Skip to content

Conversions

View or edit on GitHub

This page is synchronized from trase/data/brazil/indicators/environmental/LULC/conversions/readme.md. Last modified on 2026-06-21 06:35 CEST by GitHub Actions. Please view or edit the original file there; changes should be reflected here after a midnight build (CET time), or manually triggering it with a GitHub action (link).

Brazilian Biomes Deforestation Consolidator (Google Earth Engine)

Overview

This repository contains a Google Earth Engine (GEE) JavaScript codebase designed to consolidate deforestation data from all Brazilian biomes into a single, comprehensive multi-band image.

The script covers the timeframe from 2003 to 2024 and integrates official data from different monitoring systems, prioritizing the highest quality data available for each biome. Furthermore, it segregates deforestation events into Forest and Non-Forest categories based on historical Land Use and Land Cover (LULC) data.

Data Sources

The script relies on two primary data sources: 1. PRODES (INPE): Official Brazilian government data, used as the primary source for the Amazon and Cerrado biomes. 2. MapBiomas (Collection 10): Used to track deforestation in the remaining biomes (Mata Atlântica, Caatinga, Pampa, and Pantanal) and to provide LULC context (Forest vs. Non-Forest) prior to the deforestation events.

Key Features & Methodology

1. Forest vs. Non-Forest Segregation

Deforestation pixels are divided into two distinct bands per year (e.g., def_forest_2020 and def_non_forest_2020). * For MapBiomas and Cerrado, the script looks at the LULC classification from the previous year (year - 1) to determine if the deforested pixel was originally a Forest Formation (Class 1) or a Non-Forest/Herbaceous Formation (Class 2). * For the Amazon, a specific PRODES Non-Forest (NF) Feature Collection is used to mask and reclassify the pixels.

2. Handling PRODES Cerrado Biennial Data

Before 2013, PRODES Cerrado released data biennially (every two years). The script handles this by automatically dividing the deforested pixel values evenly (using .divide(2)) between the reporting year and the preceding year, ensuring a continuous annual time series.

3. Mosaic Prioritization

When combining the data into a single national map, the script uses a mosaic approach with strict prioritization (the last element added overlays the previous ones): * Low Priority: MapBiomas (Background / Other biomes) * High Priority: PRODES Amazon

Usage & Setup

Running the Script

  1. Open the Google Earth Engine Code Editor.
  2. Copy and paste the script into the editor.
  3. Define your Region of Interest (ROI) geometry if you plan to export the data (the script assumes a variable named geometry exists for the export bounds).
  4. Uncomment the Execution and Export blocks at the very bottom of the script.
  5. Click Run.
  6. Go to the Tasks tab in the Code Editor to start the export process to your GEE Assets.

Output Structure

The final exported image is a single multi-band asset where each year is represented by two specific bands: * def_forest_YYYY: Pixels where deforestation occurred in forest vegetation during the year YYYY (Value = 1). * def_non_forest_YYYY: Pixels where deforestation occurred in non-forest natural vegetation during the year YYYY (Value = 1 or 0.5, depending on the source).

Metadata properties such as country, territory_type, and source are automatically attached to the final exported asset.

Author

Jailson S.