This paper presents a new geographical analytical model for studies on agricultural landscape change. The model builds on the idea presented by Torsten Hägerstrand that neighbourhood configurations act as filters in change processes. The proposed approach operationalizes this idea of filtering into a geographical interpretation model that combines three different strands of research: ontologies of land use in geographical information science; local spatial contextual analysis of remote sensing data; and agency- and actor-sensitive functional time-space analysis of farming. This approach includes the use of satellite images and makes it possible to infer land-use changes as reactions to long-term policy changes, thus demonstrating how landscape configuration and land use are changing. The model is assessed by applying it to a case study in western Östergötland in central Sweden.