Forest5dplus: An Open Benchmark Data Set for the Estimation of Forest Parameters from Sentinel-1 and -2 Time Series with Machine Learning Methods
Remote sensing has become a most versatile technique for the exhaustive information retrieval. There is neither one sensor nor one platform for all conceivable applications, but a multitude of possible sensing systems for each application. The project ‘Forest5Dplus’ – forest mapped in five dimensions plus labels – is supported by the German Space Agency within the German Aerospace Center (DLR) with funds from the German Federal Ministry for Economic Affairs and Energy.This study aims to create a syntactic training dataset from multitemporal Sentinel-1 and -2 satellite imagery from the European Space Agency (ESA) and to assign semantic labels to the individual elements derived from aerial imagery. Therefore, a newly developed method of multidimensional image data fusion on hypercomplex bases [1] is applied, enabling the space-saving, but information-preserving fusion of Sentinel-1 & Sentinel-2 in spectral, polarimetric and temporal dimensions. The project generates an increased information gain through data fusion of multimodal Earth observation data from the Sentinel-1 and Sentinel-2 sensors, heterogeneous data from different domains such as UAV surveys of forest areas, and the fusion of Earth observation data with geodata from field surveys such as forest inventories. A cross-domain test dataset for training and validating AI algorithms will be published soon.
Paper
Full text
Forest5dplus: An Open Benchmark Data Set for the Estimation of Forest Parameters from Sentinel-1 and -2 Time Series with Machine Learning Methods
Semantic Scholar · Environmental Science · 2023
Abstract
Remote sensing has become a most versatile technique for the exhaustive information retrieval. There is neither one sensor nor one platform for all conceivable applications, but a multitude of possible sensing systems for each application. The project ‘Forest5Dplus’ – forest mapped in five dimensions plus labels – is supported by the German Space Agency within the German Aerospace Center (DLR) with funds from the German Federal Ministry for Economic Affairs and Energy.This study aims to create a syntactic training dataset from multitemporal Sentinel-1 and -2 satellite imagery from the European Space Agency (ESA) and to assign semantic labels to the individual elements derived from aerial imagery. Therefore, a newly developed method of multidimensional image data fusion on hypercomplex bases [1] is applied, enabling the space-saving, but information-preserving fusion of Sentinel-1 & Sentinel-2 in spectral, polarimetric and temporal dimensions. The project generates an increased information gain through data fusion of multimodal Earth observation data from the Sentinel-1 and Sentinel-2 sensors, heterogeneous data from different domains such as UAV surveys of forest areas, and the fusion of Earth observation data with geodata from field surveys such as forest inventories. A cross-domain test dataset for training and validating AI algorithms will be published soon.