The goal of this project is to analyze the dynamics of small fishing vessel fleets and provide information to help policy-makers design legislation to prevent Illegal, Unreported, and Unregulated (IUU) fishing.
The problem with monitoring the behavior of these vessels is that those which are engaged in IUU fishing do not wish to make their actions known. Therefore these vessels are unlikely to use systems such as AIS which reveal their location. We thus need remote, passive data sources to monitor their behavior, and one such source is satellite imagery. Through Stanford we have access to Planet Labs, inc.'s PlanetScope daily imagery. This imagery is potentially useful because of the its relative frequency - the entire global landmass, including most near-shore waters, are imaged daily. This allows us to model the dynamics of fishing fleets with an unmatched temporal resolution which is necessary to understand the seasonalities which affect the decisions of vessel operators.
The work necessary to extract the relevant information from this imagery and provide useful insights may be broadly divided into two tasks
- Developing a vessel detection system
- Using data generated by the detection system to design and fit statistical models which describe the dynamics of the vessels
The outputs of (2) should be novel and useful to policy-makers and other stakeholders. It is of course the job of the researchers to summarize and interpret the results of this model.
This repository is a work-in-progress. It will eventually contain a trained vessel-detection model, statistical models created using data derived from the vessel detection model, the code necessary to implement both models, and working notes from the module's creation.
Poisson was developed with the generous support of a research grant from the Stanford University Department of Statistics and continuing support from the Stanford Woods Institute in affiliation with the Stanford Center for Ocean Solutions.
Created by Richard Correro in 2020. Contact me at rcorrero at stanford dot edu.