> For the complete documentation index, see [llms.txt](https://help.globalnaturewatch.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.globalnaturewatch.org/support/faqs.md). # FAQs
What types of data power Global Nature Watch (GNW) Horizon? GNW Horizon combines datasets from Land & Carbon Lab and Global Nature Watch (formerly Global Forest Watch) to provide a powerful, multi-layered view of environmental change — including shifts in ecosystem and land cover extent, vegetation cover, greenhouse gas dynamics and land disturbances over time — supporting more informed decisions around conservation, restoration and sustainable land management. The platform integrates both near real-time and annual satellite data, delivering key insights across diverse ecosystems — from forests to grasslands, croplands and natural lands. Refer to the [Datasets page](broken://pages/wVRZOXsGSih9NnMcZDft) for an up-to-date list of supported datasets. Looking ahead, we plan to integrate additional datasets that expand the platform’s scope — from more detailed carbon-related monitoring to deeper insights on grassland condition and productivity — providing a clearer picture of how and where our planet is changing.
How can I download datasets available on Global Nature Watch Horizon? You can download data by visiting the Land & Carbon Lab [website](https://landcarbonlab.org/data/), by visiting Global Nature Watch’s [Open Data Portal](https://data.globalforestwatch.org/), or [WRI's Data Explorer](https://www.wri.org/data).
Can users upload or analyze their own datasets within Global Nature Watch Horizon? Users can define a custom area of interest by drawing or uploading a polygon. However, the platform does not currently support uploading additional user-specific data layers for analysis.
How does Global Nature Watch Horizon protect user privacy and data? GNW Horizon protects user privacy and data by applying the same rigorous standards used across WRI's platforms, including Global Nature Watch (formerly Global Forest Watch). Please refer to the following privacy notices that govern GNW Horizon: * [Global Nature Watch AI Privacy Notice](/legal-notices/global-nature-watch-horizon-privacy-notice.md) * [WRI Privacy Policy](https://www.wri.org/about/privacy-policy) Users can request to delete their accounts by emailing .
How much energy does Global Nature Watch Horizon use, and what are you doing to minimize its environmental impact? We take the environmental impact of AI seriously and design GNW Horizon to be as efficient as possible, while constantly working to improve its performance. We minimize the footprint by: * **Using efficient models.** We use AI models that balance accuracy with efficiency — such as by reducing redundant processing by storing results of common questions — so we don’t use more computing power than necessary. * **Monitoring performance.** We consistently track how the platform is used so we can spot inefficiencies, improve speed and minimize unnecessary resource use — and be transparent about its operational impact. But what does this mean for the actual footprint? Some of our [best estimates](https://www.nature.com/articles/s41598-024-54271-x) are that one query emits around 2 to 3 grams of CO₂. To put that in perspective: * **Light use** (about 10 queries per week, \~500 per year) = \~1–1.5 kilograms CO₂ per year — about the same as driving a typical car 3–5 miles. * **Heavy use** (about 20 queries per day, \~7,000 per year) = \~14–20 kilograms CO₂ per year — about the same as a one-way economy flight of 100–150 km (e.g., Washington, DC, to Philadelphia). In contrast, CO₂ emissions from land use change (such as deforestation and conversion of other natural ecosystems for agriculture and other uses) are estimated at over 4 billion tonnes in 2024. GNW Horizon is designed to ultimately reduce waste, not create it — by enabling faster, smarter decisions about land, ecosystems and climate. We are confident that, over time, the environmental benefits it enables will outweigh the emissions from running the platform, and we are committed to driving that footprint down even further.
Will Global Nature Watch Horizon provide enhanced analysis or predictive modeling? Not at this time. GNW Horizon focuses on making existing monitoring data more accessible and actionable. While it may highlight trends or surface patterns, it does not currently generate predictive models.
What AI model is this built on, and how does it work? GNW Horizon uses a multi-model AI system rather than a single model. Different models work together behind the scenes, each handling specific tasks—like conversation, reasoning or retrieving relevant geospatial data. This approach allows us to use the best tool for each job, improving speed, accuracy and reliability. We continuously evaluate and update the models we use as the technology evolves. Learn more about models here: [Models](/get-started/models.md)
How accurate is Global Nature Watch Horizon? How does it avoid misinterpreting satellite data? GNW Horizon is designed to produce analysis grounded in open, peer-reviewed geospatial datasets from Land & Carbon Lab and Global Nature Watch (formerly Global Forest Watch). Rather than relying on general web information, it uses validated satellite data, established methodologies and documented metadata. To reduce the risk of misinterpretation, the system uses a structured approach to answering questions. When you submit a query, it doesn’t generate an answer directly—instead, it first identifies the relevant datasets, methods and geographic context, then executes the analysis using WRI-managed data services. This approach (sometimes called CodeAct[^1]) ensures that results are computed from source data, not inferred by a large language model alone. Responses are also linked to underlying documentation and metadata, so outputs can be traced back to their original data sources and methods. We continuously evaluate the system using real-world queries to ensure it selects the right datasets, applies appropriate safeguards and avoids misleading results. GNW Horizon is still experimental, so some responses may be incomplete or incorrect. We are actively improving the system and encourage users to share feedback if something doesn’t look right. Learn more here: [Capabilities & Datasets](/get-started/capabilities-and-datasets.md)
What measures are in place to prevent AI bias? Global Nature Watch Horizon produces answers from peer-reviewed data developed by Land & Carbon Lab and Global Nature Watch (formerly Global Forest Watch).. All methods are open and transparent, and available for anyone to examine. Every number, graph and map on GNW Horizon comes directly from our research data, not from general information on the internet. The AI agents are used to select the relevant datasets and run the appropriate pre-defined analysis, not to predict the data — removing risk of bias.
