Effets de I'echelle sur ies proportions des classes thematiques et I'analyse des patrons de surface a partir d'images MEIS et LANDSAT

摘要

The efficient integration and use of current and forthcoming remote sensing data require an understanding of the scaling effects. At the landscape level the impact of mapping scale occurs in many ways for example changing the proportion of land cover classes or altering landscape pattern as a result of aggregation of smaller patches in large mixed-pixel entities. Fine scale observations are needed to ensure full understanding of land cover and land use derived at broader scales. In this study the effect of scales is examined using airborne data from MEIS II sensor (2.5 m spatial resolution) and LANDSAT images in the north west of New Brunswick. Airborne data were spatially degraded to increasingly coarser spatial resolution up 80 m. The original classified images were considered as a reference for evaluating changes in land cover proportions. Landscape metrics like patch shape index and square pixel index were evaluated in relation with changes in the spatial resolution.