Semi-virtual spatial simulations of land management practices in Languedoc Vineyards: a way to deal with incomplete knowledges of spatial distributions in cultivated landscapes
Résumé
In this paper, methods that aim to simulate spatial distributions of Land Management Practices (LMP) at high spatial resolution and over large extents as required by spatially distributed environmental modelling are presented. The principle of theses methods, in the scope of geostatistical conditional simulation methods, is to simulate a set of equally probable spatial patterns that all respect the knowledge that have been collected on the landscape features, i.e; spatial laws and data (descriptive approach); and/or the driven factors of the studied LMP (factorial approach). The differences between the simulated spatial patterns can be seen as a representation of the spatial uncertainties. To illustrate these methods, two examples of simulations of spatial patterns of LMP are presented. They involved specific methodological developments and can be considered as representative of descriptive and factorial mapping approaches respectively: i) the simulation of ditch networks reconstructed from an incomplete set of reaches observed by remote sensing and ii) the spatial simulation of weed control practices at plot scale from a set of driven and correlated factors.
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Loading...