GIS Blog

On the Edge

We look at creating a catchment boundary layer using existing data and nice visualisations.

https://www.nasa.gov/image-feature/dividing-line  Image Credit: NASA/Joe MacGregor

So imagine you’re walking along a mountain ridge (like the one shown above from Antarctica – hope you’ve got your thermals on).  When it starts to rain, water flows to one side or the other, depending on where it hits the ground.  The line drawn along the ridge that separates these two flow directions defines the boundary between two river catchments and, along with some other key ideas, helps us define catchment boundaries.  In this post we’ll look at how we can manually create these boundaries making use of all available data and some nice visualisations.

Many may not be aware of Lincoln’s high country sheep farm in Otago, Mt Grand Station.  Near Lake Hawea,  the station has been the focus of some recent efforts to “redesign” its future, in light of climate concerns and sustainability.  It’s also been the focus of some water quality modelling by Shyam Provost, who’s been looking into phosphorus and nitrogen levels related to land use.  As part of his research, he’s needing to know the breakdown of different land uses within his catchments and to help out with that we’ve been working up a landcover layer that suits his purposes.  Part of that has meant defining the catchment boundaries, so we’ll have to start there.

Happily, a lot of high-resolution data have been collected at Mt Grand, including elevation at a 1 m resolution and imagery at 0.125 m.  I’d venture to say that no high country station has data of such quality.  It’s a veritable gold mine.

There are three catchments that Shyam is interested in, Cameron Gully, and Lagoon Creek with one of its unnamed tributaries.

the topo map helps gives us a sense of the catchment extents but we need to be a lot more exact about this.  As a starting point, I had a look at the River Environment Classification data from the Ministry for the Environment.  The main focus of this database is to classify rivers and streams into a hierarchy of different factors such as climate and geology (among six others).  There’s a copy of the data in J:\Data\River_Environment_Classification.  The Catchments folder contains shapefiles of catchments of different sizes, or river orders.  Higher order streams are larger and have more tributaries than do lower order streams – you could think of low order streams as the headwaters; the order increases as tributaries contribute more water to a river system.  I wasn’t sure what order these streams would be but figured they would be pretty low, first or second order.  Adding the first order catchments to a map shows that they’re too small:

 

But the second order ones capture these catchments pretty well (here I’m just showing the three of interest with black outlines):

Looking more closely, the borders seem pretty jagged, almost raster-like:

This isn’t too surprising if you know where these polygons came from.  The catchment boundaries were created using some DEM based tools.  In this case, it was a 25 m DEM with the output converted to vector polygons, which explains the jagged nature.  (In another post we’ll look at how these tools work.)  The key thing with these boundaries is they have to capture the areas that contribute runoff to a particular stream.  If we were to do this manually, we’d start at a point on the stream (in this case, where the water monitoring equipment is) and then move up the land slope, perpendicular to the elevation contours until we reach the highest points of the catchment.  Doing this on both sides of the stream helps define the whole catchment area.  We ended up doing this manually, making use of as much existing data as I could.

As noted above, our 1 m DEM could offer us some critical aide in this effort.  But in it’s raw form, it doesn’t tell us that much.  We’d get a lot more out of this using a hillshade layer:

The data don’t fully cover the catchments but we can work with this.  See how the hillshade helps define changes in slope which are tied to the topography?  Another arrow in our quiver is the high-res imagery which will be useful as a reference:

This doesn’t fully cover my area of interest either but I’ve got the built in imagery in Pro plus some other satellite imagery that I can call on.  Time to get started.

I first created a new polygon feature class for my catchments.  To demonstrate how this whole process worked I’ll focus on Lagoon Creek and it’s unnamed tributary.  The symbology is set to hollow fill to make it easier the see the line as they are digitised.  From the Edit tab, I can click on Create to open up the digitising tools:

Knowing where Shyam’s water sampling instruments were gives me a starting point.  This will be an irregularly shaped area so the Polygon tool is the best one to use.  To start, I clicked on the location of the sampler and then started drawing a line that went perpendicularly up the elevation contours (since water always flows downhill):

Moving around while digisting can be a challenge – to pan it’s easiest to hold down the mouse wheel to move around and left-click/hold/move mouse to zoom in and out – gradually the boundary line gets built up – I click once every time I want to change direction:

Here I’ve made it to a local high point but still a ways to go.  When digitising I can always go back to a vertex and either delete it (right-click on it > Delete Vertex) or move it around.  I’ll do a quick double check of my progress using the hillshade:

And the imagery:

Reasonably happy so far, though I’m a little uncertain about the area around 3/4s of the way up – Might be nice to look at it in 3D just to make sure – easy to do by converting this map to a scene (View > Convert > To Local Scene):

This is with the hillshade on but I can easily turn layers on and off and change basemaps to make the most of what the data tell me.  I’m finding this 3D view helpful so I’m going to go one step further and put these two views side by side and maximise their real estate so I can see both at once:

I can edit in either window (thought it is much faster in the 2D map – I really need a decent GPU!!!) and just need to be sure I hit the Finish Sketch button, , before moving from one to the other.  To further enhance this, I can connect to the two views using Link Views:

Now when I shift the view in one, the other moves to match.  This became very handy, especially when the exact boundary became unclear.  The linked views were an efficient way to do a bit of quality control while I was working.  To me, this is one of the big benefits of using Pro over ArcMap, multiple maps and views in one project.  (I also had Google Earth open at the same time as yet another point of reference.)

My strategy here was do a relatively quick first draft and then go back and refine it where needed.  With digitising, sometimes it’s easier to get that first draft and then do further edits using the suite of editing tools, particularly the Edit Vertices tool,  .  This one lets you create new vertices, reshape or delete existing ones to refine things by right-clicking on a vertex.

In the end I’m pretty happy with the results:

For comparisons sake, here’s my Cameron Gully boundary compared to the REC catchment:

It’s fairly different in certain parts and I think mine is 1) higher resolution and 2) a better representation of the catchment that Shyam was sampling.  Digitising is a standard GIS thing to do but hopefully you can see how better data and the ability to visualise things has made this an easier (though not a quick) task .   There’s more to say about this project so stay tuned.

C

• 12/08/2021


Previous Post

Next Post

Leave a Reply

Your email address will not be published / Required fields are marked *