Using a data cube to monitor forest loss in the Amazon

Forests maintain an unlimited quantity of Earth’s terrestrial carbon and play an necessary function in offsetting anthropogenic emissions of fossil fuels. Since 2015, the world’s tropical forests may be noticed usually at an unprecedented six to 12 day interval because of the Copernicus Sentinel-1 mission.

Thousands and thousands of gigabytes of artificial aperture radar (SAR) information are acquired each day and evening, no matter cloud cover, haze, smoke or aerosols, permitting deforestation and forest degradation to be monitored not less than biweekly.

The problem, nonetheless, lies to find ample strategies to extract significant indicators of forest loss from the huge quantities of incoming radar information, such that anomalies within the time-series may be usually and constantly detected throughout tropical forests.

Such forest-monitoring strategies ought to be clear and simply comprehensible to the broader public, enabling confidence of their use throughout numerous private and non-private sectors.

The Sentinel-1 for Science: Amazonas venture presents a easy and clear strategy to utilizing Sentinel-1 satellite radar imagery to estimate forest loss. The venture makes use of a space-time information dice design (also referred to as StatCubes), the place statistical info related to determine deforestation is extracted at every level within the radar time-series.

With this strategy, the venture demonstrates using Sentinel-1 information to create a dynamic deforestation evaluation over the Amazon basin. The group had been in a position to detect forest lack of over 5.2 million hectares from 2017 to 2021, which is roughly the scale of Costa Rica.

Neha Hunka, distant sensing skilled at Gisat, commented, “What we’re seeing from space is over one million hectares of tropical moist forests disappearing every year within the Amazon basin, with the worst 12 months being 2021 in Brazil. We are able to observe these losses and report on them transparently and constantly each 12 days henceforth.”

Billions of pixels from the Sentinel-1 satellites from early-2015 to December 2021, every representing a 20 x 20 m of forest, are harmonized underneath the StatCubes design, and a easy thresholding strategy to detect forest loss is demonstrated within the first model of the outcomes.

The biggest problem within the venture was the huge quantity of information dealing with and processing. The group used a number of user-friendly software program instruments to entry the information effectively—processing over 450 TB of information to create the forest loss maps.

Anca Anghelea, open science platform engineer at ESA, added, “By offering open entry information and code via ESA’s Open Science Knowledge Catalog, and openEO Platform, we intention to allow researchers world wide to collaborate and contribute to the development of data about our international forests and the carbon cycle.

“Thus, within the final phase of the venture, a key focus shall be on Open Science, reproducibility, long-term upkeep and evolution of the outcomes achieved within the Sentinel-1 for Science: Amazonas Venture.”

Following on from the venture, the following aim is to attain a product of carbon loss from land cowl adjustments, working along with ESA’s Local weather Change Initiative group—a aim that may contribute to ESA’s Carbon Science Cluster.

The present outcomes of the venture at the moment are out there by clicking here. Sentinel-1 for Science Amazonas is carried out by a consortium of 4 companions—Gisat, Agresta, Norwegian College of Life Sciences and the Finnish Geospatial Analysis Institute. The group uniquely combines complementary and robust backgrounds in forestry and carbon assessments, multi-temporal SAR evaluation and information fusion, and large-data processing capabilities.