The Climate Policy Radar app is the world’s largest open knowledge-base for climate law and policy. Our app allows you to search and explore the full text of 12,000+ climate laws and policies, Nationally Determined Contributions (NDCs), Biennial Transparency Reports (BTRs), corporate climate disclosures and more.
This page provides information about the scope and structure of our database, our data collection methods and terminology, updates to the database and planned future developments.
As we expand and evolve, this methodology document will evolve to reflect developments.
See the list of data sources in our terms & conditions page
Our dataset contains documents published in many different languages. English-language searches will return auto-translated results from documents in all languages, helping bridge critical information gaps and blindspots. In the future, our user interface will be available in French, Portuguese & all other UN languages.
The database is continually monitored and updated by teams of experts at the LSE Grantham Research Institute and Climate Policy Radar, in order to ensure accuracy and to reflect the latest developments in climate law and policy. These updates are drawn from official sources such as government websites and parliamentary records, as well as UNFCCC and related websites, where applicable.
In the Climate Policy Radar database, documents are grouped together with supplementary documents that provide additional information, such as annexes, amendments and press releases, as well as any translations or versions of the document published in other languages.
We also indicate wider relationships between documents. For example, we indicate where documents belong to an overarching policy programme (such as the European Union’s Common Agricultural Policy) or publication cycle (such as Nationally Determined Contributions).
Documents in the database are assigned attributes (also called metadata) in order to provide information about the contents of documents and to support searchability.
Document metadata applies to entire documents and is manually assigned. Most document metadata and definitions used were developed by the LSE Grantham Institute. For the full list of attributes see our data dictionary. The table below provides some example metadata used in the Climate Policy Radar database and app.
| Attribute | Definition |
|---|---|
| Jurisdiction | A document’s jurisdiction indicates where it was published. Jurisdictions include all parties to the UN Framework Convention on Climate Change (UNFCCC) — 196 countries plus the European Union — as well as a number of territories which are not party to the UN or UNFCCC, namely Taiwan, Palestine and Western Sahara. |
| Region | Jurisdictions are grouped into regions of the world, as defined by the World Bank. |
| Year | Refers to the year in which a document was first published. Note - this may not be year of the most recent update or amendment to a document. |
Our knowledge graph maps key concepts in climate policy, the relationships between them, and their presence in climate policy and other related documents. This enables users to discover important topics, and identify connections between them. In the future, this will be used to improve our other applications like search and generative AI tools. Read the knowledge graph methodology for more information.
The principal limitations to our dataset follow from those of our third-party data providers.
We currently update our database manually, based on monitoring of developments in climate law and policy internationally. We are developing the capacity to more efficiently capture data from additional sources, including via scraping, in order to increase the pace and scope of the expansion of our dataset.
We also currently rely on data labelled manually by domain experts; this limits the pace at which it is possible for new data to be incorporated into our dataset.
When you enter a search term and enable ‘related phrases’ search, the Climate Policy Radar app uses natural language processing (NLP) to help you find relevant content without needing to use exact keywords. This is useful because many concepts can be described in different ways. For example, ‘internal combustion engine’, ‘internal combustion engine vehicle’, ‘ICEV’, ‘fossil fuel car’, and ‘gasoline car’ all refer to vehicles commonly found on today’s roads. With natural language search, you can use everyday language, and the tool will recognise related terms, delivering more comprehensive and relevant search results.
We use a machine learning method called dense retrieval alongside a fuzzy string search to perform natural language search. You can find more details in our blog post.
The model we use for dense retrieval inherits biases from both its base model, DistilBERT, and the search query dataset it was trained on, MSMARCO. It’s trained on relatively short passages of text (average 60 words in length), so may struggle with queries longer than this.
At the moment we use a general purpose model that has been trained on English Wikipedia and BookCorpus. This means it may misinterpret some climate- or policy-specific concepts, and is something that we look to improve in future.
For queries we’ve identified that could uncover harmful biases in the underlying semantic search model, semantic search falls back to using fuzzy string search only,which relies on matching words rather than returning related terms.
We work collaboratively with our expert network of partners and users to ensure our app is accurate and up to date. Should you encounter any errors in the data, we encourage you to get in touch with us at support@climatepolicyradar.org.
This map was made with the Admin o - Countries map package by Natural Earth. Climate Policy Radar’s usage of this World map does not represent an opinion on any disputed boundaries.
Climate Policy Radar contains laws and policies from the Climate Change Laws of the World dataset. Please refer to the Climate Change Laws of the World methodology to read more about this dataset.
Climate Policy Radar contains publicly available guidance documents and project documentation (such as concept notes, full proposals, implementation reports) from these Multilateral Climate Funds: GCF, GEF, CIF and AF. Please refer to the Climate Project Explorer methodology to read more about this dataset.
Climate Policy Radar now includes 900 regulatory filings, climate plans, and integrated reports published by more than 450 companies worldwide. Please refer to the Corporate Disclosures Pilot Methodology to read more about this dataset.