MapBiomas¶

MapBiomas publishes annual land-cover and land-use maps of Brazil. disscube.sources.mapbiomas reads them straight from the national GeoTIFFs on Google Cloud Storage: the files are tiled in 256 × 256 blocks, so only the pixels of the area of interest are fetched.

Dataset	Resolution	Years	Call with
Collection 11 (Landsat) — default	30 m	1985–2025	`collection=11, resolution=30`
Collection 4 of the 10 m series (Sentinel-2)	10 m	2017–2025	`collection=4, resolution=10`

For land-change modelling the 30 m collection is the natural choice: it has the long series needed for calibration and validation years, and the same resolution as the Landsat cubes of the Brazil Data Cube.

Registering a year¶

from disscube.sources.mapbiomas import register_mapbiomas_source

bbox = (-44.35, -2.62, -44.20, -2.47)          # WGS84
for year in (2000, 2020):
    register_mapbiomas_source(cube, f"lulc_{year}", year, bbox, out_dir="raw")
# raw/lulc_2020.tif               uint8 class codes, nodata 0, EPSG:4326
# raw/lulc_2020.provenance.json   dataset, collection, resolution, year, URL,
#                                 bbox, legend, license, checksum, software

Each year becomes a source with time=year, so derivations from several years load as a (time, y, x) series and go to DisSModel with to_raster_backend() (pip install "disscube[dissmodel]"):

from disscube import Derivation

for year in (2000, 2020):
    cube.derive_declarative(Derivation(target="landuse", source_id=f"lulc_{year}",
                                       operator="majority"), grid_id=grid.id)
    cube.derive_declarative(Derivation(target="urban_pct", source_id=f"lulc_{year}",
                                       operator="percentage", class_code=24), grid_id=grid.id)

backend = cube.to_raster_backend(["landuse", "urban_pct"], grid_id=grid.id)

Code 0 is nodata¶

The files declare no nodata value, but code 0 means "not observed" (for example open sea outside the mapped area). The reader treats 0 as nodata and writes the local file with nodata=0, so it never counts as a class in majority or percentage.

Level-1 classes¶

The leaf classes of the legend are grouped in five level-1 classes: Forest, Herbaceous and shrubby vegetation, Farming, Non vegetated area and Water. majority reads a map at that level with groups, in one derivation and without a reclassified copy of the map. For Collection 11 (legend of 13 August 2026):

[[derive]]
target = "l1_class"
source = "lulc_{year}"
operator = "majority"

[derive.params.groups]
"1" = [1, 3, 4, 5, 6, 7, 49]                                  # forest, savanna, mangrove, floodable forest, flooded savanna, wooded sandbank
"2" = [10, 11, 12, 29, 32, 50, 77, 84]                        # wetland, grassland, herbaceous and shrub formation, rocky outcrop, tidal flat, salt marsh, herbaceous sandbank
"3" = [9, 14, 15, 18, 19, 20, 21, 35, 36, 39, 40, 41, 46, 47, 48, 62]   # pasture, crops, forest plantation, mosaic of uses
"4" = [22, 23, 24, 25, 30, 75, 91]                            # beach and dune, urban, mining, photovoltaic, wind farm, other
"5" = [26, 31, 33]                                            # river, lake and ocean; aquaculture

A code that is in no group is ignored, as is nodata (0): a class missing from the table is a class that disappears from the map, so check the table against the legend of the collection in use. The share of a class, instead of the dominant one, is operator = "percentage" with class_code = [3, 4, 5, 6, 49] (a list is the share in any of them).

Classes¶

disscube.sources.mapbiomas.CLASSES names the classes most common in land-change studies (forest formation 3, savanna 4, mangrove 5, floodable forest 6, wetland 11, grassland 12, pasture 15, mosaic of uses 21, beach and dune 23, urban area 24, other non-vegetated 25, hypersaline tidal flat 32, water 33). The authoritative list is the legend of the collection in use: https://brasil.mapbiomas.org/en/codigos-de-legenda/.