Will there be an API or integrations with other platforms like ArcGIS or Earth Engine? The current version of Global Nature Watch GNW Horizon is focused on direct user interaction and does not include external integrations or developer access. Broader interoperability may be revisited as the platform evolves.
Will Global Nature Watch Horizon be updated regularly? GNW Horizon is experimental and designed to evolve over time. As new AI models become available, we evaluate them internally to assess potential improvements in accuracy, speed and efficiency — allowing us to refine the assistant as technologies advance. On the data side, the assistant interfaces with APIs from across Land & Carbon Lab and Global Nature Watch (formerly Global Forest Watch). This enables it to access newly released datasets with minimal engineering lift, ensuring users benefit from the latest monitoring capabilities as they become available.
How can users with low bandwidth work with Global Nature Watch Horizon? While the AI experience requires a stable internet connection, GNW Horizon has been designed with accessibility in mind. Users on low bandwidth can still view data layers and explore visual insights even without engaging the AI assistant — providing a lighter version of the experience. Ensuring greater access in low-bandwidth contexts is a priority for the team, and we’re actively exploring ways to further optimize the platform for use in these environments.
How does Global Nature Watch Horizon handle sensitive data such as Indigenous territories? The data on Indigenous Peoples and local community lands and territories that are available on GNW Horizon are sourced from [LandMark](https://landmarkmap.org/map): The Global Platform of Indigenous and Local Community lands. The data on LandMark is voluntarily contributed by partner organizations under the terms of a data sharing agreement and/or are otherwise publicly available with licensing terms. GNW Horizon does not independently collect or alter this data. In addition, any sensitive information provided by users will be handled in line with WRI’s Privacy Policy, aiming to protect and promote the responsible use of data shared through the platform.
Is Global Nature Watch Horizon available in multiple languages? Yes, users will be able to interact with GNW Horizon in multiple languages. While the user interface and on-screen instructions will initially be in English, the tool will support multilingual queries so users can ask questions in other languages and receive relevant responses. This functionality will be powered by Gemini, enabling broad language coverage.
I ran out of prompts, what can I do? If you run out of prompts for the day, you can still manually browse and view data on the map. Learn more [here](/resources/access-and-usage-limits.md).
Does Global Nature Watch Horizon have a broader strategy on AI? Global Nature Watch Horizon aligns with WRI’s broader approach to responsible AI. WRI views AI as a transformative technology with the potential to accelerate progress for people, nature and climate — if it is developed and applied thoughtfully. Given the scale and urgency of today’s environmental challenges, innovative tools are needed, and AI can play an important role when used responsibly. [*Learn more about WRI’s perspective on AI here.*](https://www.wri.org/insights/ai-promise-and-peril-for-planet) That responsibility means grounding AI applications in peer-reviewed data, transparency and clear safeguards. GNW Horizon reflects this approach by using AI to make complex, peer-reviewed science easier to access and apply for decision-making, while keeping people at the center of how the technology is designed and improved. The platform is continuously evaluated and refined, with attention to accuracy, user feedback and the environmental costs of operating AI systems.
How does Global Nature Watch Horizon relate to Global Nature Watch (formerly Global Forest Watch)? Global Nature Watch (formerly Global Forest Watch) is WRI's core initiative for monitoring the world's forests. For more than a decade, it has provided free, open and globally consistent satellite data used by millions of people to understand and monitor forest change, and it's now expanding coverage to a broader range of ecosystems, including grasslands, agricultural lands, trees outside of forests and more — leveraging advances in AI, satellite science and geospatial technologies to do so. GNW Horizon is Global Nature Watch's AI-driven platform, currently in preview, where we're exploring new ways to make land and nature data easier to access and use through a conversational interface. Responses from GNW Horizon are backed by data from Global Nature Watch and Land & Carbon Lab.\ Over time, we expect GNW Horizon to become more deeply integrated into the broader Global Nature Watch website and user experience. The Global Nature Watch user community and GNW Horizon users are vital partners in shaping what that integration looks like.\ To learn more about this transition, read [Global Forest Watch is Becoming Global Nature Watch](https://www.globalforestwatch.org/blog/data-and-tools/gfw-now-global-nature-watch/).
How can I reference the information Global Nature Watch Horizon provided me? If you're unsure how to cite a specific result, you can ask GNW Horizon directly (e.g., “How can I cite this?”), and it will do its best to provide a recommended citation based on the data used or insight provided.
How long are conversations saved for? There isn’t a time limit for conversation storage. However, given GNW Horizon’s experimental nature, users should not expect conversations to be stored for longer than a year.
Can I access myGFW saved areas in Global Nature Watch Horizon? Areas created and saved in other WRI applications are not currently available in GNW Horizon but may be at a later stage.
Can I share my conversations and/or results with others? While entire conversations cannot be shared, each prompt response can be shared by clicking the “copy” icon at the bottom each response in your conversation.
[^1]: AI agent framework that enables Large LLMs to use programming (executable Python code) as a tool to solve problems, rather than relying solely on text or predefined JSON tools