UmaTerra
One Earth to all.
A set of open source geospatial analysis tools for understanding the interactions between land, climate and society in Brazil and other countries — land cover, biodiversity, indigenous territories and conservation units, rural cadastre and climate risk, all from the same satellite data and public records. This portal is not a commercial product, but rather a research and open source development initiative, with the goal of sharing knowledge, resources and technology for territorial and environmental management.
Automated case studies using the tools from this platform:
- Naturametrics
- Camposcope
- Yvynation
- climate-shift
- 27capitals read with one ruler165 INMET stations, across every region of the countryWhat we found
- Maximum temperature rises significantly in 19 of 27 capitals. São Luís falls. 7 show no clear trend.
- Fastest warming: Porto Alegre (+0.11) and São Paulo (+0.11) °C a year.
- Annual rain: falls significantly in 8 capitals and rises in 4.
Map of the capitalsHover over each capital to see its enlarged card, with what we found in each one.Open the map → - 🇨🇦West Kelowna, 2023: one fire through four platformsBurned forest, a regional park, Crown land and the climate ahead, in one story that runs through Naturametrics, Yvynation, Camposcope and climate-shift.Read the story →
- 🇧🇷Porquinhos and Bacurizinho: what happens to the land while an extension waitsA detailed case study: Bacurizinho as ratified in 1983, its extension declared in 2008 and neighbouring Porquinhos, and what changed inside them and around.Read the story →
- 🇧🇷Ruschi's museum: from a smallholding in Santa Teresa to the National Atlantic Forest InstituteOrchids traded for forest in 1939, a museum, a biological station and a federal reserve: four places seen from space from 1985 to 2025, and what GBIF holds for each, and locations it has.Read the story →
- Three parks, the same tools: Great Smoky, Gros Morne and IguaçuA national park in the United States, one in Canada and one in Brazil, of similar size, read with the same outline, the same tree-cover loss and the same GBIF counts: where the tools agree, where the parks differ, and why a count of records measures observation effort, not richness.Read the story →
- Three growing metros, Houston, Goiânia and Kelowna, and the climate they are growing intoThe same ruler in three countries: the urban footprint from 1975 to 2020, what the growth took, the heat island measured by satellite, and the heat, drought, fire and heavy-rain scenarios in climate-shift, with El Niño, an urban and a rural place in each city, and places drawn by Naturametrics.Read the story →
- 🇧🇷Campos Gerais: a national park and a state park in the same landscapeIn inland Paraná, a federal park created over farmland and a state park around a canyon, in satellite pictures from four eras, with each clearing in the colour of the president of the day, in Yvynation.Read the story →
- 🇧🇷Jamanxim and Martírios: a federal park and a Pará state park, term by termThe same governor stayed through the switch from Bolsonaro to Lula: on the BR-163 clearing rose and fell with the federal government, while on the Araguaia the state park's fire regime changed.Read the story →
- 🇧🇷Cerrado, 2024: the fire year, from the whole biome to the Ilha do BananalForty years of fire in the Cerrado from MapBiomas, the 2024 season month by month and class by class, and the Ilha do Bananal, which burned two years running, with the climate and El Niño in climate-shift.Read the story →
- 🇧🇷The Pantanal drying out: forty years of water, year by yearThe water of the planet's largest floodplain measured every year since 1985 by MapBiomas: the floods of before, the 2024 drought, the fires and what took the water's place, with the climate ahead in climate-shift.Read the story →
- 🇨🇦British Columbia, 2021: the atmospheric river that brought a lake backThe Sumas Prairie flood read by radar through the clouds, the rivers at their gauges on both sides of the border, the farms under water, and the climate and La Niña in climate-shift.Read the story →
- 🇨🇦The Okanagan's thinning winters: from Silver Star and Big White to GlacierThe Okanagan's snow on 1 April 2026 was the lowest since 1980, in a La Niña winter. The province's snow bulletin, a reanalysis, stations and satellite at Vernon, Kelowna and Glacier National Park, a test of snow near 0 °C, and an El Niño that may be the strongest ever measured.Read the story →
- 🇺🇸Palisades, 2025: a wildfire that crossed into the cityThe January 2025 Los Angeles fire measured from orbit: what burned, how much of it was town, what the land-cover map recorded and the climate ahead, in climate-shift.Read the story →
- 🇺🇸The Guadalupe, 4 July 2025: one night's flood, before and afterThe flood that went through Camp Mystic, Texas, read by the instruments: the rain from orbit, the river at three USGS gauges, the valley before and after from Sentinel-2, and the climate and El Niño in climate-shift.Read the story →
- 🇺🇸Katrina, 2005: the city under water and what the maps keptNew Orleans nine days after the levees failed, seen by Landsat 5, and what three records of water and land cover show or miss, with the climate ahead in climate-shift.Read the story →
- 🇺🇸Houston and Harvey: the city that grew over the prairie, and the rain of 2017Forty years of city and wetland in Harris County from the Annual NLCD, Harvey's rain, the bayous past their records and the water seen by radar, with the climate ahead in climate-shift.Read the story →
How these pages are made
The walkthroughs and stories are written with Claude Opus 5.5, an AI model by Anthropic, working inside the platforms' own code. The model picks the examples, runs the same analyses the apps run and writes from the numbers they return: no number comes from the model, and every one can be redone in the app. Along the way it also checks the code and tests the pages on phones; that is how we found and fixed an error that understated the areas on the Canada pages by a third.
