Nirchio, FrancescoFrancescoNirchioForgia, V.L.V.L.ForgiaPasquariello, G.G.Pasquariello2020-09-172020-09-171995-01-01https://hdl.handle.net/20.500.13025/219The objective of this paper is to present a methodological approach devoted to the sea state parameters extraction from ERS-1 SAR data. Wave parameters (direction and wavelength) retrieval is not a straightforward task, due to the nonlinearity of mapping the sea surface into the detected image. To overcome these difficulties a neural network approach has been tested. A SAR ocean image simulator has been used to create a set of image windows for the learning of a multilayers network and the trained network has been applied to a set of independent examples corresponding to various wave directions and wavelengths. The results, obtained on simulated data, seems to be encouraging and independent of linearity or nonlinearity of the wave dataERS-1Extraterrestrial measurementsImage retrievalLinearityMulti-layer neural networkNeural networksOceansParameter extractionRemote monitoringSAR ocean image inversionSea measurementsSea surfaceSurface wavesTestingWeather forecastingdirectionfeedforward neural netfeedforward neural netsgeophysical signal processinggeophysics computingmeasurement techniquemultilayer networkneural networknonlinearityocean waveocean wavesoceanographic techniquesparameters extractionradar applicationsradar imagingradar remote sensingremote sensingremote sensing by radarretrieval methodsea statesea surfacespaceborne radarsynthetic aperture radartrained networkwavelengthconference paper10.1109/IGARSS.1995.521104http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=52110454dcce078580fe1368eeae5b