Multi-temporal satellite data maps burned areas in brazilian cerrado
A multi-temporal dataset for mapping burned areas in the Brazilian Cerrado using time series of remote sensing imagery
Machine LearningArtificial IntelligenceComputer Vision and Pattern Recognition
Summary
Wildfires in Brazil's Cerrado region are tracked using satellite pictures taken over time. The authors created a detailed dataset from these images to identify areas that are fully burned, partially burned, or not burned. They tested computer programs to classify these areas and found that considering partial burns separately improves accuracy. Their data and methods can help monitor fires better than some existing maps but still have some errors. The work shows how satellite sensors can be used directly for fire mapping without extra calibration between satellites.
What this means in practice
- •For environmental monitoring teams: Classify burned and partially burned areas in protected regions using multi-temporal satellite data from WFI sensors to improve wildfire impact assessments.
- •For forest management services: Use the provided dataset and classification approach to refine burned area detection in savanna ecosystems without requiring inter-satellite calibration.
Tested on one dataset.
Authors
Alisson Cleiton de Oliveira, Thales Sehn Körting
Abstract
This paper introduces a multi-temporal tabular dataset derived from satellite images to map burned areas in the Chapada dos Veadeiros National Park, in Goiás, Brazil, covering the years 2020 to 2022. The dataset contains blue, green, red, and near-infrared bands, as well as the BAI, EVI, GEMI, NDVI, and NDWI spectral indices from the WFI sensor on the CBERS-4A, CBERS-4, and AMAZONIA-1 satellites, organized into a regular grid. We applied the Random Forest classifier to develop and validate models based on samples labeled as totally burned, partially burned, and non-burned. Two classification approaches were tested: one combining burned and non-burned areas into binary classes and another distinguishing between totally burned (TB), partially burned (PB), and non-burned (NB) classes. Seven validation approaches assessed different post-classification combinations, focusing on accuracy, precision, recall, and intersection over union (IoU) metrics. Results showed higher IoU when TB, PB, and NB were used as individual classes and TB was reclassified as burned area (BA) while PB and NB were grouped as non-burned. Comparing the annual results of this approach to the MCD64A1 product, the errors of omission for the BA class were 22% in 2020, 28% in 2021 and 59% in 2022, while the errors of commission were 46%, 43% and 46%, respectively. The study highlights the utility of the WFI sensor for burned area mapping without inter-satellite spectral calibration and suggests further exploration with other machine learning algorithms to evaluate the dataset potential and limitations.