Drones on Goonderoo

on 04 Jul 2016 

I recently spent a few days with volunteers Rosemary Rogers and Geoff Spanner who spent a month on Goonderoo working on weeds and infrastructure. The place, especially around the homestead, is looking way better for it. Geoff is also a photographer and videographer and carries a camera drone (the DJI Phantom 3 to be exact).

This was timely as I was in the process of a 10-year review of the property and many of us are re-looking carefully at how to demonstrate changes in the reserves over time. I made the most of Geoff’s generosity and technical skill to consider the potential of drones in our tool kit.

The terrain at Goonderoo is low and rolling, and the vegetation often dense. It’s hard to get a scenic view or an overall impression of how things vary with the terrain and over time. The observer can kneel, stand or climb onto a ute or low hill. Above that it’s helicopters, planes, and satellites. When you first watch a drone lift from zero to 500m with a good wide angle lens, it’s hard not to run around getting that bird’s eye view. 

Here are my thoughts on the model Geoff had (an investment of around $2,500 and no doubt many hours learning to fly it).

Overlap with aerial imagery - recovery of tree and shrub cover

Ten years ago I flew over Goonderoo in a small plane, sadly not on a good day. In some areas the satellite images would be better, although not available for Goonderoo at the time. 

In less time than it took to drive to the airport, Geoff and the drone gave far better images. In evaluating the regeneration of Brigalow the key variables are crown cover and the ground cover. The images would allow robust sampling of canopy cover over perhaps 20 hectares, with the ability to repeat over many locations quickly. 

With ground markers and geo-rectification of the images, the fate of hundreds of individual trees could be monitored, hugely increasing the power to detect trends. The images attached from a couple of hundred metres up show the comparative regeneration on Goonderoo. Vertical images, now possible with the drone, will make for easier quantitative comparisons.

Extra-sensitive aerial imagery – past land use

Figure 5 shows the Brigalow regeneration with the tell-tale ripping lines beneath. I’ve never noticed this before - the tight light angle shows up the micro-relief. I'd seen the density of the regeneration change across the fence-line in Indians paddock and thought this was due to the effect of grazing on fire. Now differences across the fence in past land management are just as likely the cause - even at least 20 years on. Fancier cameras on drones, of course, can get imagery in wavelengths to look at other variables.

Weed mapping

I recently flew in a R44 chopper over the Mulligan River on Cravens Peak Reserve to map out major Buffel Grass infestations. I’m yet to do the ground-truthing of this flight but if the tall grass in Figure 7 is Buffel this low imagery would allow close scrutiny of changes in density, again possibly monitoring many individual plants. A drone would allow cheaper photographs at the same time as the ground truthing and the measure of weed density could be done with more robust samples for indicating the nature and degree of change.

Fire and assemblage change in grasslands

Looking at the impact of fire regimes on vegetation is at the mercy of unpredictable fire behaviour at the finer scale – you can’t put in sites and prescribe the future burn regime for them.  We’ll never have the luxury of sampling for all contingencies of fire mosaics in advance.  Figure 8 has had the colours doctored only slightly to show the variety and patchiness in the grass sward at Goonderoo. Low-level imagery used along with fire mapping and some ground-truthing may give us the sampling scale and precision needed to study or at least illustrate the benefit of different fire regimes on grassland. 

Fire and changes in woodland structure

Paul Williams and I put in 1km-long traverses on Yourka Reserve in a laborious attempt to sample the variation in woodland characteristics arising from successive patchy burns. The method quickly illustrated to us that at Yourka, in low-rolling terrain at least, you couldn't figure out vegetation patterns and responses from one hectare sites very far at all. To make the traverses quicker we did eyeball estimates of stem counts from the centre of the imaginary plot. Like many hands-off methods we could see how it could vary with the observer to a degree that would compromise any results over time. 

The drones can be programmed to fly to GPS points up to several hundred metres. A few short flights along the traverse (or whole traverses with more expensive gear) would get much better estimates of structural change more quickly – allowing more time for more botanical ground-truthing.

Site photos of vegetation structure

At Goonderoo, and elsewhere, we're looking for dense shrubby vegetation for decent woodland bird habitat.  I've often wanted an extension ladder to get photographs of that structure, wishing I could lift the camera about 4 meters of the ground, then 40, or 400- vertical or oblique. Careful flying though!

Vegetation mapping

The variety of geology, land use and terrain at Goonderoo and the implications for weed management in particular require much more detailed vegetation map that we have. Had I thought of this while up there I could soon have had the images in important areas to make that happen quickly. Figure 9, taken from about 500m in low afternoon light shows how it brings up the patterning in the vegetation.  

Pictures that tell and sell

A photo on an angle in landscape taken from about twice the height of the largest trees makes for a great lookout as you can see the small scale and the large, and maybe the context.  You can see variety in the landscape at a range of scales. Figures 10 to 11 give a couple of examples.

Overall

Clearly there are a range of technologies that greatly expand the ecologist's toolkit. For me this chance to explore drones was quite an eye-opener. They have the potential to help us compare plots across variable terrain, reasonably quickly. Their use would impact mapping, planning and review of our conservation management plans. 

I’m interested in the thoughts of others.