Data sources

Community-maintained world map. Used as the default basemap for the analyst console and the source of place names, roads, and waterways under all overlay layers.

High-resolution satellite imagery composite. Useful for visual ground-truth when interpreting DETER alerts or PRODES polygons against actual canopy cover.

Global 10 m optical imagery with a 5-day revisit. Level-2A delivers surface reflectance atmospherically corrected, ready for vegetation indices and change detection. The Element84 Earth Search STAC catalog indexes every scene on AWS Open Data.

Near-real-time deforestation, degradation, and selective-logging alerts in the Brazilian Amazon, derived from CBERS-4/AWFI and AMAZÔNIA-1/WFI imagery. New alerts arrive within days of detection. The legal reference layer in Brazil for enforcement triggers.

DETER alerts adapted to the Cerrado biome — Brazil's tropical savanna, where deforestation rates have outpaced the Amazon since 2022. Distinct detection thresholds from the Amazon product because of the very different canopy structure.

Annual cumulative deforestation polygons in the Brazilian Amazon since 1988. PRODES is the official statistic for Brazil's NDC and REDD+ commitments. Where DETER triggers enforcement, PRODES is the legal accounting baseline for carbon credit projects.

Cerrado equivalent of PRODES Amazon — annual cumulative deforestation polygons. Critical for monitoring the agricultural frontier expansion in MATOPIBA and beyond.

Annual land use and land cover classification for all of Brazil from 1985 to 2023, at 30 m resolution. Classifies 62 distinct cover classes built from a Random Forest ensemble over Landsat archives. The longest consistent national LULC time series in the world — essential for carbon baseline calculations.

Global tree cover, loss year, and gain at 30 m. Built from Landsat archives by the Hansen lab at UMD. The reference dataset for global forest change since 2000 — used by Global Forest Watch and most carbon registries.

Official boundaries of Indigenous Lands at every demarcation phase: identified, delimited, declared, and homologated. Demarcation status is critical — land in the intermediate stages sits in a legal gray zone that drives speculative deforestation.

Federal-level Conservation Units (Unidades de Conservação) — national parks, ecological stations, biological reserves, sustainable-use reserves, and extractive reserves. Combined with INPE alerts, this layer flags illegal clearing inside protected areas.

Aggregate of state-managed conservation units (parks, environmental protection areas, extractive reserves under state authority). Coverage varies by state — PA, MT, MA, BA, and TO have the densest networks.

Brazilian state (UF) boundaries from the BC250 base cartography. The reference layer for any aggregation by state — alert counts, deforestation totals, conflict reports.

Six Brazilian biomes — Amazon, Cerrado, Caatinga, Mata Atlântica, Pampa, Pantanal. Defines the policy and ecological framing for every product muda-clima ships.

Daily downscaled CMIP6 projections at ~10 km resolution from 1950 to 2050. Wraps a large CMIP6 ensemble — we use EC_Earth3P_HR, MPI_ESM1_2_XR, and CMCC_CM2_VHR4. The engineering tradeoff: lower fidelity than running ERA5-bias-corrected CanDCS or INPE Eta on our own infrastructure, but free and zero-ops.

The canonical El Niño / La Niña monitoring indices: ONI, Niño 1+2/3/3.4/4, SOI, TNI, BEST, and MEI v2, monthly back to 1870 for the longest series. ENSO phase is the dominant interannual driver of Amazon and Cerrado drought and fire seasons — the signal that turns a climate trajectory into a near-term risk call.

Monthly global observation grids served over OPeNDAP: GPCP satellite-gauge precipitation (1979–), PREC/L gauge precipitation (1948–), GPCC gauge precipitation (1891–), and interpolated OLR as a deep-convection proxy (1974–2022). We slice a single grid cell per request — the observed rainfall record at any parcel, with no netCDF stack on our side.

The largest voluntary carbon-credit registry (VCS program) — project name, proponent, status, methodology, country, and estimated annual emission reductions for 5,000+ projects, including 300+ in Brazil. Backs project pre-screening: know what is already claimed in a landscape before originating or buying credits there.