ShellEye Water Quality Event Map
Subscribe to ShellEye bulletin service
Register your interest
Using new satellite technology
22 January 2019
Dr Andrey Kurekin, from
Plymouth Marine Laboratory
, summarises the research into incorporating brand new satellite technologies into the ShellEye service.
Exploitation of new satellite Eath observation (EO) sensors, such as Sentinel-3 OLCI, sufficiently extends the capabilities for early warning of harmful algal bloom (HAB) applications by providing more accurate, detailed and regular ocean colour measurements. However, exploitation of new sensors also brings new challenges to the developers of EO data processing algorithms.
Adaptation of HAB detection methods to a new Sentinel-3 OLCI sensor was carried out at ShellEye-DEMO project with the objectives to provide a new HAB risk maps product with improved spatial resolution up to 300 meters (i.e. 1 pixel equates to 300m). The automatic HAB detection method, previously developed at PML, uses machine learning techniques to recognize HAB signatures in ocean colour data. The accuracy of this method depends on the number of example satellite images of HAB being used for training the algorithm. But the number of harmful algal blooms, available for a new EO sensor, is limited and this adversely affects the accuracy of HAB detection. To overcome this limitation alternative training strategy was developed, which is completely independent of EO sensor data and can be even used for EO sensors that do not yet exist. The new approach incorporates together several methodologies: cultivating harmful algal species in the laboratory, measuring inherent optical properties and modelling sensor measurements of ocean colour data. The methodology was successfully tested on
HAB classifier with application to Sentinel-3 OLCI sensor. The classifier was trained using the laboratory experiment data and then applied to Sentinel-3 OLCI data (see Figure 1). It demonstrated promising results in comparison with the previously designed MERIS HAB detector technique.
PML team is now working on validation of a new Sentinel-3 OLCI HAB classifier and its integration into the operational EO data processing chain. This will help to extend the capabilities of ShellEye-DEMO in providing extended monitoring capabilities of HAB in the UK coastal waters.
After successful testing of new methodology in ShellEye-DEMO, we are planning to publish our achievements in a peer reviewed research journal. We envisage that new HAB training technique will help to improve the efficiency of early warning of HAB using new EO sensors and data.