This study explores the application of unoccupied aerial vehicles (UAVs) and structure from motion (SfM) techniques for digital forest inventories, addressing key challenges in sustainable and cost-effective forest monitoring. UAV-SfM data products such as 3D point clouds and orthomosaics were generated from a study site in the Hainich National Park, Germany, to create a comprehensive digital representation of the forest structure, encompassing canopy, stems and ground components. Subsequently, this data was leveraged for the automated derivation of forest parameters such as tree stem position, individual tree crown delineation (ITCD), diameter at breast height (DBH), and coarse wood debris (CWD). The employed algorithms involve deep learning models (U-Nets), clustering techniques, and object-based methods.

This study investigates the integration of low-cost unoccupied aerial vehicles (UAVs) and community engagement in permafrost monitoring. Utilizing novel UAV flight patterns and crowdsourced data analysis, including a Convolutional Neural Network (CNN), the study enhances digital surface models (DSMs) for identifying ice-wedge polygons. Conducted in rapidly changing Arctic regions, it demonstrates an overall accuracy of 74.56% in feature detection. This approach offers improved resolution in environmental monitoring and suggests potential for broader application and rapid disaster response through community-sourced scientific analysis and consumer UAVs.

The Sentinel-1 (S-1) microwave measurements pose a unique opportunity for estimating forest parameters from satellite time series with increased data availability from overlapping orbits. Leveraging data from different orbits comes with varying viewing geometries and acquisition schedules, which can be considered in artificial neural networks.We adapt a transformer architecture for mapping the full S-1 data against median forest height values. By doing so, we propose per-orbit temporal encoders to handle different acquisition times by position, missing data by attention masking, and the addition of viewing geometry context.We show that our adjustments improve performance, with a greater impact of the additional data and a slight improvement in the masking. By optimizing the hyperparameters, our proposed method achieves a preliminary RMSE of 5.9m and an rRMSE of 35% in predicting the per-pixel vertical median forest height.

Microwave scattering from forests generates pixel geolocation shifts in Synthetic aperture radar (SAR) data that require an adequate representation within digital elevation models (DEM) for preprocessing. We analyze the impact of DEM properties on the radiometry and geolocation of radiometric terrain corrected (RTC) Copernicus Sentinel-1 imagery of forests to improve consistency in backscatter intensities for time series analyses. To account for the penetration depth of the C-Band sensor, we approximate the structure of stands in a temperate deciduous forest using height percentiles from Aerial Laser Scanning (ALS) point clouds in the Hainich National Park, Germany. Comparing the RTC results obtained using DEMs of SRTM, Copernicus, and ALS DEMs, the latter reduces topographically induced errors, resulting in visibly smaller effects from topography and spatially shifted information. Based on the P50 ALS vegetation elevation, results show homogeneous intensities within the same orbit and reduce variance from 2.4 dB2to 1.2 dB2in the difference of mid-range data from ascending and descending azimuth directions. Over forest, we observe lower intensities on sensor-facing and increased intensities on away-facing slopes and correlations with the illuminated pixel area (IPA) and local incidence angle. We reduce this bias with linear regressions of intensity on IPA. ALS DEMs in RTC and the proposed regression correction increase the consistency of images across orbits, measured by the inter-orbit range, throughout the selected year at our study site. We suggest the proposed method applies to other areas, requiring further testing under different forest types and topography.

This study explores the potential of Landsat-8 and Sentinel-2 imagery for annual grass biomass mapping in savannas. To this end, three wet season image mosaics based on Landsat-8 and Sentinel-2 were created for 2016, 2017 and 2018 over Kruger National Park (KNP), South Africa. For the purpose of calibration and validation, use was made of in situ fuel biomass values measured as part of the yearly veld condition assessment (VCA) in KNP. The satellite and reference data were fed into a random forests machine learning approach to make park-wide predictions of grass biomass and to assess the performance of Landsat-8 and Sentinel-2 predictors (i.e., surface reflectance and the normalized difference vegetation index, NDVI). Examples of the data sets used and biomass maps produced are provided together with the obtained error statistics. The latter suggest that wet season NDVI mosaics from Landsat-8 and Sentinel-2 data enable the creation of fairly reliable, annual maps of fuel biomass for KNP. These new biomass estimates represent a slight improvement over recent mapping efforts based on Sentinel-1 data [1].

