11 replies to this topic

[Gretchen Peterson]

I'm tasked with determining the width of the canopy cover along streams. I'm thinking that it'll be a raster analysis where every cell representing the stream is assigned a value equal to the distance from that cell to the furthest canopy cell (which will be in another raster dataset) as long as the furthest canopy cell is connected to the stream via other canopy cells.

In this way a variable width canopy cover buffer could be produced. I considered using euclidean distance, viewshed, and cost path tools but none of these seemed to fit the bill. I admit to being flummoxed on this one! Has anyone done something like this before?

[Bryan Swindell]

Are your input datasets already rasters? I've done a similar exercise using vector data where I simply established 500-foot cross-sections along a stream, then intersected the xsections with riparian polygon data. The lengths of the resulting cut-up lines can be summarized by type, thereby giving you the average width of your polygon data.

[Gretchen Peterson]

Bryan - The data actually will be in vector format. I was just assuming that I'd have to convert to raster. If I didn't have to convert then that would be even better. Did you have an automated method for creating the stream cross sections? It sounds like your method might be the ticket.

[Bryan Swindell]

I've attached a working graphic of a portion of my project area, just to give you a visual. The entire process is not exactly automatic, but it answers several "what is the average width/distance to edge of floodplain units?" questions.

On a big river, or a small one that is very sinuous, 500-foot xsection spacing is usually plenty. On a smaller river, 100-foot stationing might work fine. Admittedly, I use AutoCAD to automatically draw lines perpendicular to my measured stream centerline every x-feet along the stream - there may be a GIS tool out there that does this, but I'm not aware of it. Sorry!

Obviously, once the xsections are drawn some may need to be manually adjusted. Then, just intersect and summarize!



 fig2s.png   767.05K   56 downloads

[David Medeiros]

Are your input datasets already rasters? I've done a similar exercise using vector data where I simply established 500-foot cross-sections along a stream, then intersected the xsections with riparian polygon data. The lengths of the resulting cut-up lines can be summarized by type, thereby giving you the average width of your polygon data.

This sounds about right, but why wouldn't you just create a single 500' buffer for the streams before intersecting the canopy polygons?

edited to add: Answered my own question. I didn't see that Gretchen specifically needs width of canopy versus just a riparian polygon.

[Gretchen Peterson]

David - intersecting a buffer with the canopy cover polygons gives a great measure of total canopy within the buffer but doesn't get at that slightly more interesting statistic of connectivity of canopy to the stream. So if there is a patch of forest that starts at 300' from the stream, for example, it would not be counted because it does not touch the stream. In fact, both measurements are highly useful and total canopy within a set buffer is definitely one of the metrics that will come out of my analysis. It was the connectivity bit that was alluding me.

Bryan - thanks! I understand your process. I'm not sure if I can create those cross-lines in Arc but I will do some searching to figure out if it is possible. If not, I may be able to find a function in GRASS or some other open source GIS. Thank you for your help!

[Bryan Swindell]

Great - if you find such a tool, let us all know!

[David Medeiros]

David - intersecting a buffer with the canopy cover polygons gives a great measure of total canopy within the buffer but doesn't get at that slightly more interesting statistic of connectivity of canopy to the stream. So if there is a patch of forest that starts at 300' from the stream, for example, it would not be counted because it does not touch the stream. In fact, both measurements are highly useful and total canopy within a set buffer is definitely one of the metrics that will come out of my analysis. It was the connectivity bit that was alluding me.

Yeah, I was initially thinking you were only after a general riparian polygon. As for connectivity I assumed you would select the non contiguous canopy polygons and exclude them from the intersection.

[david17tym]

Hi Brian/Gretchen,
About generating lines perpendicular at set intervals, ETGeowizards can do this, see:
http://www.ian-ko.co...tationLines.htm
Dave

[frax]

Is digitizing by hand (and interpreting visually) out of the question? Sounds like it wouldn't be too much work to do that.

[Gretchen Peterson]

Thanks david17tym, I'll check that out.

frax - I thought about that but the study area is quite large and I will be doing this to about 13 river systems. If it were just one I might just hand-digitize it.

[Gretchen Peterson]

Thanks for your help everyone. I still haven't created the buffers but I do think I discovered a method that may work, when I get the proper data for it.

Given: a stream polygon layer that outlines the banks of the streams
Given: a tree raster layer (1=tree, 0=no tree)

I have the tree layer (derived via NAIP 1 meter, 2009, 4 band imagery supervised classification) but I only have stream lines, not polygons. The NHD has polygons that would work except not for every stream in my study area. We are thinking about delineating these by hand but it will take some time.

Analysis: Cost Allocation (via ArcMap 10, spatial analyst, in this case) whereby the tree layer and the stream layer are input. The output looks like this (see rough screenshot), though you can see that it would be better if the stream line represented the banks rather than the stream center line.

Attached Files

•  costallocation.jpg   150.14K   26 downloads