I4S (Intelligence for Soil) – Integrated System for Site-Specific Soil Fertility Management

Project number: 031A564A
Contact: Dr. Sebastian Vogel, Leibniz-Institut für Agrartechnik und Bioökonomie e.V. (ATB)
Mail: svogel@atb-potsdam.de
Project team: Leibniz-Institut für Agrartechnik und Bioökonomie e.V. (ATB), Bundesanstalt für Materialforschung und -prüfung, Forschungsverbund Berlin e.V., Geophilus GmbH, Hahn-Schickard-Gesellschaft für angewandte Forschung e.V., Technische Universität München, Rheinische Friedrich-Wilhelms-Universität Bonn, Martin-Luther-Universität Halle, Universität Potsdam, Leibniz-Zentrum für Agrarlandschaftsforschung Müncheberg e.V.

Website: https://i4s.atb-potsdam.de/en/project
Duration: 09/01/2018 – 09/30/2021

Project aim

I4S is targeted towards the design of an integrated system for fertilizer recommendations and improvement of soil functions regarding nearly every square meter of soil. 

Motivation 
A detailed assessment of soil properties and a deeper understanding of soil processes are prerequisites for a site-specific and thus resource-saving and ecofriendly soil management.In agricultural practice, however, technologies for site-specific fertilization and tillage are still not widespread. The high demand on input data is one of the main reasons why the modeling and prediction of soil nutrient dynamics is not regularly used in practice. In particular, cost-efficient methods for detecting relevant soil properties such as soil type, nutrient, and humus content are lacking as well as easy-to-use decision support systems. 
Expected results 

The outcome of the project will include new soil sensors and new mobile sensor platforms for soil mapping. A set of soil and crop models will be developed specifically adapted to process the sensor readings. These models will form the core of a decision support system which will provide guidance on site-specific fertilization of nitrogen, phosphorus, potassium, and lime. 

 

Results from phase 1 & 2:

In the first and second phase of the project, several sensors systems are tested for mapping soil properties, in the laboratory as well as in the field. Furthermore, sensor modules are developed for field application. A mobile multi-sensor platform for topsoil mapping (RapidMapper) was developed and assembled with the aim of integrating all sensor modules in a single device.

Sensor-based soil maps were integrated into soil-process models to simulate soil water and nitrogen dynamics and predict crop growth and yield in higher spatial resolution. However, our model results point out the great importance of valid subsoil information as input variable.

Based on field experiments, nutrient contents and related yield data are recorded to compare a constant and a variable-rate fertilization strategy for lime, phosphorous and potassium. In the short term, results demonstrate higher cost of the variable-rate strategy. However, in the long term, constant-rate fertilization will generate lower yields, due to suboptimal nutrient supply.

A decision support system (DSS) for generating demand-oriented fertilizer recommendations for lime and macro nutrients using high-resolution sensor-based soil maps is currently under development. It will contain a systematic sampling protocol for sensor data calibration.

 

Expected results from phase 3:

During the third phase of the project, the RapidMapper will be equipped with all sensor modules and arranged for extensive mapping campaigns on different test fields. By means of sensor data fusion and machine learning technologies, the prediction of different soil properties can be further improved. The DSS for the transfer of the sensor data into information, in order to make appropriate management decision, will be completed and tested at the test fields. For meeting the needs of the soil-crop models for high-resolution subsoil information, a second multi-sensor platform, able to extract and measure soil cores, will be developed and built (RapidProfiler). Finally, based on our results from nine years of I4S project, we will make best-practice recommendations for sensor systems and sensor combinations for high-resolution soil property prediction.