In this paper, we propose a methodology for upscaling fractional vegetation cover (FVC) estimates derived from unmanned aerial vehicles (UAVs) to a larger area using freely available Sentinel-1/2 and Landsat-8 satellite data in the semi-arid Nama-Karoo biome of South Africa. To the best of our knowledge, such an approach is still lacking yet critical for understanding human-environment interactions, degradation, and the impacts of climate change in this vulnerable dryland ecosystem. The proposed approach utilizes ultra-high spatial resolution UAV imagery (i.e. <; 5 cm) to develop a high quality FVC product. The resulting product is then used to upscale and validate satellite-based estimates of FVC. The rationale for upscaling UAV estimates to the satellite-scale is not only to cover larger areas but also to exploit historical satellite time-series data, in an attempt to understand past trends and vegetation dynamics in the Nama Karoo-Grassland ecotone. The proposed method is expected to produce the first-ever high-resolution continuous maps of FVC and its changes in the above-mentioned ecosystem. Such information is of vital importance as it could help decision makers to gain a better understanding of the extent at which mechanisms such as bush encroachment, grassland expansion, or degradation are occurring in dryland ecosystems.

The Radar Vegetation Index (RVI) is widely applied to indicate vegetation cover. The index includes the backscattering intensities of co- and cross-polarization that do not only contain information coming from vegetation scattering at longer wavelength (L-band), but also from the soil underneath. A forward modelling approach using active and passive microwave-derived parameters to obtain the scattering contribution of the soil is pursued. The idea of this research study is a subtraction of the attenuated soil scattering contribution from the measured backscattering intensities, to provide a clean vegetation-based solution, called improved RVI (RVII). For latter analysis, the vegetation volume is forward modeled to calculate vegetation-only RVI-values without any soil scattering contribution. It reveals that, the pre-factor of the standard RVI leads to values up to 1.2, unfavorable for a normalized index running between zero and one. Hence, improvements for the standard RVI equation are proposed here to obtain a better suited value range and for incorporating soil scattering influences and filtering of regions with dominant soil scattering. Moreover, the improved RVI (RVII) is compared with datasets of vegetation and soil parameters (e.g. vegetation water content) for correlation analysis to find the physical parameters contributing to the index.

In this work we estimated Mexican forest aboveground biomass (AGB) using Synthetic Aperture Radar (SAR) and optical remote sensing data and two different reference data: (1) extensive national forest inventory (NFl) and (2) airborne Light Detection and Ranging (LiDAR) data. For the second modelling scenario, we applied a two-stage upscaling approach: firstly AGB for the LiDAR transects were estimated and used then to calibrate satellite imagery. Furthermore, we propagated uncertainties from field measurements to LiDAR-derived AGB and to the national wall-to-wall forest AGB maps. The estimated AGB maps (NFI- and LiDAR-calibrated) showed similar goodness-of-fit statistics compared to the independent validation dataset. However, the AGB map based on two-stage up-scaling method (i.e., from field AGB to LiDAR and from LiDAR-AGB to satellite imagery) showed much higher uncertainties compared to the traditional field to satellite up-scaling.

Today, very dense synthetic aperture radar (SAR) time series are available through the framework of the European Copernicus Programme. These time series require innovative processing and preprocessing approaches including novel speckle suppression algorithms. Here we propose an image transform for hypertemporal SAR image time stacks. This proposed image transform relies on the temporal patterns only, and therefore fully preserves the spatial resolution. Specifically, we explore the potential of empirical mode decomposition (EMD), a data-driven approach to decompose the temporal signal into components of different frequencies. Based on the assumption that the high-frequency components are corresponding to speckle, these effects can be isolated and removed. We assessed the speckle filtering performance of the transform using hypertemporal Sentinel-1 data acquired over central Germany comprising 53 scenes. We investigated speckle suppression, ratio images, and edge preservation. For the latter, a novel approach was developed. Our findings suggest that EMD features speckle suppression capabilities similar to that of the Quegan filter while preserving the original image resolution.

This paper proposes a method for quality assessment of synthetic aperture radar (SAR) scenes to facilitate selection of a master scene for persistent scatterer interferometry (PSI). Parameters like perpendicular and temporal baseline or weather conditions at acquisition time are evaluated using a scoring system. By defining scoring rules every available information can be used for rating image quality thus helping to make a quick decision to find the best possible scene in a large data stack. The method can be further improved by weighting the single results as needed. The approach is successfully tested on an exemplary data stack containing 57 SAR images of ERS.

