Résumé
In recent years, Costa Rica has become the world's leading producer and exporter of pineapple, with more than 65.000 hectares of its territory dedicated to this crop. The development of these intensive pineapple crops is mainly concentrated in the north of the country, especially in the border area with Nicaragua. This rapid expansion in these regions has many economic, social and environmental consequences. In order to understand the local impact of this phenomenon, a first analysis of land use changes for the benefit of pineapple monocultures must be performed. For this purpose, pineapple crop delineation layers in 2000 and 2019 were provided by MOCUPP, a United Nations agency responsible for monitoring agricultural areas in Costa Rica. Landsat 5 satellite images from 1998 and 2001 were also used to generate land cover classifications. The pineapple crop delineation layers produced by MOCUPP are characterized by heterogeneous granularities and positional accuracies, complicating the analysis of land cover changes through the generation of artefacts. To address this problem, an artefact removal procedure was conducted, based on the use of minimum shape and size indicators. The assignment of land use changes to pineapple crops allowed the assessment of deforested areas between 2000 and 2019. Thus, over this period, more than 6300 hectares of forest were lost to pineapple crops, mainly in the cantons of Los Chiles and San Carlos, in the northern part of the country. Finally, this article illustrates the problems associated with the use of heterogeneous data to analyse land cover changes. To avoid these problems, the use of prior data integration procedures, such as data matching algorithms, is recommended