Sustainable forestry solutions powered by satellite data

Forest burned areas detection

Assessment of burnt forest areas determines the location of burnt forest due to fires in a particular area.

Methodology

• Currently, it works in a semi-automated mode, but we are working on automating the solution further.
• EOSDA can analyze the area of interest, and with the help of various sources of satellite data, evaluate the burnt forest territories and classify the degree of damage.
• Different monitoring frequencies may be considered depending on the type of input data and the needs.

Inputs required

• Area of interest (AOI) in .kml or .shp format.

Expected output

• Raster or vector layers with marked burned forest areas (ha/acre, %).
• Analytical report.

Forest health analysis

Forest health analysis can include historical and/or current data about changes in forest health, based on remote sensing.

Methodology

• The current solution version is based on satellite images from Sentinel-2. Other satellites can be used, if required.
• EOSDA uses vegetation indices for the assessment of plants' health status and its changes (current imagery or time series of images and clustering by indices deviation values).
• Different monitoring frequencies may be considered depending on the type of satellite data and the needs.

Inputs required

• Area of interest (AOI) in .kml or .shp format.

Expected output

• Raster or vector layers with calculated indices or deviation indicating forest health changes (ha/acre, %).
• Analytical report.

Tree species classification

Tree species classification is a solution that allows the detection of different tree species in large areas of forest to optimize forest inventory management.

Methodology

EOSDA utilizes a multi-level Conv-LSTM architecture for crop detection. The model combines spatial (CNN) and temporal (LSTM) analysis. The current solution version is based on satellite images from Sentinel-2.

Inputs required

• Area of interest (AOI) in .kml or .shp format.
• Ground truth data (examples of polygons with the different tree species).

Expected output

• Raster or vector layers with the marked detected classes (species) and additional attribute data (examples).
• Analytical report.

EURD-related analysis

EUDR-related analysis is a solution developed to support companies in preparing their product for compliance with EUDR requirements (confirmation of no deforestation or forest degradation after 2020).

Methodology

EOSDA uses data from Sentinel-1/2, Landsat, or other satellites (including the EOS SAT-1 satellite). Detection is carried out based on the RGD analysis and the spectral characteristics of the forest (by vegetation indices).

Inputs required

• Area of interest (AOI) in .kml or .shp format.

Expected output

• Raster or vector layers with the marked fields and forest/deforestation.
• Raw satellite imagery used for analysis.
• Analytical report, part of EUDR compliance report (examples).