The SoilCares Soil
Testing Concept

AgroCares has developed two tools, the Scanner and the Lab-in-a-box, that bring the knowledge of our Soil analysts and agronomists into the hands of the farmer in a quick, easy and affordable way. But how does it work? From the moment the soil is scanned with either the Lab-in-a-box or the Scanner, the spectrometry data produced by the spectrometers inside our technologies is started on a journey through several processes and checks until it is returned to the user as a soil status report.

It all starts with a scan from the sensors of the AgroCares Lab-in-a-Box or Scanner, that produces a spectral image. From these spectra, several regression models produce the numerical predictions that are returned to the client as a soil status. Indeed, the real intelligence of our solutions lies in the database and its algorithms.

It is really by creating and training machine learning regression models that Agrocares has made it possible to predict the content of a soil sample from a spectrum.

These regression models are developed country by country by our team of experts. Our agronomists first determine the number and location of samples required to cover the full spectral range of a country using data like soil type, land use, fertiliser and crop residue management, satellite crop development images, climate and elevation.

Sampling team in the Philippinnes

These samples are collected following very strict procedures and shipped to our Golden Standard Laboratory in the Netherlands where they are analysed using regulated traditional wet chemistry techniques and scanned with the sensors of the Lab-in-a-Box (Mid Infra Red and XRF) and the Scanner (Near Infra Red).

Analysis of a Sample by the MID Infrared Spectrometer of the Lab-in-a-box in the Golden Standard Laboratory.

Machine learning is, in essence, the process of applying algorithms to identify patterns in the data that correspond with the ground truth of that data. In our case, the ground truth is the reference values obtained in the Golden Standard Laboratory, and the patterns are the spectra for each sample obtained from the Scanner and the LiaB.

The regression model calculates a function to transform a spectrum into each of its reference values. For example, the presence of a significant peak in the spectrum could correspond with a high Potassium concentration.

The farmer then receives a full soil management report that includes soil analysis results in classes for N, P, K, pH and organic matter with the Scanner and in values for all Macro and Micro nutrients with the LiaB. But also, tailor-made , locally adapted fertiliser recommendations for a better yield