Rural properties (CAR) inside protected areas: selecting and measuring the overlap

Abstract

Rural properties from the Rural Environmental Registry (CAR), downloaded from the public SICAR service and overlaid with protected areas, which ones only touch the boundary, which ones are inside, and how much of each protected area is declared as a rural property. Removing cancelled records changes the conclusion.

Level Intermediate Reading 8 min Data October 2026 Programs ArcGIS

Brazil’s Rural Environmental Registry (CAR) is self-declared: the landholder draws the property. That is why there are properties inside protected areas, including strict-protection ones, where private property should not exist. Knowing where this happens helps prioritize land regularization and enforcement.

The script I used only selected the properties that touched the protected areas. Here you will also measure how much of each property and of each protected area overlaps, which separates the property that only touches the boundary from the one lying entirely inside.

Map of the Novo Progresso region. Protected areas have a green outline and CAR properties are colored polygons: active in blue, pending in orange, suspended in purple and cancelled in grey. Many properties lie inside Jamanxim National Forest and the Tapajós Environmental Protection Area. — CAR properties overlapping the protected areas of the study area, colored by registration status.

Where the CAR comes from

SICAR publishes the properties of each state in an open WFS service:

Address: https://geoserver.car.gov.br/geoserver/sicar/wfs
Layers: sicar:sicar_imoveis_pa, sicar:sicar_imoveis_mt etc. (one per state)

The most useful fields are cod_imovel (property code), status_imovel (AT active, PE pending, SU suspended, CA cancelled), tipo_imovel (IRU rural property, AST settlement, PCT traditional peoples and communities), condicao (review status) and area (declared area, in hectares). The articles’ script downloads the properties of the study area into the car layer (how to download).

Note

SIGEF (properties certified by INCRA, the land reform agency) works the same way: replace the car layer with the SIGEF one and the code below works unchanged. The files are in INCRA’s Land Archive (https://acervofundiario.incra.gov.br).

What you will do

Select the properties that touch the protected areas (what the old script did).
Cut each property by the protected areas and measure the overlapping area.
Compute the percentage of the property inside the protected area and drop those that only touch.
Summarize by protected area: number of properties by status and area of the protected area covered by CAR, without double counting properties that overlap each other.

The code

Names in Portuguese: toca = touches, dentro = inside, sobreposicao_ha = overlap in hectares, area_imovel_ha = property area, pct_imovel = percentage of the property, coberta = covered.

import arcpy

arcpy.env.overwriteOutput = True
gpkg = r"C:\dados\artigos.gpkg"
gdb = r"C:\dados\resultados.gdb"
ucs = arcpy.analysis.Select(gpkg + r"\main.ucs", r"memory\ucs", "regiao = 'Amazonia'")

# 1. Select by location (what the old script did)
car = arcpy.management.MakeFeatureLayer(gpkg + r"\main.car", "car")
arcpy.management.SelectLayerByLocation(car, "INTERSECT", ucs)
arcpy.management.CopyFeatures(car, gdb + r"\car_toca_ucs")

# 2 and 3. Intersection, overlapping area and percentage of the property
arcpy.management.CalculateGeometryAttributes(
    gdb + r"\car_toca_ucs", [["area_imovel_ha", "AREA_GEODESIC"]], area_unit="HECTARES")
car_ucs = arcpy.analysis.PairwiseIntersect([gdb + r"\car_toca_ucs", ucs], gdb + r"\car_ucs")
arcpy.management.CalculateGeometryAttributes(
    car_ucs, [["sobreposicao_ha", "AREA_GEODESIC"]], area_unit="HECTARES")
arcpy.management.CalculateField(car_ucs, "pct_imovel",
                                "100 * !sobreposicao_ha! / !area_imovel_ha!", "PYTHON3",
                                field_type="DOUBLE")

# 4. Summary by protected area and status (only properties with at least 1% inside)
dentro = arcpy.analysis.Select(car_ucs, r"memory\dentro", "pct_imovel >= 1")
arcpy.analysis.Statistics(dentro, gdb + r"\car_por_uc",
                          [["cod_imovel", "COUNT"], ["sobreposicao_ha", "SUM"]],
                          ["nome_uc", "status_imovel"])

