Peskas Coasts is the automated data pipeline powering the coastal fisheries analytics at Peskas.org.
This project processes raw ocean tracking data and generates accessible web dashboards for the Western Indian Ocean (WIO) region, including Kenya, Mozambique, and Zanzibar, see the dashboard here
Peskas Coasts is an automated data pipeline. Every two days, a scheduled GitHub Actions workflow performs a series of data gathering, classification, and spatial modeling tasks:
- Data Ingestion: Fetches the latest boat GPS tracks from Pelagic Data Systems (PDS) and combines them with human-reported survey data (KoboToolbox).
- Fishing Activity Prediction: Uses a statistical model
(
ssfaitk) to classify parts of the boat’s journey as fishing activity. - Spatial Modeling: Translates GPS pings into standardized hexagonal grids (H3) and calculates fisheries metrics, like Catch Per Unit Effort (CPUE).
- Dashboard Delivery: Exports the final results into web-ready formats (JSON/GeoJSON) and pushes them to MongoDB to update the maps on Peskas.org.
We’ve designed this documentation to be accessible to stakeholders, researchers, and developers alike:
- 📖 How the Pipeline Works: A plain-English walkthrough of our automated GitHub Actions workflow. Learn how data travels from a boat’s GPS tracker to our web portal.
- 🗺️ Understanding the Models: Discover how we calculate Catch Per Unit Effort (CPUE) and why we use Hexagonal (H3) gridding to protect fisher privacy while highlighting ocean hotspots.
- 🛠️ Reference: For developers looking to interact with the underlying R functions and APIs.
By automating data cleaning, model prediction, and spatial aggregation, Peskas Coasts provides updated spatial datasets for coastal monitoring.
Peskas Coasts is proudly developed as part of the WorldFish Center’s Peskas initiative.