Caption: Sea surface temperature of a 15 km by 15 km patch of the ocean recorded by the Landsat 8 satellite. The image on the left is the original image acquired by the satellite, whereas the image on the right is the outcome of the filtering process developed by the SAT-FIRE team.
Sea surface temperature recorded by satellite systems are very useful in weather forecasting and monitoring the effects of climate change. The USGS Landsat-8 satellite can capture images of sea surface temperatures at a high resolution of 100 metres every 16 days. However, the captured images are usually affected by stripes caused by sensor mis-calibration. A team from the Data Science Research Platform (DSRP) at the University of Malta has developed a filtering technique that significantly reduces the striping artefacts present in Landsat-8 data. Preliminary results were presented during the 20th meeting of the Group for High Resolution Sea Surface Temperature held at the European Space Agency - European Space Research Institute (ESA-ESRIN) in Frascati (Italy) in June 2019.
The work is part of the SAT-FIRE (SATellite data Fusion and Imaging Resolution Enhancement for coastal areas) project, which has received 150 000 Euro in funding by the first-ever National Space Research Fund. It aims to use deep learning and artificial intelligence to improve the spatial resolution of current Earth Observation satellite systems. The European Copernicus’s Sentinel-3 provide a thermal image of the same location on Earth every two days, therefore allowing for frequent updates for forecasting purposes, however each pixel in the image is of 1 km length. This means that it would not be possible, for instance, to distinguish between certain areas along the coast, such as Marsamxett Harbour from the Grand Harbour. Therefore, by fusing information from the Sentinel-3 and Landsat 8, more accurate predictions of marine currents can be provided, aiding divers, search and rescue operations and coastal monitoring. A team of three researchers (one post-doctoral researcher and two Masters students) have been hired to work on the project. The principal investigator on this project is Dr Ing. Gianluca Valentino, with the co-investigators being Dr Ing Reuben Farrugia from the Department of Communications and Computer Engineering and Dr Anthony Galea from the Department of Geosciences at the University of Malta.
About the project
Project SAT-FIRE financed by the Malta Council for Science & Technology, for and on behalf of the Foundation for Science and Technology, through the Space Research Fund.
Media enquiries
The Data Science Research Group can be reached at dsrp.research@um.edu.mt
Dr Ing. Gianluca Valentino can be reached at gianluca.valentino@um.edu.mt
For further information about the group, please see https://www.um.edu.mt/research/researchprojects/dsrp