Naturametrics, Camposcope and Yvynation each have a Brazil version and a Canada version, built on different data sources and cadastres. On this page they open in the Canada version, in English; the link under each panel opens the Brazil version. Naturametrics and Yvynation also have a United States version (Camposcope does not: there is no public national rural cadastre there).
Study interface: analysis point, MapBiomas layers and the National Forest Inventory.
Naturametrics🇧🇷🇨🇦🇺🇸🌐
Land use history and landscape analysis
Reconstructs land-use history and the state of the vegetation around any point on the map, in concentric radii up to 10 km.
- Canada
- AAFC Annual Crop Inventory (2009–2025), NTEMS forest stand age and Hansen forest loss, at radii from 1 to 10 km. The crop inventory is agricultural and stops at about 58° N.
- Brazil
- MapBiomas land cover (1985–2024), the National Forest Inventory, vegetation age and IBGE biomes, at radii from 0.5 to 10 km; drawn areas too.
- United States
- USGS Annual NLCD (1985–2025) across the lower 48 (Alaska has three snapshot years; Hawaii has no NLCD land cover), with a two-year swipe comparison on the map; disturbance history (LCMS), time since disturbance in place of stand age, fire (MTBS), tree canopy and Hansen forest loss, in radii from 0.5 to 10 km. Walkthrough on Paradise, California.
Who it's for
- Forestry & ecological research
- Land use analysis
- National Forest Inventory
Example use: Compare land cover between 1985 and 2024 within a 10 km radius around a farm.
Open Naturametrics (Canada) →Open Brazil in EnglishOpen the United StatesOpen the world (any point on land)See it working, step by step: one point in the forest above Kelowna · one point in Paradise, California
Search by CAR code, coordinate, or municipality, with CAR and IBGE biome layers.
Camposcope🇧🇷🇨🇦
Rural properties in public cadastres
Finds a rural property in the public cadastre and analyses land cover, forest loss and fire inside it and in rings out to 5 km. It describes what the data show and issues no verdict.
- Canada
- British Columbia only: ParcelMap BC parcels, searched by PID, coordinate or municipality; AAFC crop inventory, NTEMS stand age, biomass and MODIS burned area. Clicks off a parcel fall back to protected areas.
- Brazil
- Brazil's Rural Environmental Registry (CAR): search by CAR code, coordinate or municipality; MapBiomas, 2008 SPOT imagery, and the series split at the property's registration date.
Who it's for
- Rural landowners
- Communities
- General public
Example use: See how much forest was lost around a property before and after its registration.
Open Camposcope (Canada) →Open Brazil in EnglishSee it working, step by step: one parcel in Kelowna
Home panel, with the four available analysis paths.
Yvynation🇧🇷🇨🇦🇺🇸
Indigenous lands, conservation units and protected areas
Combines satellite imagery, land-cover data and forest-loss detection (Hansen/GLAD) to analyse land use and ecosystem dynamics in protected territories, with maps, charts and reports.
- Canada
- Protected and conserved areas from the federal CPCAD register, starting with British Columbia: AAFC crop inventory, VLCE2 land cover (1984–2022), Hansen forest loss, forest age, biomass and fire; areas made of several parts can be analysed part by part. Canadian Indigenous lands are not analysed, in keeping with the First Nations principles of OCAP®.
