Combining and processing large heterogeneous datasets for bathymetry developmentc

When working with very large datasets, using traditional GIS software, one quickly reaches the limit of the processing capabilities of a high-end desktop computer. Hence, in order to manage potentially overlapping, extensive data, combined from different sources both in terms of accuracy and spatial resolution, a novel approach to bathymetrydata development, using the terrain module of the open-source package Gerris, has been developed and is presented. Each individual bathymetrydataset is converted into the kd-tree database format used by Gerrisfor efficient handling of bathymetrypoints. Besides its efficiency at recovering data point statistics for any arbitrary region, this format also allows easy blending and weighting of the individual bathymetrydatasets. Gerrisis used to automatically create a bathymetrygrid of local resolution dictated by a predefined criteria based on the local data density, the resolution of the desired output or any scale of interest. Once the grid is created, Gerrisgathers the statistics of all the points of all the datasets in each cell of the grid and reconstructs the bathymetryusing a least-square bilinear approximation technique. From the continuous bathymetrycreated, depths can be output at any location in the domain, such as the nodes of a finite element grid. This method was initially used to create a bathymetrydata for the Hauraki Gulf, New Zealand, combining electronic navigation charts, multi-beam and single-beam data, LiDAR (Light Detection and Ranging) and broad scale datasets such as GEBCO and SRTM. Following this exercise and fine-tuning of the methodology a bathymetrydataset has been compiled for all New Zealand waters. This illustrates how the management and use of all available datasets can be based within a single framework, with the ability to easily integrate future data.