부산대학교 도서관

Geographic data can roughly be divided into two main categories: vectors (including polygons, lines and points) and rasters (composed of uniform grid squares). Rasterisation is necessary for geographic tools that operate on raster data when the original data are represented using vectors. Our research group is interested in modelling the spread of disease through heterogeneous woodland landscapes: we employ simulation tools on rasterised landscapes, but forestry information is typically provided as polygons. We must therefore determine, for each grid square, how much of its area is made up of a particular species of tree. Up until now, Esri's ArcGIS has been used to calculate the intersection between polygons and grid squares, but this approach is unfeasibly slow and requires workarounds for large data sets and finely resolved grids. In this paper, we introduce a new approach towards solving this problem using the Clipper Library, a free open-source implementation of Vatti's clipping algorithm. We show that Clipper generates grid square rasterisations of representative Ash and Larch data sets between 10 and 20 times faster than ArcGIS. Clipper produces results that are at least as accurate as ArcGIS and can be applied to larger data sets without the need for workarounds.