California 30x30 work in progress

About this project

Welcome to the 30x30 Biodiversity Assessment Geospatial Companion Tool. This tool is designed to explore and analyze conservation status and distribution of habitats and other important ecological features across the state to enhance knowledge of biodiversity conservation in support of 30x30. It is intended to be a companion tool to the California Biodiversity Network's (CBN's) 30x30 Biodiversity Assessment and draws upon datasets vetted by the CBN Biodiversity Assessment working group. It is currently based upon the 2025 Conservation Area data.

This tool reduces the technical barrier to data access by allowing plain-language queries (through an AI-LLM interface) for users to explore beyond the statewide assessment, for example, by local jurisdiction or managing agency. Users can explore the data through a web map itself or by asking questions such as:

• "What is the most under-represented habitat in the Central Coast ecoregion?"
• "Where are the most species rich areas that are unprotected in Merced county?"
• "What agency manages the most oak woodlands in lands that count toward 30x30?"

In 2020 California committed to conserving 30% of its lands and coastal waters by 2030 — the "30x30" goal — to protect biodiversity, expand access to nature, and build climate resilience. Meeting it depends on knowing what is already conserved, where biodiversity and habitat connectivity are concentrated, and how a changing climate is reshaping the landscape.

This is a work in progress — layers, models, and capabilities are being added and refined.

Testing, guardrails & your responsibility

The tool was developed by the Boettiger Lab at UC Berkeley, a part of the Eric and Wendy Schmidt Center for Data Science & Environment (DSE). While the tool has undergone initial testing and the developers have implemented guardrails to keep the tool from making value judgements and to stay focused on CBN vetted datasets, LLM interface tools are extremely difficult to comprehensively test because they can provide boundless queries. Feel free to explore the application and please let the developers know if you are seeing any inappropriate results (see proposed tool benchmarks).

The report this app accompanies

This app is the interactive companion appendix to the California Biodiversity Network's 2025 biodiversity assessment, published in August 2026. The report presents the analysis; this app lets you explore the same conservation, biodiversity, habitat and climate datasets behind it, ask your own questions of them, and take the numbers further than a printed table can go.

DSE at UC Berkeley is among the report's authors and contributed to its analysis, and this app follows the report's conventions — the same conserved-areas inventory, the same definitions of habitats and features, and the same measure of representation. Figures here are computed live from the source data, so they may differ slightly from the report's published tables where data vintages or methods differ; the assistant will always tell you which dataset and method produced a number.

To see the report's own tables for a particular place, use the 30x30 Assessment Explorer. Pick California, an ecoregion, a county, a congressional or legislative district, or draw your own area, and it shows land by conservation status and the representation of habitats, species richness, connectivity, freshwater and climate features, as in the report's Tables 3–13. It uses no AI model: every number comes from fixed SQL that you can read and export, and the statewide view sets each value beside the published one.

How it works

This app is powered by GLEN, DSE's open-source framework for AI-assisted geospatial exploration. Open datasets stream straight to your browser from cloud-native formats — no tile server in between — and an AI assistant answers questions by writing live queries against the same data. Full technical details are in the GLEN documentation.

Exploring the map

The map is built on MapLibre GL — the open-source engine behind the familiar Mapbox-style interactive maps. Scroll or pinch to zoom, drag to pan, and hold Ctrl and drag to tilt and rotate the view. Use the locate-me button in the top-left corner to jump the map to your current position.

The Overlays panel lists every data layer, organized into groups — conserved areas, biodiversity, connectivity, climate, and more. Toggle a layer on or off with its checkbox, and hover any feature for a tooltip of its live attributes — names, categories, ranks, acreages — straight from the source data. When layers overlap, use the send-to-back button (↩) in the Overlays header to drop the topmost layer beneath the others.

Asking the assistant

The Data Assistant understands plain English and does two kinds of things: it can drive the map for you, and it can answer analytical questions about the data. Just type what you want — if it can't do something, it will say so and usually suggest the closest thing it can.

Map operations happen right away on the layers already loaded — no hunting through menus:

• Filter a layer: "Show only GAP status 1 and 2 conserved lands."
• Recolor or restyle a layer: "Color conserved areas by managing agency", "Make the habitat layer 50% transparent."
• Fly somewhere: "Fly to the Sierra Nevada", "Zoom to San Diego County."
• Show or hide layers: "Turn on bird species richness", "Show the historic fire perimeters."

Analytical questions are answered by writing live spatial queries over the underlying data, returning a number or a downloadable table:

• "How many acres of California are conserved at GAP status 1 or 2?"
• "Which ecoregion has the most conserved acreage?"
• "Where is rare-species biodiversity highest relative to its ecoregion?"

Data layers & sources

Every layer draws from a public, primary source. Click "STAC" beside any entry to see the full machine-readable metadata — schemas, column descriptions, license, lineage, and download URLs.

Browse the full STAC catalog at radiantearth.github.io / stac-browser to explore additional datasets not yet enabled in the app.