Gordon Morris, Remote Sensing Scientist
Over the past few months Lynker Analytics together with our partners Action Aviation and UAV Mapping NZ have flown the entire length of Aotearoa New Zealand capturing aerial photography for the Ministry for the Environment.
The goal of this project is to survey and classify around 7,000 distinct areas of possible deforestation for climate change reporting. This project is similar to a survey we completed in 2020 for that reporting cycle.
The imagery collected from a Cessna 172 aircraft is destined for our machine learning model that has been pre-trained to classify land cover and vegetation types common in these areas including plantation seedlings, native forest and regenerating exotic forest.
Lake Coleridge, Canterbury
Our Wellington team support the air team and manage the data logistics using several management tools that provide stakeholders with confidence on the capture and safety of our crew.
To manage the large distances we first organise the image collection locations into a systematic and fuel-efficient order of capture. Once daily areas have been planned, a “classic traveller” least-distance analysis is run. This feeds into the on-board flight planner, while still allowing the pilot options to get into the best approach for each target.
Based from Rangiora, our team captured much of Canterbury and Christchurch. Flight path (blue crosses), Forest targets (orange boxes)
We track the progress of our planes in the air with connection to an API from FlightAware through ArcGIS Online. The dashboard also allows us to track the project as a whole, and update stakeholders with current events. Editing is done to a minimum, with the dashboard taking live feeds from spatial data.
Short messages from the flight crew’s daily reporting are added which can highlight areas of concern, such as weather conditions and controlled CAA airspace. Meanwhile, the ground crew send back imagery regularly for processing by hard drive, given the large size and volume of data. The data is processed and dispatched to our deep learning training system for model fine tuning and eventual classification of each target area.
Flight Operations dashboard
In the first 3 months of 2022 persistent clouds and unfavourable weather conditions have been present throughout. However, our crew are now turning north again to complete the homeward leg. This work is scheduled to complete by the end of June.