Request for Proposals: Enhancing Global Fishing Activity Predictions Through Data and Model Development

Request for Proposals: Enhancing Global Fishing Activity Predictions Through Data and Model Development
Organization name: Global Fishing Watch
Website: www.globalfishingwatch.org
Submission deadline: January 6, 2025
Please see submission requirements below.

Global Fishing Watch (GFW) is an international nonprofit organization dedicated to advancing transparency of human activity at sea, with the goal of improving ocean management. By combining cutting-edge technology, satellite data, and machine learning, we publicly share insights into human activities at sea to foster fair and sustainable use of our ocean.

Founded in 2015 through a collaboration between Oceana, SkyTruth, and Google, GFW became an independent organization in 2017. Our groundbreaking efforts have resulted in the first-ever global map of industrial fishing and offshore energy activity, using satellite data and cloud computing.

We are now expanding this work with the Open Ocean Project (learn more here), which will map and monitor all major human activities at sea. This includes industrial and small-scale fishing vessels, offshore energy infrastructure, aquaculture operations, and more. By applying deep learning and data fusion techniques to massive datasets, GFW aims to provide actionable information that governments, researchers, NGOs, and the private sector can use to protect our oceans and the communities that depend on them.

Project Description

We aim to release an updated general fishing model in 2025 to enhance the accuracy and utility of fishing activity predictions on vessel tracking data. Priority areas for improvement include reducing false positive fishing predictions in MPAs, EEZs, ports, and anchorages; improving predictions near coasts; and increasing accuracy for VMS. Additionally, we will aim to provide improved accuracy metrics across different dimensions such as fishing gear, ping rate, region, fleet, vessel size, and device type. These metrics will provide users insight into the reliability of the datasets for their specific use cases. 

We are looking for consultant support for an approximate 12 months period (January to December 2025)  to support the timely delivery of those objectives.

Deliverable 1: Labelled data collection (Q1 to Q3)

To improve the model and to evaluate its ability to predict fishing activity, we need to collect labelled vessel tracking data. This task will involve selecting representative vessel tracks to label, training and coordinating a labelling team, and validating labels with expert feedback, acknowledging potential differences in expert opinions.

Tasks:

• Review existing labelled datasets for AIS and VMS.
• Identify representative tracks to label to expand the existing training datasets.
• Collaborate on data labelling efforts.
• Train team members on labeling fishing tracks, emphasizing consistency and accuracy.
• Help select test sets to assess model performance.
• Ensure datasets are well-documented and reflect diverse fishing patterns, gear types, regions, fleets etc.
• Conduct analyses to determine the minimum required ping rates for accurate predictions across different fishing gear types.

Deliverable 2: Fishing hours analysis (Q1 to Q2)

Conduct a comprehensive analysis to understand differences in fishing activity outputs between the updated model and its predecessor. This analysis will identify significant changes, their root causes, and their implications for users.

Tasks:

• Analysis to compare estimated fishing hours between the new and old models. Highlight key differences and trends.
• Investigate how variables such as AIS data reception impact total fishing hours.
• Investigate consistency and alignment between GFW’s fishing events dataset and fishing effort dataset. 

Deliverable 3: Communication and Documentation (Q4)

Support the release of the updated fishing model through engaging communication materials, detailed technical documentation, and public outreach efforts.

• Write clear and comprehensive documentation for fishing model code repositories, ensuring accessibility for developers and collaborators.
  
  

Required Skills & Experience

• Experience evaluating machine learning models
• Demonstrated experience and data science skills with Python or R
• Experience with SQL
• Understanding of fishing practices and modeling fishing activity
• Ability to work with large datasets and visualize data effectively
• Experience with version control software and collaboration tools such as git and GitHub
• Analytical, detail oriented and self-motivated
• A deep curiosity and dedication to critical thinking
• Willingness to take ownership of projects and effectively communicate updates in a transparent and proactive manner
• Ability to manage multiple priorities under tight deadlines while performing in a fast-paced, collaborative environment

Preferred Skills & Experience

• Experience in geospatial analysis of remotely sensed data
• Experience engaging with academic researchers and the peer-review process
• Familiarity with issues of marine ecology or policy
• An understanding of the complexities and collaboration required to work effectively for a globally remote organization

Submission requirements

Interested parties should submit a proposal with a descriptive plan and budget via email to research@globalfishingwatch.org. 

The maximum budget available for this project is $100k USD.

The proposal should consist of:

• A cover letter explaining the provider’s interest and qualification for the consultancy including a description of previous experience with the types of work proposed in the RFP.
• A narrative proposal outlining how the provider intends to fulfil the responsibilities outlined in the previous section.
• A descriptive budget outlining the total cost of fulfilling the terms of the RFP.
• Resumes/CVs of key project personnel.

Submission deadline: Proposals should be submitted by January 6, 2025

• Shortlisted proposals may be invited to an interview by Jan 10, with a view to finalize the agreement and begin the work no later than January 31
  
  

Consultants must be based in