Research questions accuracy of drone data in agriculture

Research questions accuracy of drone data in agriculture

Farmers keen to adopt drone technology to improve crop health should be wary of ‘off the shelf’ aerial products, according to researchers at the University of Queensland and the University of New England, in Australia.

The warning comes after new research which appears to show that aerial imaging kits – hardware and software platforms sold to farmers for up to $4,000 apiece – are “unlikely to provide accurate information about the health of crops.” The research was conducted by Yu-Hsuan Tu from the Joint Remote Sensing Research Program at UQ’s Remote Sensing Research Centre.

Multi-spectral concerns

Drones have seen widespread adoption in the agriculture sector because they allow farmers to monitor the progress of crops quickly and effectively from the air, sometimes linked to robotic tractors which can irrigate or fertilise fields in areas identified by the drones.

More often than not this process involves multi-spectral imaging, which measures reflected energy coming off of crops within specific sections of the electromagnetic spectrum.

But the University of Queensland has urged caution, suggesting that the results may not be as reliable as first thought, depending on how the drone’s data is processed.

“Our research has shown that the high-tech, multi-spectral sensors used to collect images from the drone must be processed in a certain way to obtain correct information for horticultural farming applications,” said Yu-Hsuan Tu.

According to the study, inconsistencies in a drone’s altitude and even the angle of the sun could result in farmers receiving the wrong results. The obvious solution is to ensure that aerial imaging is carried out by experienced professionals and processed the right way. But even that might not be enough to guarantee reliability and accuracy.

Distorted comparisons over time

As well as providing on-the-spot health checks for crops, drones fitted with multi-spectral sensors can help a farmer to build a picture of operations over time. However, the slightest inconsistencies could render this comparative process unreliable, suggests the new study.

“To compare reflectance across tree canopies, we flew a drone at different heights and angles to the sun above an avocado orchard, and found huge variations in the images collected,” said Yu-Hsuan Tu.

“This means that while differences in tree health might be determined from tree to tree in imagery from one flyover, subsequent flyovers will show the vegetation very differently unless the drone is exactly the same height and angle to the sun – something almost impossible to achieve,” he continued.

“In practice, this inconsistent imagery makes it impossible for farmers to compare vegetation conditions over time because the light variations are not corrected for height or angle and the product will be distorted.

“This could lead them to draw conclusions about the decline or improvement of their crops that may be incorrect – assumptions that could prove expensive to farmers already under pressure from recent weather events.”

Farmers should take care when selecting equipment and hiring drone operators, says Yu-Hsuan Tu – at least until more research has been done on the topic.

“While the science behind our work is complex, the message to farmers is simple: use drone technology at your own risk. More work needs to be done before drone technology can accurately determine the health of crops,” he said.

Internet of Business says

The research paper has not been published yet. We’ve reached out to Yu-Hsuan Tu for clarification on several points and will update this article in due course.