Smallliolder agriculture provides 90 % of primary food production in developing countries. Its mapping is thus a key element for national food security. Remote sensing is widely used for crop mapping but is underperforming for smallholder agriculture due to several constraints like small field size, fragmented landscape, highly variable cropping practices or cloudy conditions. This research aims to develop an original approach combining remote sensing and spatial modelling to improve crop type mapping in complex agricultural landscapes. The method is based on the joint use of high spatial resolution satellite imagery (Spot 6/7, to characterise the landscape structure through image segmentation), and high revisit frequency time series (Sentinel 2, Landsat 8, to monitor the land dynamic processes), combined with spatiotemporal rules (STrules) expressing the strategies and practices of local farmers. The spatial dynamics modelling is done using the Ocelet language. This combination is processed in 3 steps: (1) Each crop type is defined by a set of general rules from which a model-based map of crop distribution probability (MDist) is obtained in Ocelet, (2) a primary crop type map is produced by satellite image processing based on a combined Random Forest (RF) and OBIA classification scheme and each object is labelled with the classes' membership probabilities, (3) the STrules are applied in Ocelet to identify objects with classes locally incompatible with known farming constraints and strategies. The result is a map of the spatial distribution of crop type mapping errors (omission or commission) that are corrected through the joint use of spatiotemporal rules and RF classes' membership probabilities. Combining remote sensing and spatial modelling is thus a future way to better characterize and monitor complex agricultural systems.