are constantly moving. These situations used to be covered by ambulatory boundaries, which move with the stream but as farms on one side of a river loose ground and farms on the other side gain ground, these have become unpopular with farmers. They would rather have the river split their land than loose land.

I just thought that this is a really interesting GIS problem. The solution that popped into my head would be to select a line of a fixed length, say 100 m and start the line at a vertex in the scalloped buffer with the far end being located at a point on the buffer 100 m away. If the midpoint of the line was inside the buffer, nothing would happen, but if it was outside the buffer, a new vertex would be created at the midpoint. The unchanged and changed points would be captured in a new component but it would smooth the line and get rid of the spikes, something which simplifying geometry would not do.

tjhb

8,805 post(s)
#25-Jul-19 05:33

No.

If there is a legal requirement, then that is it. Being clear about this is where to spend all the time and money.

Even if it is geometrically stupid, it will count as being right. It is law + maths.

Adding nice GIS smoothness is irrelevant and wrong (and potentially very, very expensive for those undertaking it, so don't).

oeaulong

203 post(s)
#25-Jul-19 09:06

No.

?? No to what? Was it my last paragraph trying to decipher the origin example photo's layers?

I don't understand why this was in response to my suggestions on a geometric fix.

Are we talking about a clean polygon giving a bufferage to a stream so field operators don't get their feet wet or worse, or are we speaking of similar allowance of vegetation growth for erosion mitigation, or are we speaking of legal ramifications of a parcel boundary on an stream meander? Right now, I'm not so sure what this discussion is really about after re-reading the comments.

If there is a legal requirement, then that is it. Being clear about this is where to spend all the time and money.

Even if it is geometrically stupid, it will count as being right. It is law + maths.

Adding nice GIS smoothness is irrelevant and wrong (and potentially very, very expensive for those undertaking it, so don't).

My turn to be dumb here. I thought that the geometric smoothness *was* the solution desired, not the legal ramifications. I liked the second option I presented as a simple, scriptable, automaton based around the streams center-line, however they were made and updated. I am not making any proper claims on legality nor budget here at all, as I have no position in background nor context. I just don't appreciate the attribution of my response as "wrong" when offered honestly for a possibility of a manifold-based graphical solution as presented.

My only experience with GIS concerning alluvial plains is in the Mississippi Delta where projects were annually flooded and islands with old land parcels can be removed in a few months of flooding. Were it my project, I would probably approach it quite differently in terms of meandering parcel and administrative bounds, source of the meander, layering of exclusionary areas based on fixed or seasonal streams. I would like to hear more of the project and how answers to issues get addressed and less of the curt responses. Thank you.

tjhb

8,805 post(s)
#25-Jul-19 09:16

In short, look at the first sentence of the first OP post.

Your inner buffer of an ordinary buffer was a great idea... but only if the legal requirement is different from what was said.

And we still haven't seen any actual centrelines (except one fictitious example from Dimitri, helpful although too easy). Not from Forest, for example.

oeaulong

203 post(s)
#25-Jul-19 19:04

And we still haven't seen anyactual centrelines (except one fictitious example from Dimitri, helpful although too easy). Not from Forest, for example.

It was not based on a centerline, but either two centerlines or an area defining the water course.

attached is my deconstruction using buffers of 40m around the points defining the perimeter. This is approximate using a local scale of 0.44m for the image.

Attachments:
deconstructedStreamCourse.PNG

Forest
594 post(s)
#26-Jul-19 01:31

Some actual centrelines.

There are two areas with different stream morpologies here. For one set of streams, the inner buffers idea works well and for the crowded area, it does not work so well. At least, I would need to check the results and decide if I wanted to keep the original or the smoothed version. I unioned the buffers as the stream centrelines are in small separate segments and produced overlapping buffers.

Attachments:
InnerBuffers.jpg
InnerBuffers_Crowded.jpg
StreamCentrelines.map

tjhb

8,805 post(s)
#25-Jul-19 10:00

To put it another way, I am sure that lawmakers everywhere are restricting land use in unrealistic or unworkable ways, partly because they have little GIS experience or incompetent GIS advice.

But after they pass such a law or regulation, "better" smoothing by a person who has more GIS experience is only a legal liability for that person. No one can gain.

adamw


8,634 post(s)
#25-Jul-19 13:59

In the old days of manual drafting, one would just slide a circular object of the correct size up to to the buffer until it touches the buffer line at two points, then trace around the edge of the object.

This might not be a bad idea for a geometry function - take an area and a smoothing radius and smooth area boundary so that a circle of the specified smoothing radius can reach the entire boundary while being completely outside of the area.

I think there is a fairly close approximation to this that you can do right now - say, you want to smooth out everything to the radius of 10 meters. You take the original area A, create yet another 10 meter buffer of it: B = buf(A, 10). Then you take the boundary line of the buffer and create a 10 meter buffer of that line: C = buf(boundary(B), 10). Then you take B and subtract C from it. The result should be 10 meter smooth.

I guess this subtraction of buffers in my second paragraph might actually be one possible implementation of the imaginary smoothing function I describe in my first paragraph.

PS: ...and I guess that's exactly approach 2 in oeaulong's post above. I should have read the thread first.

Manifold User Community Use Agreement Copyright (C) 2007-2017 Manifold Software Limited. All rights reserved.