# Area of the protected area covered by CAR, without counting overlaps between properties
coberta = arcpy.analysis.PairwiseDissolve(dentro, r"memory\coberta", ["nome_uc"])
arcpy.management.CalculateGeometryAttributes(
    coberta, [["coberta_ha", "AREA_GEODESIC"]], area_unit="HECTARES")

import processing
G = "C:/dados/artigos.gpkg"
AREA = "area(transform($geometry, 'EPSG:4674', 'ESRI:102033')) / 10000"

def rodar(alg, **p):
    p.setdefault("OUTPUT", "memory:")
    return processing.run(alg, p)["OUTPUT"]

ucs = rodar("native:extractbyexpression", INPUT=f"{G}|layername=ucs",
            EXPRESSION="regiao = 'Amazonia'")

# 1. Select by location (predicate 0 = intersects)
car = rodar("native:extractbylocation", INPUT=f"{G}|layername=car", INTERSECT=ucs,
            PREDICATE=[0])
print("properties touching protected areas:", car.featureCount())

# 2 and 3. Property area, intersection, overlapping area and percentage
car = rodar("native:fieldcalculator", INPUT=car, FIELD_NAME="area_imovel_ha",
            FIELD_TYPE=0, FIELD_PRECISION=4, FORMULA=AREA)
car_ucs = rodar("native:intersection", INPUT=car, OVERLAY=ucs,
                OVERLAY_FIELDS=["cnuc", "nome_uc", "grupo"])
car_ucs = rodar("native:fieldcalculator", INPUT=car_ucs, FIELD_NAME="sobreposicao_ha",
                FIELD_TYPE=0, FIELD_PRECISION=4, FORMULA=AREA)
car_ucs = rodar("native:fieldcalculator", INPUT=car_ucs, FIELD_NAME="pct_imovel",
                FIELD_TYPE=0, FIELD_PRECISION=2,
                FORMULA='100 * "sobreposicao_ha" / "area_imovel_ha"')

# 4. Summary by protected area and status (at least 1% inside)
dentro = rodar("native:extractbyexpression", INPUT=car_ucs, EXPRESSION='"pct_imovel" >= 1')
resumo = rodar("qgis:statisticsbycategories", INPUT=dentro, VALUES_FIELD_NAME="sobreposicao_ha",
               CATEGORIES_FIELD_NAME=["nome_uc", "status_imovel"])
for f in resumo.getFeatures():
    print(f["nome_uc"], f["status_imovel"], f["count"], round(f["sum"]))

# Area of the protected area covered by CAR, without counting overlaps between properties
coberta = rodar("native:dissolve", INPUT=dentro, FIELD=["nome_uc"])
coberta = rodar("native:fieldcalculator", INPUT=coberta, FIELD_NAME="coberta_ha",
                FIELD_TYPE=0, FIELD_PRECISION=1, FORMULA=AREA)
for f in coberta.getFeatures():
    print(f["nome_uc"], "covered by CAR:", round(f["coberta_ha"]), "ha")

-- 1 to 3. Properties touching protected areas, overlapping area and percentage of the property
DROP TABLE IF EXISTS artigos.car_ucs;
CREATE TABLE artigos.car_ucs AS
SELECT c.cod_imovel, c.status_imovel, c.tipo_imovel, u.cnuc, u.nome_uc, u.grupo,
       ST_Area(c.geom::geography) / 10000                         AS area_imovel_ha,
       ST_Area(ST_Intersection(c.geom, u.geom)::geography) / 10000 AS sobreposicao_ha,
       ST_Intersection(c.geom, u.geom)                            AS geom
FROM artigos.car c
JOIN artigos.ucs u ON u.regiao = 'Amazonia' AND ST_Intersects(c.geom, u.geom);

ALTER TABLE artigos.car_ucs ADD COLUMN pct_imovel numeric;
UPDATE artigos.car_ucs SET pct_imovel = 100 * sobreposicao_ha / area_imovel_ha;

-- 4. Summary by protected area (only properties with at least 1% inside)
SELECT nome_uc, grupo,
       count(*)                                         AS imoveis,
       count(*) FILTER (WHERE status_imovel = 'AT')     AS ativos,
       count(*) FILTER (WHERE status_imovel = 'PE')     AS pendentes,
       count(*) FILTER (WHERE status_imovel = 'SU')     AS suspensos,
       count(*) FILTER (WHERE status_imovel = 'CA')     AS cancelados,
       round(sum(sobreposicao_ha))                      AS soma_sobreposicao_ha,
       round((ST_Area(ST_Union(geom)::geography) / 10000)::numeric) AS coberta_ha
FROM artigos.car_ucs
WHERE pct_imovel >= 1
GROUP BY nome_uc, grupo
ORDER BY coberta_ha DESC;

import geopandas as gpd

G = "dados/artigos.gpkg"
ucs = gpd.read_file(G, layer="ucs").query("regiao == 'Amazonia'")
car = gpd.read_file(G, layer="car")
ALBERS = "ESRI:102033"                       # equal-area projection

