Creating an accurate 3-D image of the provincial road through Verloren Valei is the first step to rehabilitating it, according to Andre Beetge, Working for Wetlands .
Over the course of the years, the old provincial road that links the R540 to Mashishing with the R577 to Roossenekal has been graded so frequently that it has lost its design integrity. As a result, runoff from the road brings large amounts of silt and rocks into the large wetlands in the eastern portion of the reserve. Efforts to restore these wetlands have been ongoing since 2004, but it is clear that the root cause of the problem has to be addressed.
In order to do so, it is first necessary to survey the road and the surrounding area so that future restoration plans would include both the road and the wetland. The project is being undertaken by Working for Wetlands, in collaboration with the MTPA.
Tights budgets meant that traditional surveying by human surveyors was out of the question, so the decision was taken to use a drone. The project was also something of a test case to see whether drones could be used routinely for this kind of surveying.
The drone used was a DJI Phantom 4 Pro, piloted by the MTPA’s Dr Mervyn Lötter. This quad-copter drone is valued at around R30 000, and has a flying time of approximately 30 minutes on a single battery. The resolution of the imagery captured and the products produced ultimately depend on the sensor of the drone camera and the height flown during the survey: the higher the flight, the lower the resolution. The DJI Phantom 4 Pro has a 1-inch 20-megapixel sensor which captures images of a very high quality, and each image also includes the GPS coordinates and altitude. This information is then processed using advanced software such as Drone2Map.
The area to be surveyed was roughly 3 kilometres by 400 metres, roughly 109 hectares. It was broken up into three areas. The drone flew in a grid flight pattern at an altitude of 100 metres, planned using the free Pix4Dcapture app. Each flight took 20 minutes, and the entire operation took two hours to complete.
Thirteen ground control points (GCPs) were surveyed prior to the flight using the ArcGIS Collector field app with 30 second GPS averaging enabled. GCP altitude readings were adjusted after the survey using Airbus 24 DEM imagery. The purpose of using the GCPs was to try and ensure that the horizontal and vertical accuracy of the imagery was as accurate as possible without the use of survey-grade techniques.
A total of 1173 images were captured for processing. They were imported into Drone2map for processing (Fig. 2). The GCPs were also added and linked to the imagery and the image processing took just over 14 hours to run on powerful computer with a good graphics card.
Combining the two data sets allowed the creation of the digital surface model as shown clearly showing the road and other salient features such as rocky outcrops, culverts and so on.
The completed images have now been supplied to the engineering firm that is helping to develop the rehabilitation plan. It will then become clear just how useful drones will prove to be as a way to survey small natural areas to establish existing biodiversity and/ or infrastructure. But the signs are promising.