Aerial photography proves an effective method for searching for Malleefowl mounds

We searched an area of 70km² in the mid-west using high definition aerial photography for Malleefowl mounds. This area was searched on the ground in 2008, and all more recently discovered Malleefowl mounds had been added to the database that we used in this assessment.

The aerial survey captured the area using a Microsoft Ultracam D large format camera mounted in a Cessna 421. The ground sample distance (GSD) was 4cm, or put another way, each pixel on the computer screen represents 4cm on the ground. This very high definition photography was post-processed to produce images suitable for searching on the computer.

We initially assessed numerous approaches to viewing the aerial photography. We then systematically searched a few small areas using a variety of approaches to determine which method produced the most accurate result. Accuracy was determined by turning off the layer of known mounds, searching the area and comparing the number of mounds that were recorded (and missed) with known mounds based on on-ground grid searches.

We then settled on a protocol that provided the most accurate result. We opted for a conservative approach, as we considered it was better to include false negatives, which could be eliminated by ground-truthing, rather than discard possible mounds.

Findings:

• Very old, weathered and long-disused mounds can be easily missed;
• Recently used mounds that have the typical shape are easily detected;
• 3D image searching, compared to 2D, provides a higher level of accuracy;
• Searching black and white images records most of the Malleefowl mounds, however, it also records a higher number of false positives and fails to record mounds that are under dense vegetation. False positives were mostly the shadows from a tree with a circular canopy falling on an open section of substrate. It was also difficult to separate foliage and tree vegetation above a mound from the mound profile; and
• Searching coloured 3D images provides the most accurate record of mounds.

Our search areas included eucalypt woodland with a sparse understorey on red sandy-clay substrate, low shrubs on a banded ironstone ridge and dense shrubs and low trees on a sand plain. We would now like to assess this methodology in other habitat types, in particular in dense heath.

Our preliminary look at the costs and outcomes indicate that this method is as effective in locating recently used mounds and inactive mounds with some structure as conducting on-ground grid searches. Aerial photography searches with follow up ground-truthing are likely to be cheaper than grid-searching an entire area, particularly if the area has limited access and supports dense shrubs about 1.5m.

The size of the area, vegetation density and abundance of mounds are all factors that influence the cost-effectiveness considerations. This approach becomes more cost-effective if the aerial photograph is produced for another purpose and is then used to search for Malleefowl mounds. Discounting the cost of the aerial photography because of other uses is where the real cost saving can be made.

Images below are of two different mounds in 2D.  Images in 3D are not possible without special computer screens.