# 1. Select by location
toca = car[car.intersects(ucs.union_all())].copy()
print("properties touching protected areas:", len(toca))

# 2 and 3. Intersection, overlapping area and percentage of the property
toca["area_imovel_ha"] = toca.to_crs(ALBERS).area / 10_000
car_ucs = gpd.overlay(toca, ucs[["cnuc", "nome_uc", "grupo", "geometry"]],
                      how="intersection", keep_geom_type=True)
car_ucs["sobreposicao_ha"] = car_ucs.to_crs(ALBERS).area / 10_000
car_ucs["pct_imovel"] = 100 * car_ucs["sobreposicao_ha"] / car_ucs["area_imovel_ha"]

# 4. Summary by protected area (at least 1% inside)
dentro = car_ucs[car_ucs["pct_imovel"] >= 1]
resumo = dentro.pivot_table(index="nome_uc", columns="status_imovel", values="cod_imovel",
                            aggfunc="count", fill_value=0)
resumo["soma_sobreposicao_ha"] = dentro.groupby("nome_uc")["sobreposicao_ha"].sum().round()
resumo["coberta_ha"] = (dentro.dissolve("nome_uc").to_crs(ALBERS).area / 10_000).round()
print(resumo.sort_values("coberta_ha", ascending=False))

Checking the result

The QGIS, PostGIS and Python versions gave the same property counts and the same areas (PostGIS differs by less than 0.01%). Since the CAR was only downloaded inside the study area, the percentages refer to the part of each protected area within the study area:

CAR properties overlapping the protected areas of the study area (SICAR, queried in October 2026).
Protected area	Group	Properties (≥ 1% inside)	Cancelled	Area covered by CAR (ha)	% of the area in the study area	% without cancelled
Jamanxim National Forest	Sustainable use	415	119	631,419	49%	37%
Tapajós Environmental Protection Area	Sustainable use	517	171	511,783	72%	48%
Iriri State Forest	Sustainable use	70	63	146,701	80%	4%
Jamanxim National Park	Strict protection	89	80	124,953	19%	2%
Rio Novo National Park	Strict protection	20	17	51,462	17%	2%
Altamira National Forest	Sustainable use	40	29	29,644	5%	1%

The “without cancelled” column changes the story: in Iriri State Forest, 63 of the 70 records have already been cancelled, and coverage drops from 80% to 4%. In the two National Parks (strict protection), 12 active, pending or suspended records remain.

Pitfalls

“Intersects” includes properties that only touch

Selection by location with intersects also picks the property drawn along the protected area boundary, with 0.01% of its area inside. In the study area, 1,196 properties touch the protected areas, but only 1,135 have at least 1% of their area inside. Measure the overlap before concluding that “there are N properties in the protected area”.

Properties overlapping each other

Because the CAR is self-declared, several properties cover the same piece of land. Adding up the overlap of each one gives more area than exists: compare the sum column with the covered (dissolved) area in the table above.

Cancelled records

Properties with status_imovel = 'CA' (cancelled) remain in the service. Decide whether they enter the analysis and say so in the report. In this article, they appear in a separate column.

Personal data

SICAR’s public service has neither the name nor the tax ID of the landholder, only the property code. Even so, avoid publishing lists of property codes with accusations: a CAR record overlapping a protected area is a lead for analysis, not a proven violation.

Updates

2026-10-29: first published, with data from INPE, MMA, IBAMA and SICAR from October 2026.

Versão em português

Reuse

CC BY 4.0

Citation

BibTeX citation:

@misc{alves2026,
  author = {Alves, Hendesson},
  publisher = {Hendesson Alves},
  title = {Rural Properties {(CAR)} Inside Protected Areas: Selecting
    and Measuring the Overlap},
  date = {2026-10-29},
  url = {https://hendesson.github.io/en/articles/08-car-protected-areas.html},
  doi = {10.5281/zenodo.23780260},
  langid = {en},
  abstract = {Rural properties from the Rural Environmental Registry
    (CAR), downloaded from the public SICAR service and overlaid with
    protected areas, which ones only touch the boundary, which ones are
    inside, and how much of each protected area is declared as a rural
    property. Removing cancelled records changes the conclusion.}
}

For attribution, please cite this work as:

Alves, Hendesson. 2026. “Rural Properties (CAR) Inside Protected Areas: Selecting and Measuring the Overlap.” In Geoprocessing Articles for Environmental Monitoring. Preprint, Hendesson Alves, October 29. https://doi.org/10.5281/zenodo.23780260.