University of Twente Student Theses
The heat under our peat : using transfer learning for peat fire prediction in the arctic region of North-East Siberia with Sentinel-2 imagery
Schreurs, A. (2021) The heat under our peat : using transfer learning for peat fire prediction in the arctic region of North-East Siberia with Sentinel-2 imagery.
|
PDF
2MB |
| Abstract: | Over 90% of peatland can be found in boreal and subarctic regions. Unfortunately, these peatland areas might be more vulnerable to burning than previously assumed. Eventually northern peatlands could turn from carbon sinks to sources. New tools are necessary to respond to fire danger in peatland areas prone to burn. Remote sensing can view remote locations such as the Arctic regions of North-East Siberia. For peat fire prediction the combination of Sentinel-2 satellite imagery and the usage of a convolutional neural network (CNN) is applied. As labelled data in the field of remote sensing is not easily found and is costly, transfer learning will be applied. This leads to the main objective to explore the viability of peat fire prediction for the remote Arctic area of North-East Siberia, using Sentinel-2 imagery and transfer learning of a CNN network. The EuroSAT dataset is used to create a pretrained network closer to the target task of peat fire prediction and ResNet50 is chosen as the network architecture. The results show that predicting peat fire in the region of North-East Siberia with the help of transfer learning is a novel approach that is still far from being able to be used as an early warning in the form of, for example, a susceptibility map. Transfer learning cannot be applied with the approach and data in this thesis; however, it is not excluded as an option for peat fire prediction and needs further investigation towards which features are important and which approaches are best applicable per region and timespan. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Interaction Technology MSc (60030) |
| Link to this item: | https://purl.utwente.nl/essays/88821 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page