2022 SAG Award Winners

Swedish Forest Agency (SFA)

View Photo(s) of Award Presentation

Project Goal

The goal of this project has been to build a system that can use the benefits of the increased access to Sentinel-2 satellite images from the European Copernicus Program. When a new satellite image is available, we wanted a system that automatically performs a change detection analysis over areas where we have felling notifications. Other important goals were to build a generic system that can handle different types of areas to monitor, and in the future also be able to handle other sensors.
Learn more about us

Business Problem Solved

Administrators at the SFA have information on harvested areas soon after harvesting, which leads to more rapid follow up of notifications.
Forest owners gain feedback on harvested areas more rapidly.
The SFA gain better quality information on when harvesting occurs.
Fewer manual processing steps compared with previous methods give better quality in statistics.
Fewer manual processing steps gives the SFAs administrators more time to focus on qualitative feedback.
The system is built in a way that will enable the SFA to follow up on other types of changes in the landscape rather than just final fellings once a year.

Technology Implemented

The ArcGIS Image Server is the primary platform for the whole chain of events. The server holds the Sentinel-2 data from the daily orbits. However, during the dark winter (November through February) no data is downloaded. By using the server’s capacity to calculate statistics on the pixel values within a given polygon and combine that with the strength of chained raster functions, the SFA can get reliable median values from the pixels without too much interference from the clouds and cloud shadows.

The median values are calculated every time there are cloud free pixels within the polygon. These values are stored in a database and analysed as a time series. When a significant and steady change is detected, the Image Server computes a change detection image, thresholds it and calculates a polygon surrounding the area where the harvest took place.
Currently more than 50 000 polygons are active and included in the analysis every time there is a new Sentinel-2 image available.

Development Team Biography

Catarina Roempke, project leader
Joel Reisek, end user/problem owner
Jenny Olsson, end user/problem owner
Maria Strand, end user/problem owner
Anders Persson, remote sensing specialist
Johannes Berghällen, IT-architect
Stefan Wennborg, software developer/IT-architect
Wei Smedberg, test leader
Anton Lundkvist, software developer
Kjell Timalm, hardware technichian
Martin Nylander, statistics engineer
Morgan Ståhl, software developer/IT-architect
Arne Dahlman, GIS developer
Pär Nyman, GIS-specialist
Simon Johansson, data scientist