Mapping Seagrass Meadows with Satellite Imagery and Computer Vision

Local Chapter Trieste, Italy Chapter

Coordinated by Italy ,

Elena Andreini

Apply here

Status: Upcoming

Project Duration: 08 Jan 2024 - 09 Feb 2024

Project background.

Seagrasses, forming vast marine meadows in shallow salt waters from tropics to the Arctic, are vital for biodiversity. They provide habitats for fish and shellfish, supporting local coastal economies. Moreover, they stabilise sediment, absorb wave energy, and contribute significantly to carbon absorption, making them crucial allies in combating climate change.

The problem.

The declining health of Posidonia oceanica meadows in the Mediterranean Sea is a pressing concern, attributed to climate change and various human activities. These meadows, crucial for their ecological significance, are in jeopardy due to factors such as warming, ocean acidification, coastal urban development, fishing, and aquaculture.

This decline has led to a substantial loss of goods and services provided by these ecosystems. While P. oceanica is the most vital and studied seagrass species in the region, there has been a limited effort to collate and provide a comprehensive distribution of these meadows. This lack of information impedes our ability to effectively address the regression of these critical habitats.

Project goals.

Our project aims to develop accessible and efficient methods for mapping seagrass meadows using readily available satellite imagery and  computer vision. We envision these results as valuable tools for long-term seagrass monitoring, including tracking restoration and replanting efforts. Our primary goal is to create a pixel-level classification and segmentation model to map seagrass distribution, with a focus on the Mediterranean, especially Italian waters. Using computer vision techniques, we'll identify seagrass regions by analyzing satellite images and classifying pixels based on data from public databases indicating seagrass presence or absence. An essential aspect of this project involves comparing our model's outcomes with established habitat suitability models for P. oceanica presence. Habitat suitability models predict species presence in a location by analyzing the relationship between observed occurrences and environmental conditions. They assess marine habitat status, forecast species distribution changes from human and environmental impacts, and guide restoration efforts by identifying optimal areas.

Project plan.

  • Week 1

    Literature research
    Success metric definitions
    Data collection

  • Week 2

    Data collection
    Data cleaning and preprocessing
    Modelling

  • Week 3

    Data preprocessing, Modelling, Comparison with habitat suitability models

  • Week 4

    Deployment : public dashboard for data visualisation

Share project on: