Abstract
The volume of geospatial data acquired by satellites, from the air, on the ground, on the sea or in the sea using image, lidar or sonar technology is continuously growing. Surveys are often conducted in locations that have existing data from before, providing updated datasets. The data acquired can either confirm already existing information, or it can represent changes important to detect and understand. In the fp7 IP www.iqmulus.eu (2012-2016) technology for the representation of huge datasets of the ocean floor was developed using a novel technology called locally refined splines. The work is continued in the national Norwegian project www.sintef.no/en/projects/analyst/ (2017-2021). The smooth component of the data is represented using spline surfaces where local degrees of freedom are added where required by the data behaviour. The result is a spline surface representing the smooth component and a point set that is not part of the smooth behaviour. Data compression experienced are typically two orders of magnitude. The representation is well suited for GPU accelerated visualization. The spline surface representation also makes it easy to check if new data sets confirm the already existing information, or if deviations are found that must be checked and analysed in more detail, with a possible update of the spline model. So far, the approach has been tried out on lidar and sonar data. However, we see a great potential for testing the approach on sets of multi-spectral image satellite data to represent the data as a compact locally refined spline in a consistent coordinate system.