Comparison, harmonization and integration of bathymetric datasets from multiple sources; finding the most accurate method to create an improved bathymetric dataset by integrating datasets from different sources for the BASE-Platform

Organised by Laboratory of Geo-information Science and Remote Sensing

Tue 12 July 2016 09:30 to 10:00

Venue Lumen, gebouwnummer 100
Room 2
By Maarten van Doornik (the Netherlands)

There are many applications that require information about the water depth of oceans, seas and lakes. The study of water depth is called bathymetry. Bathymetry measurements are mainly done by echo sounding from ships. However, because this is very time and cost consuming, Earth Observation (EO) data is becoming increasingly important. This approach, called Satellite Derived Bathymetry (SDB), is an innovative, rapid and cost effective approach to determine bathymetry from space. There are different techniques for measuring bathymetry from space that are useful at different depths and which all have their own opportunities and limitations. The challenge is to combine these different techniques in such a way that a complete bathymetry dataset is generated extending from coast to deep ocean that takes advantage of the benefits of every single technique. That is what Deltares, an institute for applied research in water and subsurface, is aiming to achieve in the BASE-Platform project, of which this thesis research is part. The main objective of this thesis research is to compare and harmonize different bathymetric datasets and to test methods for integration to create an improved bathymetric dataset. The data that were used are the open low-resolution grids from GEBCO and EMODnet, together with SDB data from Landsat 8 and some ship measurement data collected by Crowd Sourced Bathymetry (CSB) and the Indian Navy. Five different methods for integration were tested: averaging, weighted averaging on pixel level based on the variance of the prediction error, weighted averaging on layer level based on the variance-covariance matrix, multiple regression and Regression Kriging. It was found that Regression Kriging produced the highest accuracy. This can be explained by the fact that bathymetry is a spatially distributed variable with spatially correlated residuals. By predicting these residuals using a simple Kriging interpolation and adding these to a prediction done by a multiple linear regression, an output result is generated which is more accurate than each of the input datasets. The techniques were tested in two totally different study areas: one shallow coastal area and one deep ocean area with some shallow part and rapid changes in depth. However, only in the shallow coastal study area a reliable result was obtained, because there was a lack of data in the other study area. Still, the results are promising and are of interest for the BASE-Platform project.

Keywords: Bathymetry; Deltares; BASE-Platform; Satellite Derived Bathymetry; Accuracy; Regression Kriging