The GIS Detective – Have LiDAR, Will Travel

In this post we’ll see how LiDAR data were used to help understand a tricky problem.

So I’ve got a bit of an issue at my house – and here it is:

We live out on the Lyttelton Harbour on steep loess soils, which are prone to tunnel gullies – sub-surface channels that develop in the soil.  What’s shown above is the outlet of a tunnel gully below our house.  Over time, water flowing out of the outlet has carved out a rather large hole.  Hard to tell without something for scale but at its worst the hole carved by flowing water is about a metre and a half deep and over a metre wide.  It’s quite common on the peninsula and one doesn’t need to look far to see evidence.  Arguably, they are one of the dominant modes of erosion on the peninsula.  Here’s a picture of a spur near Motukarara with some significant erosion and collapse of tunnel gullies visible – they are the long lines running down the face of the slope:

Tunnel gullies can cause problems with with drainage, as they tend to collect water into those sub-surface channels, sort of like natural pipes, that must have an outlet somewhere.  Since our tunnel gulley is at the bottom of our section, it can cause the occasional problem for our downslope neighbours when we get a lot of rain.

One thing I’ve observed over the years is that there are several areas where water gets into the tunnel gully, and two in particular.  When we or our neighbours above us water our respective gardens in certain places, we get water flowing out of the hole.  I’ve often surmised that somewhere uphill from us is an inlet – a hole in the ground where it starts, but I’ve never found it.  Wouldn’t it be nice to know where that is?

Enter LiDAR data.  We covered the ideas behind LiDAR in detail in a previous post.  It works a bit like radar, only using laser light instead of radio waves.  Imagine an airplane flying over Christchurch.  The plane has a highly accurate GPS system and an onboard “inertial measurement unit” which keeps track of the pitch and yaw of the aircraft – these two help the airplane know where it is and what direction it’s facing.  The plane also has a laser system mounted underneath, which shoots out a beam that scans from left to right at high frequency.  This unit also has a receiver, which collects any of the laser light that is reflected back to the airplane.

A highly precise clock measures the time it takes for the pulses of light to return to the airplane.  From high school physics, distance = velocity x time, so the travel time measurements allow the unit to determine how far away  whatever the laser beam bounced off of is from the plane (given the speed of light as the velocity).  Taking that a step further, since the plane knows where it is from the GPS, on board computers can determine the location (x, y and z) of the points on the ground.  (Hope that all makes sense.)

So what we get out of a LiDAR system is a “cloud of points”, each with x, y and z (elevation) coordinates.  And to boot, these are typically at very high resolutions (around 0.5 m between points).  At this scale, we can resolve trees, buildings, even cars as well as smaller objects.  From these points we can then derive high resolution elevation models, on the order of 1 m grid cells.

In the wake of the Christchurch earthquakes, several LiDAR missions were flown over the city.  These data allowed us to know, for example, that parts of the Avon-Heathcote estuary have risen by 500 mm while others have sunk by a similar amount.  LiDAR is quite simply revolutionising the way we capture elevation data, very much for the better.  It’s not cheap, but high-resolution elevation data can be captured over large areas in short periods of time.

For our advanced GIS courses (ERST310 and 607) last year, the students were working with LiDAR data and it got me thinking…could I use LiDAR to find the source of my tunnel gully?  Nah – the chances of actually finding something have to be slim to nil, right?  But I had to have a go anyway (a sad commentary on what GIS people do for fun…).

So, with too much time on my hands one day, I started by looking at some LiDAR data from September of 2011.  With a bit of pre-processing, and using the LAS Dataset toolbar,

,

I produced this high-resolution terrain model of the area around where I live, shaded to show the changes in elevation:

There’s some exquisite detail to these data.  In the image below I’ve zoomed in to Corsair Bay – you can easily make out roads, individual trees, houses, walking tracks.  Look closely and you’ll see that when the data were collected, the tide was out:

And here’s a selection of the points that fed into the elevation surface:

(No points over the water because the laser light is absorbed by water rather than reflected.)  So now let’s zoom in to where my problem is (not to suggest that I’ve only got one problem…) – here’s a satellite image for a start:

You can see three houses – mine is in the lower left hand corner; elevation increases from left to right.  You can also see a driveway in the image centre going uphill to the north, then cutting hard right and heading southeast.  The three red dots are related to our tunnel gully – the leftmost one is where the gully surfaces (this is where the first photo above was taken) and the other two are places where I know water can get in to the tunnel gully.  So next, using some elevation contours, I drew a few lines – one that’s just a straight line connecting the three points and another to show the rough fall line (i.e the line along which water would flow following gravity – starting at the house and moving uphill perpendicular to the elevation contours (the thin blue lines are 5 vertical metres apart, and 25 vertical m between the bright blue ones):

Next I turned on the LiDAR surface and just looked for anything interesting.  I’ve set this to show the “bare earth” surface, i.e. just those points that have come from ground level.  This allows me to filter out the buildings and trees at the touch of a button:

One thing that stood out to me immediately was that slight depression just along the fall line and between the driveway sections.  There’s was another, slightly more subtle linear feature oriented roughly east-west that crossed the fall line – both are enclosed by the purple areas below:

Hmmm…interesting…  Worth a look?  Why not.  Here’s what I found.

At the place where my LiDAR data showed the depression, there was in fact a significant depression and, much to my surprise, a bloody great hole!  (This is on a neighbour’s section so I couldn’t uncover it extensively, but the hole goes down almost two metres!)  The linear feature turned out to be an old fence line which you can partially see from the satellite image above.

To be fair, this hole probably is not the actual inlet.  In all likelihood, the tunnel gully continues further uphill and probably doesn’t have a single identifiable inlet.  These channels collect water infiltrating down through the soil, so it’s really taking in water along its whole length.

While I’m quite surprise to have found anything, this hasn’t really helped solve my problem.  Even if I were to block this hole, water would still get into the channel and flow out beneath our house.  But, the great thing is that the level of detail the data showed was enough to find this hole.  I won’t be publishing this in a paper anytime soon, but it did prove to be an interesting exercise, right in my own back yard.  All thanks to LiDAR.

C