GET TO KNOW THE STEPS OF THE MAPBIOMAS ALERT METHODOLOGY
Below you will find the main features, structure, and methodology of the MapBiomas Alert (v.1) to validate and refine deforestation alerts with high-resolution images in Brazilian biomes.
All alerts and respective deforestation reports are produced from supervised analysis and classification of Planet satellite images with 3 m resolution and daily frequency.
The whole process is done with extensive use of machine learning algorithms through the Google Earth Engine platform that offers tremendous processing power in the cloud.
The work is organized in tiles of 5 x 5 km. The definition of tiles to be analyzed is based on the deforestation alerts generated by the systems DETER / INPE (Amazon and Cerrado), SAD / IMAZON SimparSAR/Censipam(Amazon) and GLAD / University of Maryland (other biomes).
HOW WE ORGANIZE
To validate and refine alerts, teams of programmers, specialists in remote sensing, conservation and land use are organized into teams for each biome and experts in technology and system support.
OVERVIEW OF THE METHOD
The diagram below illustrates the main steps in the validation and refinement of deforestation alerts in Brazilian biomes.
SOURCE OF CROSS DATA
The alerts are crossed with land tenure and fiscalization information, such as indigenous territories, protected areas, settlements, Rural Environmental Registry (CAR) areas, including Permanent Preservation Area (APP) and Legal Reserve (RL), as well as areas of embargo, suppression authorizations and the forest management plan of Sinaflor from IBAMA. Alerts are also located in geographical boundaries such as municipalities, states, biomes, and watersheds.
Alerts can be viewed on different base maps (backgrounds) such as Google Earth images (non-associated date), Google Maps 2D, and Brazil's land cover and land use maps between 2012 and 2017 from MapBiomas (Collection 3.1).
This information qualifies the alerts and complements the reports with relevant information to the users.
|Layer||Year||Description||Reference||Site to download|
|Countries||2015||Map of Brazil (political boundary)||IBGE, 2015||ftp://geoftp.ibge.gov.br/organizacao_do_territorio/malhas_territoriais/malhas_municipais/municipio_2015/Brasil/BR/|
|States||2015||Map of the States of Brazil||IBGE, 2015||ftp://geoftp.ibge.gov.br/organizacao_do_territorio/malhas_territoriais/malhas_municipais/municipio_2015/Brasil/BR/|
|Municipalities||2015||Map of the Municipalities of Brazil||IBGE, 2015||ftp://geoftp.ibge.gov.br/organizacao_do_territorio/malhas_territoriais/malhas_municipais/municipio_2015/Brasil/BR/|
|Biomes||2016||Map of Brazilian biomes in 1: 5,000,000 (IBGE, 2004) refined RADAM vegetation map to scale 1: 1,000,000 and state boundaries in 1: 250,000 (IBGE, 2015)||MapBiomas, 2016||https://drive.google.com/open?id=0Byp5eRWoQ-PkWUpXNHRGMTdkcTg|
|Watersheds level 1||Map of watersheds level 1, scale 1: 1,000,000||ANA||http://dadosabertos.ana.gov.br/datasets?group_ids=084346aa5c18467782432f48bb687f83|
|Watersheds level 2||Map of the Indigenous Lands of Brazil||ANA||http://dadosabertos.ana.gov.br/datasets?group_ids=084346aa5c18467782432f48bb687f83|
|Indigenous Lands||2017||Mapa das Terras Indígenas do Brasil||
|Protected area||2019||Map of Conservation Units of Brazil||MMA, 2019||http://mapas.mma.gov.br/i3geo/ogc.htm|
|Map of settlements||INCRA, 2017||http://acervofundiario.incra.gov.br/acervo/acv.php|
|CAR||2018||Areas Rural Environmental Registry||Serviço Florestal, 2018||https://sistemas.florestal.gov.br/geoserver|
|Authorizations||2019||Permits for suppression and forest management||Sinaflor, SEMA-MT, SEMA-PA|
|Rural real estate - INCRA SIGEF||2019||Delimitation of rural real estate||
WHAT ARE THE PLANET IMAGES?
PlanetScope images come from a constellation of +130 satellites capable of daily recording the entire earth surface (equivalent to a collection capacity of 340 million km² / day) with A a spatial resolution of 3 m and four spectral bands.
For each alert, dozens of images are selected and activated in a 5 x 5 km cell. From this total, two images are selected, one before and other after deforestation, recording the satellite identifier, date and time of acquisition.