Child pages
  • Abstract: Geophysical information from high resolution satellite synthetic aperture radar (SAR) imagery, Y1
Skip to end of metadata
Go to start of metadata

High resolution sea surface roughness data from synthetic aperture radars (SAR) aboard space-borne satellites capture important features of the ocean at global scales. The recent launch of 2 identical SARs from the European Space Agency’s (ESA) Sentinel-1 (S1) mission in April 2014 and May 2016 are now provide a huge amount of data (>500 Tb) that can used for oceanographic research. The intended research is focused on developing algorithms to extract ocean wave information from S1 in both open ocean and marginal ice zone environments. The first application is dedicated to estimating significant wave heights in the open ocean. We will use a proven satellite technology, altimetry, as reference to develop machine learning techniques that use SAR parameters as input. A large motivation of this work is to improve our ability to extract wave heights in extreme conditions like tropical cyclones and extra-tropical cyclones. The second application focuses on extracting wave spectra from the SAR imagery in ice-covered waters in polar regions. This information is highly valuable and allowing more detailed studies of the momentum and heat budget between the wave-ice-ocean interfaces.

SAR measures sea surface roughness at very fine spatial resolution (<10 m). The high resolution SAR images from the 2 recently launched Sentinel-1 satellites are revealing stunning structures of small scale processes never observed as regularly in time and space as now. Various oceanic and atmospheric processes leave their imprint, much like a fingerprint, on the ocean surface. This information are highly valuable to study air-sea interaction, ocean wave propagation, wave-ice interaction, and small-scale atmospheric weather phenomena. SAR is the unique contemporary technology able to estimate directional wave information at global scales. Wave spectra (energy for each wave direction and frequency) can be extracted from the imagery. However, high-frequency random motions create destructive signals that the radar cannot resolve. Therefore, SAR derived wave spectra using standard image modulation mapping techniques cannot resolve high frequency waves and integrated wave parameters such as significant wave heights cannot be estimated. Under storm conditions, this distortion is stronger and only a reduced set of sea state parameters can be extract from SAR which is a clear limitation of these methods.

1) We are developing an improved method to empirically estimate the significant wave height (and potentially other sea state parameters like wave average periods or wave energy) from the SAR imagery. We are using another satellite technology, namely altimetry, as a reference to develop machine learning techniques like neural networks to estimate significant wave heights from the SAR modulation spectra. The significant wave height from altimeters is a proven technology and accurately estimates tall wave heights in extreme storms; so, it is an ideal reference dataset. There are now an unprecendented6 altimeter missions in operation. This allows for a sufficiently number of satellite co-locations (ie satellites view the same patch of ocean in a sufficiently same window in time and space). We expect to improve our estimation of extreme waves from SAR directly within storms. This work is timely because there is project in Europe funded by the European Space Agency to use all previous satellite missions including both altimetry and SAR to the best quality estimates of the sea state with focus on generating suitable time series for climate studies. There will be a competition within the next 6 months on which SAR empirical model performs the best. This method will then be used to for all previous missions as well: ENVISAT, ERS2, and ERS1.

  • No labels