8. Precision fertilization in field crops using sensing technologiesAUA in collaboration with Augmenta (CNH)

Short Description of the technology

The DJI Phantom P4 Multispectral is the drone used by AUA to assess the current vegetation status of crop fields. It is equipped with an RGB sensor for visible light imaging and five monochrome sensors for multispectral imaging, each with a 2.08 MP resolution, covering the blue, green, red, red edge, and near-infrared bands. The drone captures multispectral images in TIFF format, which can be processed using photogrammetry software such as Pix4Dmapper. This software applies photogrammetry algorithms to convert the images into vegetation maps such as NDVI. By analyzing the data from these vegetation maps, we can determine the variable fertilizer requirements of the fields, enabling variable rate fertilizer application (VRA). The main goal of this approach is to optimize fertilizer use, reduce input costs, and enhance agricultural efficiency. In addition, Augmenta (CNH) has developed an innovative technology named Augmenta® Field Analyzer which is a tractor-mounted multispectral camera system that collects Red/Infrared high-resolution data at 16 fps over an up to 42-meter (depending on mounting height) field-of-view to compute an NDVI-based Vegetation Index (VI), enabling immediate, site-specific fertilization based on proprietary agronomic algorithms that are self-calibrated. The technology leverages remote sensing and advanced data processing to optimize Real-Time fertilizer application based on field variability, thereby improving nutrient management, crop yields, and environmental sustainability.

Innovative Features

AUA utilizes commercial drones, such as the DJI Phantom P4 Multispectral, to assess the fertilizer input needs of crop fields. In collaboration with Augmenta (CNH), it implements an innovative experimental design for precision fertilization. On the other hand, Augmenta (CNH) provides an innovative technology equipped with hardware and software designed to optimise Real-Time fertilizer application.

On a hardware level the Augmenta® Field Analyzer is a proximal sensing system that features:

  • Multiple spectrum 4K cameras (Red, Infrared, RedEdge, RGB stereo) providing resolution of 12 pixel/cm
  • Embedded NVidia GPU
  • Environmental sensors (solar radiation, sun elevation, cloud coverage) to and
  • Internal GPS sensor.

On a software level, advanced multispectral Computer Vision and Machine Learning (CVML) techniques enhance the performance of the system’s agronomic algorithms. The agronomic algorithm requires a minimum input from the end-user (the maximum rate to be applied) and are autonomously adjusted to the conditions encountered during each session, with no further need for farmer’s actions.

Embedded sensors coupled with ML mitigate the impact the adverse impact of changes in light intensity (due to presence of clouds, sun angle, etc.) on data gathered, recognise patterns (glares on plant canopy) and exclude them from further processing, and ensures consistent performance of the system at a fixed rate when unfavorable visibility conditions (fog, dusk, etc.) are detected.

It is a plug and play technology that seamlessly connects with various agricultural machinery using ISOBUS and LH5000 protocols while storing data in the cloud for real-time analysis.

Whenever necessary, Augmenta® Field Analyzer work in tandem with a DJI Phantom P4-multispectral UAV working in strong agreement in the vegetation index observed validating their consistency for improved crop management. In addition, remote sensing data (NDVI values) gathered with the use of the UAV are being used as input to the agronomic algorithm in order to generate offline prescription maps.

In conclusion, both real time and prescribed variable rate application can be evaluated and compared with the conventional fixed rate applications.

 

Type of Contribution

VRA technology contributes significantly to reducing the environmental impact of nitrogen leaching. By dynamically adjusting fertilizer applications to match field variability, it has demonstrated input savings of 4.4% to 6.3% without compromising crop growth and even showing a tendency toward increased productivity. This reduction in nitrogen inputs can help mitigate nitrate contamination in groundwater and surface water, as well as decrease harmful ammonia emissions, making farming practices more sustainable and environmentally friendly.

 

Benefits for farmers, industry, and the environment

Data analysis suggests that this technology could contribute to more sustainable crop production, as a reduction in in-season nitrogen (N) fertilizer inputs was achieved during the production process without compromising productivity, resulting in lower environmental impact and reduced production costs for farmers. It also offers multiple benefits, including cost savings from reduced nitrogen fertilization; these savings, when factored into overall production costs, can serve as a key indicator in a farmer’s decision to adopt the technology. Moreover, the observed tendency for increased productivity acts as a complementary indicator, with higher yields directly translating into increased income. The benefits of this technology are generally proportional to farm size and highly dependent on crop type, meaning that both cost reduction and income increase should be evaluated within this context to estimate the respective return on investment (ROI). Furthermore, by utilizing different types of variable rate application (VRA) operations, farmers can gain valuable insights into their farms, enabling informed decisions about future cultivation practices, for example, using nitrogen VRA to identify low-productivity areas. Beyond the economic advantages, this technology also brings socio-environmental benefits; reduced nitrogen fertilizer inputs lower the risk of farmers being exposed to harmful substances, thereby contributing to a healthier working and living environment, while fewer refills translate into shorter working hours for operators. However, these benefits are not always easily quantifiable, which can make it challenging for end users to fully recognize the added value of adopting this technology.