Detecting Coastal Aggregation of Macro-Plastic Litter Using Satellite X-Band SAR Imagery
Journal
IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING
ISSN
1939-1404
Date Issued
2026-03-18
Author(s)
Verlanti, Anna
di Michele, Luigi
Fanti, Alessandro
DOI
10.1109/JSTARS.2026.3675344
Abstract
This study is to summarize the lesson learned during
the July 2024 experimental campaign where targets, consisting
of floating plastic and natural debris, were jointly observed in a
South Italy lake by drone and satellite instruments. The satellite
layer consists of the X-band Synthetic Aperture Radar (SAR)
equipped onboard the Italian Cosmo-SkyMed Second Generation
(CSG) constellation, while the drone layer consists of an optical
camera.
A set of multi-polarization features, namely the co- and crosspolarized normalized radar cross section (NRCS), the SPAN and
the like-to-cross-polarization ratio, are extracted from the CSG
SAR measurements. Their sensitivity to the plastic target, along
with their ability to discriminate plastic from natural floating
debris, are investigated using objective metrics.
Experimental results, obtained under calm water conditions,
show that floating plastic calls for co- and cross-polarized signals
well-above the system noise and well-distinguishable from the
surrounding water using all the features. The plastic target
appears as a spot brighter than the surrounding water in both
the co- and the cross-polarized NRCS imagery. The co-polarized
NRCS calls for the largest plastic-to-water contrast and the
largest separation between the plastic and the natural target.
However, this separation is such that the discrimination is a
challengin
the July 2024 experimental campaign where targets, consisting
of floating plastic and natural debris, were jointly observed in a
South Italy lake by drone and satellite instruments. The satellite
layer consists of the X-band Synthetic Aperture Radar (SAR)
equipped onboard the Italian Cosmo-SkyMed Second Generation
(CSG) constellation, while the drone layer consists of an optical
camera.
A set of multi-polarization features, namely the co- and crosspolarized normalized radar cross section (NRCS), the SPAN and
the like-to-cross-polarization ratio, are extracted from the CSG
SAR measurements. Their sensitivity to the plastic target, along
with their ability to discriminate plastic from natural floating
debris, are investigated using objective metrics.
Experimental results, obtained under calm water conditions,
show that floating plastic calls for co- and cross-polarized signals
well-above the system noise and well-distinguishable from the
surrounding water using all the features. The plastic target
appears as a spot brighter than the surrounding water in both
the co- and the cross-polarized NRCS imagery. The co-polarized
NRCS calls for the largest plastic-to-water contrast and the
largest separation between the plastic and the natural target.
However, this separation is such that the discrimination is a
challengin
