The United Nations Environment Programme (UNEP) is the leading global environmental authority that sets the global environmental agenda, promotes the coherent implementation of the environmental dimension of sustainable development within the United Nations system, and serves as an authoritative advocate for the global environment.
Description of Activities on AI
Project 1: Use Case/Problem Statement
To show how water ecosystems are changing over time. A water related ecosystems monitoring project, aided by Google Earth Engine and the European Commission’s Joint Research Center. It works through the application of computer vision and machine learning algorithms to recognize water bodies in satellite image data and map reservoir trends over time.
Project 2: Smart match making of stakeholders
The project aims to do smart matching making of stakeholders in a database based on stakeholder profile, and the various interactions between stakeholders, such viewed, connected.
Project 3: UNEP Q&A Chatbot
Conversational agent designed to initiate dialogue with and respond to user queries through an electronic interface as responding to Frequently Asked Questions (FAQs) manually is cumbersome and time consuming.
Project 4: SDG Meter
A myriad of textual documents produced / consumed by UNEP need to be mapped to SDGs (project proposals, reports, briefings, etc). Such mapping exercises demand extensive expert time and rely on personal knowledge of interlinkages among topics and SDGs. While UNEP counts with experts in several topics, interlinkages with SDGs outside our expertise can be missed out. A web platform is proposed as an aid tool to analyse text document via an algorithm and rate relation to each of the 17 SDGs.
Project 5: Machine Learning applied to chemicals in products and the environment
Interoperability is a key challenge for exchanging information and interlinking knowledge domains. This project aims to dynamically learn from chemical, industrial and environmental taxonomies, ontologies, and data sources to establish linkages among nodes and bridge knowledge gaps as well as to indicate risks and hazards for people and the environment.
Project 6: Spatio-Temporal forecasting of Methane Super-emitters using heterogeneous data stream
Data on methane super emitters is available via different monitoring methods and formats. This project’s aim is two-fold: initially to create a global map combining available live data sources on methane emissions (satellite images, numeric values, self-reporting, text, etc.) to then predict where and when methane super-emissions will take place.
Project 7: Promotion of Countermeasures Against Marine Plastic Litter in Southeast Asia and India (Counter MEASURE project)
The project aims to identify a region-based model for monitoring and assessment of plastic leakage and pollution reduction targeting land-based plastic leakage entering waterways such as rivers and canals or drainages to the sea. As a part of monitoring practice, we developed a machine learning algorithm to detect plastic pollution from aerial images using a drone in order to establish a standardized and cost-effective survey method.
Project 8: Marine Litter Digital Platform
UNEP has been mandated by Member States to establish a new digital online platform for marine litter that serves as a global observatory, solutions and collaboration center that integrates data, assessments, risk, knowledge, while prioritizing action and facilitating access to technology solutions and innovative financing
Challenges and Opportunities
- More datasets are required to make a prediction in funding trends which drives the need to collaborate with external providers.
- Current lack of in-depth expertise within the organization. Hence a need to engage AI experts to grow UN Environment’s capacity in AI through training and project consultancy.
- The visual modelling / algorithm tool has limitations for complex issues.
- Lack of systems infrastructure necessary for building and deploying these applications.
- There is potential to scale the application of the machine learning tools to similar projects to the use cases above within the organization.
- Staff members across the organization at UNEP have shown a great interest in understanding how AI can support their work and applying its capabilities in support of the environmental agenda.
- Acquisition of an expert from an established organization in the AI domain to facilitate AI capacity building, consultancy and training within the organization.
- Strategic partnerships with industry players with the relevant technical expertise, as well as earth -related big datasets. As the leading global environmental authority that sets the global environmental agenda, UNEP has the capacity to provide the partners with substantive knowledge related to the environment.