About this project
In progress 2019 - 2022
Purpose and goal
The project aims to streamline the planning of forestry measures by using satellite data with reference data from implemented forest clearing measures to map future needs. At the end of the project we will have achieved the following: 1. Developed and fully trained AI method. Implemented at service companies 2. Mapping method that can give good results over at least 90% of Sweden´s forest land (the evaluation needs to cover different types of forest). 3. Validated results from three forest companies with clear feedback on utility.
Expected results and effects
The proposed solution will generate benefits for Sweden´s forest owners by mapping provides increased clearing activity and better profitability in future timber businesses. Other benefits that the increased clearing activity entails are increased service sales to small and medium-sized companies, both for those who sell mapping services and forestry companies that offer clearing. The continuous collection of forest data may be combined with remote sensing and support AI strategies at the forest companies and at the same time create positive social effects.
Planned approach and implementation
Within the project, the following data sources are planned to be used: 1. Multispectral and multitemporal satellite data 2. Field data collected within the project 3. Actually harvested forests with age information 4. Images taken with drones for verification The combination of data sources is used in machine learning to achieve a graded mapping, but it is also central to handle the integration of neuro-symbols as well as a domain adaptation to handle the variations in input data. The models are validated both in terms of performance and operationally by participating forest companies.