Evaluation of a decision support system for the recommendation of pasture harvest date and form

Tobias Reuter; Juan Carlos Saborio Morales; Christoph Tieben; Konstantin Nahrstedt; Franz Kraatz; Hendrik Meemken; Gerrit Hünker; Kai Lingemann; Gabriele Broll; Thomas Jarmer; Joachim Hertzberg; Dieter Trautz

In: Informatik in der Land-, Forst- und Ernährungswirtschaft -- Referate der 43. GIL-Jahrestagung. GIL-Jahrestagung (GIL-2023), February 13-14, Osnabrück, Germany, Köllen Druck & Verlag GmbH, 2/2023.


The task of generating automatic recommendations of pasture harvest date and form was previously addressed through a knowledge-based decision support system (DSS). The system follows expert rules and exploits data such as the weather history and forecast, the growth stage of grass and legumes, plant height and crude fibre content. In this paper we present the results of our evaluation of this DSS on 26 fields in West and Northwest Germany. We compared the suggestions made by the DSS with the decisions of expert farmers and obtained an accuracy of R²=0.746 and RMSE=7.83 days. The best results occurred for intensively managed fields for dairy cows, with an R² of 0.891 and RMSE of 3.20 days. We conclude our DSS and its underlying methodology have the potential to support farmers and secure high-quality fodder.


Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence