INSANE - Innovative Spatial information products for forest Applications using new satellite technologies
Slovenian Forestry Institute
Funded by DAAD / 2020-2021
The INSANE project is located at the interface of remote sensing, forestry and ecology and aims to develop methods, workflows and concepts to use latest remote sensing technologies and ground reference and survey data to derive operational spatial information products for forest applications. With spatial information products, we here refer to geo-datasets derived from remote sensing data that can be integrated into existing forest information systems being used by forest practitioners.
As first step of the INSANE project, an online survey in form of an interactive questionnaire will be designed and distributed amongst forest experts in Slovenia and Germany. This online survey will collect information to compile a list of spatial information products desired by forest practitioners that could potentially be derived from existing remote sensing technologies. This list of spatial information products will contain detailed description of the minimum requirement that practitioners have on these products (e.g., accuracy, spatial detail, frequency of updates) and some concepts (to be developed within the project) how these requirements could be matched with the current state of the art in remote sensing, image processing and the new options provided by synthetic datasets, deep learning and machine learning algorithms.
Next, the complementary key-expertise of the two partners in Slovenia and Germany will be bundled to develop an exemplary forest information product, namely a tree species map for complete Slovenia and the Federal State of Baden-Württemberg based on freely available satellite data. Spatially explicit tree species information is still rare in most forest administrations and enterprises throughout the European Union (EU). While in some cases, particularly in state forests, stand-wise coarse estimates of tree species are available, in many smaller private forests, the species composition is unknown. Furthermore, even the information available in state forests is known to contain a comparably large uncertainty. The missing spatial detail within stands may furthermore negatively affect management decisions. Hence, a detailed tree species map at a fine spatial grain could be very useful for forest practitioners. In the INSANE project, we will develop such a tree species map based on multi-temporal Sentinel-2 data provided by the European Union’s Copernicus program and reference datasets provided by the state forest administrations of Slovenia and Baden-Württemberg. We will make use of machine learning, deep learning and cloud-computing approaches to derive the tree species map.
In summary, INSANE aims to support the ongoing development in remote sensing to progress from case-studies to operational approaches. The project will achieve this on the one hand side practically by developing a workflow to derive an operational tree species map and on the other hand theoretically by compiling a list of potential spatial information products desired by practitioners. We believe that these tasks could path the way to further collaborations focusing on the development of further spatial information products and the refinement of the workflow to derive the tree species map.