Viability Study of Using Satellite to Monitor Plastic in Marine Enviorments

Our client wanted us to investigate using machine learning for detecting plastics in the ocean. There are clear environment concerns about the number of plastics entering the ocean and an urgent need to help non-profit organizations and researchers identify plastics over large areas. Satellite imaging provides a non-traditional approach and, if open-source imaging can be used, a possibly economical means, for detecting and tracking plastic objects in the ocean. If a viable method can be demonstrated, there are non-profit organizations and research organizations who could benefit from the information and could make a positive environment impact.  

Problem Background

Plastics pose a global threat to ocean health and current methods used to remove plastics in the ocean can be extremely expensive [1]. One report estimates there being 7.25 trillion kg entering the ocean annually[1].  A growing number of organizations are working to remove plastics from the ocean, but it’s difficult to know where that help is needed most. Current methods are also generally passive and don’t track where plastics are coming from or where they are going. Given the importance of removing plastics and difficulty locating them, if satellite imagining can be used for identification, it would offer a promising alternative to existing methods.  

Our Solution

Our client solicited Northwestern Analytics to investigate using satellite imaging for identifying plastic debris in the ocean. Northwestern Analytics reviewed the current literature on the topic including research on multispectral imaging for object detection, existing platforms used for geographical information services (GIS) such as QGIS and ArcGIS, and machine learning efforts in the GIS space. Our initial research identifying Sentinel-2 satellite and its accompanying Copernicus platform as a promising means for using multispectral imaging for object detectionPreliminary research also indicated success in using a combination of spectral bands and matrices to identify specific targets and anomalies. Northwestern Analytics recommended pursuing machine learning methods for conducting unsupervised and supervised clustering as well as in-depth investigation in the spectral signatures of known plastics to determine viability of plastic identification.  

Project:
Research Survey to determine viability of the use of satellites to detect plastics

Client:
London based For-Profit Company

Sponsor:
GIS Practice

Year:
2020

Data Scientist
Brennen Chadburn, Partner

Project Manager:
Andrew D’Amico, Partner

Team:
Michael Purvis, Partner
Antonius Tran, Partner
Erik Meija, Consultant

    Our purpose is to assist individuals and organizations in making a positive difference in their industry and in their community through education, data strategy, and innovation.

    Our mission is to create industry-leading, tailored solutions powered by sound research to assist companies in making data driven decisions and innovating in their field, engaging brilliant people, and cultivating diverse and forward-thinking perspectives.

    Operating Cities

    Chicago, IL
    Miami, FL
    Kansas City, MO
    Salt Lake City, Utah
    San Francisco, CA
    New York, NY

    EMAIL

    info@northwesternanalytics.com

    Contact Us
    +1 866 989 DATA
    +1 312 820 0340
    Cart