Organizations involved: SINTEF (Norway) and HRW (UK)
Marine Services and Workflows
A number of the marine workflows are based on a new approach for surface representation called Locally Refined B-splines (LR B-splines). In IQmulus LR B-splines approximation/regression methods have been developed for geospatial big data, allowing millions of points to be approximated by a locally refined spline surface. This is a major innovation compared to traditional multivariate spline approximations/regression methods that a tensor product approach refinements of the representation leads to an explosion in data volume of the representation. The LR B-splines refinements only add degrees of freedom locally when needed, and consequently the storage footprint is kept low. For ocean floors and sea beds that have a large smooth areas we have observed a reduction of data volume by two orders of magnitude. By sampling the resulting surface traditional surface representation such as GeoTIFF can be generated and the sampled points used as input to other workflows.