Efficient Constrained Delaunay Triangulation implementation in Java for spatial hydrological analysis

6pTo quantify the impact of increased urbanization on the peri-urban hydro-systems, it is necessary to design adapted models. Currently, most of the hydrological models, based on classical raster approach, suffer from a lack of efficiency because they essentially do not take into account the urban artifacts and do not match complex geometries such as urban ones (heterogeneous objects, strong space's partition, roads, pits, buildings...). That is the reason why, we have decided to choose a Triangulated Irregular Network (TIN) model. This approach requires a Constrained Delaunay Triangulation (CDT) preprocessing phase. Also CDT algorithms are numerous, we must admit that there is a real lack of free, efficient and robust implementations adapted to our technological choices. Indeed, all the hydrological simulations we have developed, had been made in the context of the OrbisGIS platform (an open source GIS developed in Java in the French IRSTV Institute). The main objective, with OrbisGIS, is to remain coherent and give to the community a consistent set of simulations tools and methods for the spatial analysis of the urban environment in the context of a single and cross-operating systems GIS. In this paper, we will first present the OrbisGIS platform. The remainder of this paper is dedicated to the CDT algorithm we choose, its implementation and some comparisons with other implementations. The last part of this paper is dedicated to the hydrological process and the first results we have obtained