Scientists automated the search macroplastique from satellite imagery

Scientists with the program ACOLITE processed lots of pictures of coastal zones obtained from satellites Sentinel-2A and Sentinel-2B, and developed a new index that allows to distinguish the spot from macroplastique related components on the surface of the ocean. According to the results of their study, published in the journal Scientific Reports, the accuracy of this method reaches 86 percent, and now it can be used in environmental monitoring.

Plastic debris falling into the ocean some time floating on the surface, and then can grow into living organisms, and sink. And in that and in other case it collapses to a particle size of less than five millimeters, which are called microplastics. To track and eliminate pollution of the sea by microplastics is impossible, so scientists are looking for better ways to detect and eradicate large plastic fragments before they sink or break down into small particles.

Have great potential of remote sensing techniques, however, until recently, they were uncomfortable to use for continuous monitoring plastic pollution. For example, widely used in environmental research satellite, Landsat 8 makes multi-spectral images of 30 meter resolution in increments of 16 days, that is, may miss svojeobraznaja plastic spot small size.

The European space Agency a few years ago launched satellites Sentinel-2A and Sentinel-2B, which make pictures a resolution of 10 meters in increments of 2-5 days. These satellites are focused on surveying the land, but also cover large areas of coastal waters that often become the primary foci of plastic pollution.

Scientists led by Lauren Biermann (Biermann Lauren) from the Marine laboratory of Plymouth conducted a study to assess the possibility of monitoring spots macroplastique using images of the satellites Sentinel series. On the basis of information about the regular appearance of the plastic debris of research articles, media, and social networking as a test of the platforms they have chosen the coastal waters of Ghana, West North America, Vietnam and Eastern coast of Scotland.

The method is based on the different reflectivity of objects on the ocean surface relative to the radiation in the near infrared range (NIR). Clean the surface of the sea water it absorbs, and floating objects on it (sea foam, seaweed, pieces of wood and plastic) — reflect. The study authors have determined the spectral signatures of these components using the index NDVI (it is convenient to distinguish vegetation from other objects — water, buildings, asphalt, etc.) and a new index FDI (Floating Debris Index), which allows more accurate identification of spots of plastic floating on the sea surface. The researchers trained the software ACOLITE (it is working with images obtained from satellites) ability to distinguish between materials with the help of machine learning methods (they are not cited in the article are no data on what algorithm was used and on what samples studied). General scheme presented below.

Leave a Reply

Your email address will not be published.