Satellite imagery is a rich source of information about the coastal system now and in the past. This information is essential to an understanding of system dynamics and to estimate the effects of interventions. Deltares is developing algorithms that can produce information from satellite imagery about coastal systems at basically any location in the world.
Coastal systems often consist of a combination of interacting morphological features such as bars, channels, shoals, coastlines and dunes. These features are dynamic in space and time and determine to a large extent the long-term behaviour of the coastal system, as well as the impact of human interventions on the coast. In most areas in the world, long-term data about the dynamics of morphological features – and therefore an understanding of the system – is inadequate or lacking. Satellite imagery is available for some decades into the past and it therefore opens up opportunities to fill in at least some of those gaps and so supplement other data sources, especially in data-poor environments.
Satellite imagery is becoming more and more freely available at increasing resolutions in time and space. In addition, technology for the processing of these images is developing rapidly. Google recently launched Earth Engine as a platform for Earth Science and data analysis. This technology opens up enormous opportunities for the detection of coastal features and the analysis of the associated trends.
Using Earth Engine, Deltares developed algorithms for the automatic detection of coastlines, shallow shoals, crescentic sand bars, flood plains and vegetation in the coastal zone. Analyses are mainly based on free satellite imagery from the
Landsat missions initiated by USGS and NASA, but they can also be performed on higher-resolution, purchased, imagery. On the basis of a user-defined spatial extent and temporal range, morphological trends can be analysed within seconds or minutes. An example of this is shown for the Sand Motor, a mega-nourishment project on the western coast of the Netherlands. The automated feature detection technique proved to be capable of detecting the temporal dynamics in this artificial sand feature. Verification based on jet-ski surveys produced very promising results regarding accuracy.
The automated feature detection technique proved to be very useful in several commercial and research projects at different locations in the world such as the Netherlands, South Korea, Africa and Colombia. The Colombia project focused on building a validation dataset for long-term coastline changes with satellite images in order to identify coastal zones at risk. Since long-term survey data was limited for this location, the use of satellite imagery was essential to perform this assessment.
Future work will focus on the quantitative validation of the technique for cases where in-situ data is available alongside satellite data. Corrections for the effects of clouds, tide, waves and storm surge will also require further attention. We also aim to make the algorithms easily accessible to experts and end users.
wiebe.deboer@deltares.nl
T +31(0)6 4691 1209
Automated feature detection
using satellite imagery
Sand Motor detection from satellite imagery
Sand Motor
Contact
Deltares | R&D Highlights 2015 Delta Infrastructure
Further reading:
https://earthengine.google. com/