Drones aid in the fight against grass lodging
By measuring the height of the grass in the field, drones can help identify areas in seed grass fields that are often affected by grass lodging. The method allows farmers to counteract the lodging in time and thus avoid yield losses.
Grass is not just something we have in our gardens, on golf courses or in parks. Grass is an important resource for feed in agriculture, just as it e.g. can be used for green biorefining and even for building materials, clothing fibers and maybe sometime in the future also as food on our dinner tables. Grass also has the advantage that it is a perennial crop, and thus helps to reduce agricultural emissions of greenhouse gases and nutrients as well as build up the soil's carbon content.
Worldwide, grass is grown on approx. 72 million hectares of land, and within the EU alone, grass seeds are grown annually for 23 billion. €, and Denmark has a large part of this production. But growing seed grass does not necessarily mean high seed yields. Many farmers suffer from early and heavy grass lodging, especially if they do not have control over their growth regulation and nitrogen supply.
"Grass lodging is when the seed stalks bend. Early and strong grass lodging can result in large yield losses," explains Associate Professor René Gislum from the Department of Agroecology at Aarhus University.
Often seen in the same place
When the grass lodging happens early in the season, it can affect the pollination that will heavily affect the yields. To avoid early and heavy grass lodging, growth regulation is used and the farmer must have complete control over both the timing and the amount of growth regulation in order to achieve the highest seed yields.
“You can also refrain from fertilising as much, because over-fertilisation often creates problems with grass lodging. But if you fertilise less, the yield will also be less, so this may not be the right solution either,” says René Gislum, who has been involved in a project where drones have been used to identify areas with problematic grass lodging.
Grass lodging often appears in the same places in the field year after year, which is why the researchers behind the project got the idea that they can use drones to find out exactly where these areas are.
"But of course, it required us to make the measurements with our drones several times. In this project, we measured grass lodging up to 5 years in a row, so we could collect data and find the plots that were most exposed, ”he explains.
As soon as the farmer has a picture of where in his fields the crop is affected by early grass lodging every year, he can implement mitigation strategies.
“You often see grass lodging in places in the field where there has been applied too much fertiliser. There may be double fertiliser when the tractor turns at the end of the field or when it crosses another track. In such cases, the farmer will be able to make a different fertiliser strategy when he has a complete picture of the risk areas of the field,” explains René Gislum, who adds:
“However, it requires quite a high resolution to be able to designate these areas, so we have not could not use satellite photography. Their pixels are at least 10x10 m and that is way too big. So, you have to use some equipment that can get in closer, and since the measurement must take place at a time when we cannot to drive in the field, the choice naturally fell on drones.”
Measures the height of the grass
In fact, it was not so advanced to find the right algorithm for the model that should be able to figure out where in the field the grass lodging appears. Grass that is not bending is significantly higher than grass that lies down.
"We used the height as an indicator, and we looked at a lot of different field trials or plots across years, varieties and treatments to get as much variation in data as possible," explains René Gislum.
In practical terms, the height of the crop inside the plot was measured, and then the height outside the plot as well, the two subtracted gave the height of the crop inside the plot. Despite the simplicity of the method, it worked.
“We also tested another method, where we looked for specific properties in the images. That is, we studied the structure of each image and identified areas where the structure was different. It will be different depending on whether the grass stands up or lies down. But that method is quite difficult to handle in the image processing, it is not something you just reach in and do, ”explains René Gislum.
Three different levels
Both methods were used to assess and divide the plots into three different levels of grass lodging:
- No or little grass lodging (0-30%)
- Medium grass lodging (30-70%)
- Heavy grass lodging (70-100%)
Both measurement methods were able to identify areas where the level of grass lodging was either strong or almost non-existent, but they had more difficulty measuring the medium level. It also turned out that a combination of the two methods, neither improved accuracy nor precision, which is why the researchers recommend the height method, as this is the simplest to use.
“Our study has the potential for use in precision agriculture to generate maps of grass lodging and thereby increase seed grass yield at farm level. It should be noted that we used grass as a model crop and it cannot be applied directly to other crops. Different crops have different properties, just as the effects on yield may also vary. In the future, we will continue to develop our models to cover more years, so that we can make the model even more robust,” concludes René Gislum.
We strive to ensure that all our articles live up to the Danish universities' principles for good research communication (scroll down to find the English version on the web-site). Because of this the article will be supplemented with the following information:
|Study type:||Field experiment supplemented with image processing and development of algorithms|
Department of Agroecology at Aarhus University and South China Agricultural University
|Funding:||The field trials were funded by projects that had various purposes, including the use of drones for the determination of biomass and grass lodging. The field trials are funded by the Seed Tax Fund and GUDP, among others|
|Conflict of interest:||None|
|Extern commenting:||Only reviewers from the journal have had access to the article and its results|
|Read more:||The article "Assessment of grass lodging using texture and canopy height distribution feature derived from UAV visual-band images" is published in the journal Agricultural and Forest Meteorology. It is written by Suiyan Tan, Anders Krogh Mortensen, Xu Ma, Birte Boelt and René Gislum.|
|Contact:||Associate Professor René Gislum, Department of Agroecology, Aarhus University. Tel .: +45 20542092 or email: firstname.lastname@example.org|