Jun 14, 2020

Afforestation activity across New Zealand has increased since 2019, encouraged by regional initiatives, national climate change policies, and market demand. Grants targeting planting of erosion-prone land (class VII and above) are available through various channels, including regional councils.

A characteristic of any grants scheme is the requirement to show funded areas are planted. Often supporting evidence includes field inspections, photographs of planting and spraying activities. Collection of such information can be challenging if blocks are scattered, steep and have limited access points. In the worst case, failed areas may go unnoticed until the first mapping round.

Indufor has developed a staged approach that integrates repeat satellite observations and aerial capture, presenting a streamlined and structured approach to monitoring afforestation efforts.

Over each area, Indufor's algorithms analyse repeat passes of Sentinel 2 to track changes associated with land preparation and planting progress. The results provide evidence that pre-plant activities such as tracking, and post-plant release spraying have occurred and assist in focusing attention on areas of concern. The following example tracks change before and after planting. The activity is confirmed by the images (the image swipe) and marked decrease in the vegetation index post spraying.

Snow
Snow

Temporal change in Vegetation Index before planting (0.61), June 2019 and after release spraying (0.32), August 2019.

If further information is required, the area is flown using a Cessna-mounted camera. The system developed in partnership with Canterbury Aviation and SKYVUW allows the rapid and timely capture of aerial orthophotography with a spatial resolution of at least 5 cm. The pilot-operated system works in two modes - either triggering and capturing images at pre-determined waypoints, or set up to provide full area coverage. A recent trial tested the system and over two days, 4000 ha were covered, by contrast, tasked SkySat satellite imagery took two months to fully acquire the same area due to persistent cloud, missing the window of opportunity over most areas.

image unavailable - please update or try a different browser

Example of a sub 5 cm aerial photograph, this image is used to verify establishment success and used for planning and operational purposes.

The solution can be used to track the tree establishment process efficiently. The Sentinel 2 images capture roading activities and changes in grass condition before and after spraying, while the Skysat provides a snapshot in time and aerial imagery the highest level of information. Analysis and image layers are easily transferred to online dashboards or mobile devices and used to improve the efficiency of post-plant inspections.

Further collaboration with Scion is planned to trial advanced deep learning tree detection algorithms. Here is some footage of work from Scion, which shows results from their trials applying deep learning algorithms to UAV data.