Deforestation (PRODES/DETER)

Overview

Field	Value
Provider	INPE — Instituto Nacional de Pesquisas Espaciais
Programs	PRODES (annual) and DETER (daily alerts)
Access	Public WFS API (TerraBrasilis GeoServer)
Format	CSV via WFS outputFormat + GeoJSON for geometry
Authentication	None
License	Federal government public data
Time Series	PRODES: 2000+, DETER: 2016+ (Amazonia), 2020+ (Cerrado)

Data Origin

INPE operates two complementary deforestation monitoring systems:

• PRODES: Consolidated annual mapping of clear-cut deforestation. Uses Landsat imagery (30m) to generate deforestation polygons with a minimum area of 6.25 hectares. Official result used by the federal government.

• DETER: Daily alert system for enforcement actions. Uses imagery from sensors such as CBERS-4, AMAZONIA-1 and Landsat with variable resolution. Detects deforestation, degradation, mining and burn scars.

Access via TerraBrasilis

Data is accessed via the TerraBrasilis GeoServer WFS with outputFormat=csv and filters via CQL_FILTER.

PRODES — Workspaces by Biome

Biome	Workspace	Layer
Amazonia	prodes-amazon-nb	yearly_deforestation_biome
Cerrado	prodes-cerrado-nb	yearly_deforestation
Caatinga	prodes-caatinga-nb	yearly_deforestation
Mata Atlantica	prodes-mata-atlantica-nb	yearly_deforestation
Pantanal	prodes-pantanal-nb	yearly_deforestation
Pampa	prodes-pampa-nb	yearly_deforestation

DETER — Workspaces by Biome

Biome	Workspace	Layer
Amazonia	deter-amz	deter_amz
Cerrado	deter-cerrado-nb	deter_cerrado

Geometry (prodes_geo)

The prodes_geo() function returns consolidated PRODES deforestation with geometry polygons as a GeoDataFrame.

Field	Value
Geometry column	geom (uniform across all 6 biomes)
Format	MultiPolygon EPSG:4326
maxFeatures default	10,000 (tabular: 50,000)
outputFormat	application/json (GeoJSON)

Geometry (deter_geo)

The deter_geo() function returns DETER alerts with geometry polygons as a GeoDataFrame.

Field	Value
Geometry column (AMZ)	geom
Geometry column (Cerrado)	st_multi
Format	MultiPolygon EPSG:4326
Volume per feature	~1.1 KB with geometry
maxFeatures default	10,000 (tabular: 50,000)
outputFormat	application/json (GeoJSON)

The geometry column is biome-specific in the GeoServer. The parser normalizes both to geometry in the output GeoDataFrame.

Biome Normalization

The bioma parameter accepts variants with/without accents and is case insensitive:

• "amazonia" or "amazônia" → "Amazônia"
• "cerrado" → "Cerrado"
• "mata atlantica" or "mata atlântica" → "Mata Atlântica"

Normalization is applied automatically in prodes(), prodes_geo(), deter() and deter_geo().

Usage Example

import agrobr

# PRODES — consolidated annual deforestation
df_prodes = await agrobr.desmatamento.prodes(
    bioma="Cerrado",
    ano=2022,
    uf="MT",
)

# DETER — real-time alerts
df_deter = await agrobr.desmatamento.deter(
    bioma="Amazônia",
    uf="PA",
    data_inicio="2024-01-01",
    data_fim="2024-06-30",
)

# With metadata
df, meta = await agrobr.desmatamento.prodes(
    bioma="Cerrado", ano=2022, return_meta=True
)
print(meta.records_count, meta.fetch_duration_ms)

# PRODES with geometry (requires pip install agrobr[geo])
gdf_prodes = await agrobr.desmatamento.prodes_geo(
    bioma="Cerrado",
    ano=2022,
    uf="MT",
)

# DETER with geometry (requires pip install agrobr[geo])
gdf = await agrobr.desmatamento.deter_geo(
    bioma="Amazônia",
    uf="PA",
    data_inicio="2024-01-01",
    data_fim="2024-06-30",
)

Limitations

• DETER only available for Amazonia and Cerrado
• WFS limits 50,000 features per request — filter by state and/or year (warning desmatamento_*_truncated when the cap is reached)
• The Source API (agrobr.desmatamento.*) returns individual polygons (fine granularity); the datasets.desmatamento dataset delivers the annual aggregate by uf/class/biome according to the contract
• After the BiomasBR migration (03/2026), the PRODES layers for Amazonia, Pantanal, Caatinga and Mata Atlantica are temporarily broken in the INPE GeoServer (ServiceException for any client); Cerrado and Pampa operational
• DETER is an alert system, not a consolidation one — there may be overlap
• prodes_geo() and deter_geo() return geometry (~10x more volume than tabular) — use filters to reduce data

Cache and Updating

• PRODES: TTL 24h (consolidated annual data, updated ~1x/year)
• DETER: TTL 24h (daily alerts, updated frequently)
• Recommended: use state and year filters to reduce data volume

Links

• TerraBrasilis
• PRODES
• DETER