Why we are moving from satellite imagery to soil sensors

For many years, I have been intensively involved in evaluating and responding to spatial unevenness of soil conditions, which are most evident in crop yields. This unevenness is best documented within a plot through yield maps. However, a one-year yield record can be affected by the influence of the vintage, and therefore, during long-term data collection, the so-called long-term production potential of the plot is determined, which eliminates the influence of the vintage. Unfortunately, for many reasons, most farms do not have historical yield maps, and therefore satellite images are used in the vast majority of cases to determine the production potential of the plot.

I have long considered satellite images to be the most reliable information on spatial unevenness of land. Unfortunately, in recent years, in connection with anti-erosion strategies, changes in land boundaries (division into 10 and 30 ha) and the advent of regenerative agriculture, we have begun to encounter new limits of remote sensing. Fig. 1 Complex creation of production potential with frequent changes in land boundaries.

The production potential is based on the analysis of a multi-year series of satellite data and is calculated as a percentage of the achieved productivity compared to the average value of the entire plot. Therefore, it is also necessary to calculate with the yield level of the entire plot (t/ha). In our SAS Cropwise Operations platform , the customer has the opportunity to select any satellite images with the NDVI index for calculating the production potential in unlimited quantities. For example, some customers have compiled different production potentials for winter and spring crops.

Fig. 2 Satellite image of the agricultural company's lands in June 2024.

Nevertheless, in the vast majority of cases, we compile production potentials within the framework of SAS Cropwise Operations customer support . The main reason is that the influence of the previous crop is very difficult to detect at changed land boundaries (see Fig. 2 and 3). Last but not least, we compare vegetation indices with the development of vegetation and look for deviations. Frequent deviations include, for example, waterlogging of part of the land or focal weedy crops. In addition, we use the vegetation EVI index , which is less known to farmers, with which we have better experience and results.

Fig. 3 Satellite image of the agricultural company's lands in June 2023.

Creating a digital landscape

Changing borders not only brings problems with the creation of production potentials, but also the problem of maintaining a data trail of work operations - mechanization works at the land level and cannot distinguish between parts of land blocks and agricultural plots. At the same time, " internal" headlands are created and unevenness increases due to agricultural technology. These factors led us to on-the-go sensors.

What are on-the-go sensors?

On-the-go sensors are mounted on machinery or vehicles and indirectly measure soil properties as they move across the land. Although there are a number of sensor concepts (more on these later), the most commonly used are electrical and electromagnetic sensors. The term soil electrical conductivity has been coined for this type of measurement. It is a simple and relatively inexpensive tool for measuring soil properties. The measurement is based on the ability of a substance to conduct an electric current and is expressed in units of millisiemens (mS/m).

Soil conductivity serves as an indicator of the physical and chemical properties of the soil, correlating with soil texture (content of clay, sand, loam), soil moisture, soil compaction, organic matter or soil profile depth. Unlike production potential, it does not provide information about the production potential of the land, but about spatial variability, which must be defined.

Fig. 4 I have been working with on-the-go sensors in many projects since 2020.

After approximately 5 years of getting to know on-the-go sensors, I decided to advance our services and add the Austrian Topsoil Mapper sensor (Geoprospectors), which will help us better characterize the spatial variability of our customers' lands.

What is Topsoil Mapper?

Topsoil Mapper is an active soil sensor that works on the principle of measuring soil electrical conductivity (EC). The device is usually attached to the front of a tractor or, in our case, an ATV. The sensor continuously records data from several depth levels while driving. The result is detailed maps of soil conductivity that correlate with soil structure, clay content, moisture and other physical properties.

Fig. 5 Topsoil Mapper can work effectively throughout the year in no-till technologies.

The great advantage of the Topsoil Mapper sensor is its ability to work in almost any weather and is not affected by vegetation, which we have verified when measuring soil properties on farms using catch crops. On some plots with higher dead crops, this sensor with yield maps is the only chance to determine spatial unevenness of the plot.

Fig. 6. Production potential compiled from a ten-year time series of satellite data.

We have presented the sensor to about 15 of our customers so far and we are very satisfied with the results. At the same time, they do not only give us a map, but in combination with yield maps or production potential they raise a number of questions about the relationship between yield and soil conductivity.

Fig. 7 Visualization of the adjusted yield map from the 2023/2024 season.

In addition to the soil conductivity map at four depths, we are also interested in the soil water content and compaction layers. Unfortunately, I must say that our expectations were not met and on some plots, due to the high load on the protective strips (slurry application, etc.), a different response from the rest of the plot is visible.

Fig. 8 The soil conductivity map almost copies the yield map of the plot.

While satellite imagery remains a valuable tool in remote sensing and vegetation monitoring, Topsoil Mapper offers deeper and more detailed information about the spatial unevenness of the land and I firmly believe it will help us move our work forward.

Interpolation and working with acquired data

Have you ever wondered where the production zone of a plot ends and begins? We have, and that is why, after long discussions, we decided to process the data with your participation and ask you whether it corresponds to your ideas and knowledge about the plot. We will then upload this data in the form of map data to SAS Cropwise Operations , where you can use it for further work - such as creating prescription maps for variable applications or determining a sampling network for soil sampling, which you can read more about here.

We would like to get closer to the plot and understand the connection between yield and soil properties, so we will supplement the sensor with a penetrometer and a soil sampler during the month.

What do you think about the sensor and the approach? Are you interested in our work and the sensor, or are you interested in trying it out? Write to me .