Precision farming – digitalisation in agriculture

Precision farming is an approach that is now over 30 years old and aims to support and optimise agriculture with the help of digital technologies. Central technologies are GNSS (satellite positioning), GIS (geographical information systems), data transmission technologies such as CAN (Controller Area Networks) and the use of sensor systems. Earth observation satellites are an important sensor system. The use of GNSS, which is widely used for the automatic steering of tractors and harvesters and the automatic switching on and off of machines (section control), is currently of outstanding importance.

Earth observation is an excellent basis for monitoring the growth of plants in agriculture, but not only there. The special feature is that earth observation in the form of satellite images is mostly image-based and therefore captures spatial differences. Site-specific land management is one of the most important sub-disciplines of precision farming: it aims to adapt the management of sub-areas of a field in terms of sowing, fertilisation and crop protection. The aim is to save costs on the one hand and to minimise the negative effects of fertilisation and crop protection on the environment (soil, water, groundwater) on the other. Satellite images provide an excellent basis for this because they capture the small-scale differences in crop growth.

The sensors (cameras) installed in satellites have the ability to detect non-visible light beyond red (near-infrared). This range of light is particularly sensitive to differences in vegetation (chlorophyll content, biomass) – another advantage of satellite-based earth observation in agriculture.

The sensors detect the reflection of sunlight in different wavelength ranges. However, these can also be generated or altered by clouds. This requires the application of appropriate correction values or means that the data cannot be used at all. Moreover, even in the most unfavourable case, this back radiation (reflection) does not allow any direct conclusions to be drawn about plant properties, but must be transformed using models. The models are crop-specific, usually require the analysis of time series (several consecutive images) and are only valid to a limited extent for larger areas. In recent years, many new models have been developed, most of which are based on AI. It is to be expected that the models will continue to improve in quality and become more transferable with improved AI methods and a growing data basis.

Comparability with analogue phenomena

Over the past millennia, agriculture has adapted to both large and small-scale climatic conditions and, above all, to soil differences. This adaptation was based on traditional knowledge and the farmers’ own experience of cultivating the land, often from childhood onwards. They observed the development of different crops in the various areas of agricultural land and related this to the weather. At that time, their own eyes and the aiming of clearly visible objects in the surroundings (trees, church towers) replaced satellite navigation. They also served as sensors (plant height, number of ears of corn, discolouration of leaves, weed cover). This form of knowledge- or experience-based agriculture is no longer sufficient and is in decline due to the ever-increasing areas and farms and, above all, due to structural change. The knowledge is not available for purchased or leased land. The same applies to contractors and temporary drivers who carry out work on land on behalf of farmers. Data-based or digital land management is therefore becoming increasingly important.

Social relevance

The optimisation of agricultural production ensures the production of high-quality food and contributes to food security. At the same time, optimisation can reduce production costs and minimise negative effects on the environment. Earth observation can make a significant contribution to this by revealing small-scale differences in plant growth and thus creating opportunities for customised measures.

---