Choosing the correct satellite image important when calculating soil and clay content

Satellites can collect large amounts of data and do not require major investments, compared to conventional soil sampling and mapping. The technology is therefore an interesting option for farmers who want to make their operations more efficient, reduce costs and increase yields. However, for satellites to be used to calculate, for example, soil carbon content, the right conditions are needed - and they can be difficult to achieve.

“For the calculations to be accurate, the top layer of soil should preferably be dry and free of vegetation. In addition, the atmosphere needs to be cloud-free. We don't often have these conditions", says Johanna Wetterlind.

Partially very accurate calculations

In a new study, Johanna Wetterlind and her European colleagues have tried to address these problems. Using a new selection method, images of dry land areas were selected from time series of satellite data from farms in Italy, Turkey, Lithuania and the USA. This is a new way of selecting and interpreting satellite data to obtain accurate soil content estimates. Most commonly, satellite images from only one date are used.

"It was fascinating that some of the results were so good. At best, the models could explain over 90 percent of the variation in topsoil and clay content, but in some places the explanation was only 50 percent, and sometimes it did not work at all.

The size of the errors in the models was on average about 0.35 percentage points of topsoil and from a few to seven percentage points of clay.

"It is a step in the right direction to be able to estimate the soil carbon content in a cheap and efficient way. You can also automate the process, which would further streamline the calculations", says Johanna Wetterlind.

Conventional sampling still important

An important reason why the results are so good in this study is that satellite models have been made for each individual field using many conventional soil samples.

Conventional soil sampling is necessary to determine soil content and clay content using satellite data. The samples serve as comparison points, known in scientific terms as ground truthing, and are used to build the computational models.

"The difficult part is to get clear relationships between satellite data and soil sample data, which is needed to calculate the soil carbon content, but we have now shown that it is possible. However, the correlations we have managed to obtain are very local and cannot be generalized to other locations, which is a limitation of this method", says Johanna Wetterlind.

Swedish soils next in line

The method is currently being tested on Swedish soils in the region of Skaraborg, in another project that Johanna Wetterlind is involved in. The researchers are investigating how the results differ from those in other countries. This will provide further knowledge of the method's potential in different types of climate, soils and applications.

"It is, after all, an interesting method that could complement traditional soil mapping and improve the decision-making basis for cultivation measures in agriculture", says Johanna Wetterlind.

Read the study: Assessing the capability of Sentinel-2 time-series to estimate soil organic carbon and clay content at local scale in croplands - ScienceDirect

The STEROPES project

The project looks at how satellite imagery can best be used to estimate soil carbon/clay content and what the limitations are. This includes finding good methods for filtering out fields where there is no bare soil and investigating how variations in soil type and water content affect the estimation of the amount of carbon. An important part of the project is to be able to study how well the technique works for different soils and under different climatic conditions in Europe, as it is conceivable that the impact of different factors varies depending on local conditions. The aim is to develop a method to make the best possible carbon estimates based on satellite data.

Read more about STEROPES: https://ejpsoil.eu/soil-research/steropes/