- Brazil
- Indigenous lands (FUNAI) and conservation units: MapBiomas, Hansen/GLAD forest loss and land-cover transitions; drawn areas too, and batch processing of many areas at once.
- United States
- PAD-US protected areas (USGS): national parks, forests, refuges and state and local areas, plus tribal lands — these under a notice that the numbers describe land cover and that Yvynation does not speak for the peoples. NLCD land cover (1985–2025), Hansen forest loss, biomass and fire, inside the area and in rings around it, with the PAD-US overview layer on the map.
Who it's for
- Indigenous land managers
- Conservation units
- Researchers and journalists
Example use: Follow forest loss year by year inside a protected area and in the land around it, and export a report.
Open Yvynation (Canada) →Open Brazil in EnglishOpen the United StatesSee it working, step by step: an Indigenous land · a park near Kelowna · an area you draw · many areas in one batch
Impact analysis: the circle, the climate model's cell and the exposure story.
climate-shift🇧🇷🇨🇦🇺🇸🌐
Climate exposure of land use, point by point
Click anywhere in Brazil, Canada or the United States (or on any land in the world, on the World page): climate-shift reads the land use around it (MapBiomas, AAFC, NLCD), the terrain (fire, slope, low ground) and measurements (drought, rain, gusts) and crosses them with nine CMIP6 climate models to show which area is exposed to more heat, drought and heavy rain in 2041–2070. Deforestation alerts, ENSO and Sentinel-2 imagery remain as supporting tools.
Who it's for
- Governments & policy
- Conservation efforts
- Communities & NGOs
- Researchers and journalists
Example use: See how many hectares of cropland and forest within 10 km face more days above 35 °C and longer dry spells under SSP2-4.5.
Open climate-shift (Canada) →Open Brazil in EnglishOpen the United StatesOpen the world (any point on land)See it working, step by step: the climate exposure of a place
Support UmaTerra
All four tools are open and free: no ads, no sign-up, no data sold. The code is open source and the sources are public — MapBiomas, Hansen/GLAD, Copernicus, IBGE, INMET and SICAR. What keeps the project running is volunteered time and a cloud bill.
Cost transparency
- Compute for the four tools (Cloud Run)
- ~US$ 34/mo
- Database, container images and storage
- ~US$ 18/mo
- Google Earth Engine — 350 to 700 EECU-hours
- ~US$ 150–300/mo
- True cost of running it
- ~US$ 200–350/mo
None of this is billed today: the infrastructure runs on Google Cloud Research Credits and Earth Engine is used under a noncommercial-use registration, within the free quota. Institutional support is what carries the project when those programmes change — and what keeps the tools free, open and ad-free.
What would this analysis cost?
Open source doesn't mean cost-free.
UmaTerra's tools are free to use. Below is an estimate of what the same processing would cost if you had to pay for it — the value the platform absorbs to keep research and knowledge accessible.
- Earth Engine processing
- Equivalent Earth Engine cost
- Equivalent server cost
- Covered by the platform
- You pay
- US$ 0
An estimate, not an invoice. Derived from the platform's own measured usage: US$ 0.42 per Earth Engine EECU-hour and US$ 0.24 per server hour, assuming roughly 4 analyses per hour of use and a per-analysis cost that grows with area. Longer time series and larger areas consume more.
Ways to help
Institutional support
Universities, NGOs, institutes and public agencies can back the project with an annual contribution. Access stays free and open to everyone — supporters are credited on this page and help set what comes next.
Talk about institutional support →Individual support
Recurring contributions help cover the infrastructure once the research credits run out.
GitHub Sponsors →Cite and credit
If you use data, charts or figures in research, teaching or reporting, cite the platform and the original sources. Each tool ships ready-to-use citation text.
Contribute code
Fixes, translations, bug reports and new analyses are welcome in the open repositories.
Repositories on GitHub →
Supported by
- Google Cloud Research CreditsCloud infrastructure
- Google Earth EngineNoncommercial-use registration
- UBC graduate study grantsResearch funding
- Climate Action and Awareness Fund (Canada)Research funding
- inma.gov.brTechnical validation & feedback
- mgi.gov.brTechnical validation & feedback
- florestal.gov.brTechnical validation & feedback
- MapBiomasLand use and cover data
- GLAD / Global Forest Watch (University of Maryland)Land use and cover data
Institutional supporters are credited here.