This article describes the use of a case-based reasoning approach on spatial organization of farms and their management. Spatial organization graphs (SOG) model observed data. They can be recorded within a computer device and automatically compared. Functioning explanations are associated to SOG and are organized within a case base. Thus, according to graph similarity, explanations can be proposed and adapted to new SOG. The ROSA system, still in progress, has been used to evaluate the similarity calculation and the proposed explanations. The inter-disciplinarity framework of this study allows to draw first lessons on knowledge formalization and reasoning on spatial organization of farms.