In this paper, we present some investigations based on X-band interferometric coherence for retrieving forest Growing Stock Volume (GSV). Exanimating first temporal decorrelation, a comparison indicated total (γ=0.1-0.2), (γ=0.1-0.4) and low (γ=0.4-0.9) temporal decorrelation for TSX (11 days repeat pass), CSK (1 day repeat pass) and TDX (single pass) respectively. After studying also the spatial decorrelation, it could be pointed out that the increase of the perpendicular baseline enhanced the sensitivity to the volume of the forest canopies. CSK and TDX were further investigated in order to highlight their sensitivity to GSV. Volume decorrelation seemed to be dominant, as the coherence decreased with increasing vegetation. However, TDX seemed more suitable than CSK to retrieve GSV as it showed higher correlation (r2TDX=0.64, r2CSK=0.13).

With the enhancing availability and variety of space borne Synthetic Aperture Radar (SAR) data and a growing number of analysis algorithms the need for a vital user community is increasing. Therefore the German Aerospace Center (DLR) together with the Friedrich-Schiller-University Jena (FSU) and the Technical University Munich (TUM) launched the education initiative SAR-EDU. The aim of the project is to facilitate access to expert knowledge in the scientific field of radar remote sensing. Within this effort a web portal will be created to provide seminar material on SAR basics, methods and applications to support both, lecturers and students.

Numerous studies demonstrated the potential of interferometric coherence for forest stem volume estimation in boreal forests. Coherence derived from winter images acquired at frozen conditions proved being favourable against unfrozen conditions. This also applies to PALSAR coherence, although featuring relatively large temporal baseline of at least 46 days. However, when using data acquired at unfrozen conditions, a large spread of coherence was observed at all stem volume levels. This scatter negatively effects the correlation of stem volume and coherence, and thus prevents using summer coherence for stem volume estimation. Besides environmental conditions, the different tree geometries can be assumed having an impact on the magnitude of decorrelation. So far, this impact has rarely been investigated in boreal forests. This paper presents the results of a study investigating the impact of tree species on PALSAR coherence. The results show increased scatter of coherence in summer. They also show that the coherence of dense forest is larger in summer than in winter. Eventually, a slight impact of the tree species on coherence was observed. This impact is larger in summer.

For many aspects of the human life the world's forests are crucial. Only microwave remote sensing provides the means of all day weather independent monitoring of those pristine areas. Multi-temporal ENVISAT ASAR and ALOS PALSAR data for mapping boreal forest in central Siberia proved to be quite successful in diverse research projects. So far, TerraSAR-X High Resolution Spotlight images have been effectively used in the verification process of the resulting land cover area maps. Part of this knowledge will be transferred within the framework of the Remote Sensing Survey of the next global Forest Resources Assessment. Altogether 350 TerraSAR-X scenes all over the cloudy tropics will be analysed in a project called FRA-SAR 2010. The combined analysis of ALOS PALSAR, ENVISAT ASAR and TerraSAR-X on five areas inside the project will foster the expertise on forest structure parameters.

The availability of reliable land cover information is crucial for a wide range of applications, like for example monitoring of land use change and land degradation as well as administrative matters in global, regional and local scales. In this paper the potential of SENTINEL-1 C-band SAR data for land cover applications, e.g. generating level-2 land cover classification products has been investigated. Therefore, the planned short revisit and dual polarization concept of SENTINEL-1 has been simulated using multi-temporal ERS-2 and ENVISAT ASAR AP C-band backscatter intensity data. For classification, several multi-temporal metrics and the minimum amount of SAR data acquired during one growing season have been analyzed to derive five basic land cover classes with accuracies greater than 85%.

Focusing on Siberia, the feasibility of using Advanced Land Observing Satellite Phased Arrayed L-band Synthetic Aperture Radar (PALSAR) summer-intensity and winter-coherence images for large-area forest monitoring was investigated. Fine beam dual horizontal/horizontal and horizontal/vertical (polarization) intensity strip images were acquired during the summer of 2007. The processing consisted of radiometric calibration, orthorectification, and topographic normalization. The coherence was estimated from interferometric pairs with 46-day repeat-pass intervals. The pairs were acquired during the winters of 2006/2007 and 2007/2008. During both winters, suitable weather conditions that allow for low temporal decorrelation had been reported. By using PALSAR intensities and winter-coherence data, areas of forest and nonforest were separated. By combining both data types, a minimal overlap of the class signatures was observed, even though the analysis was conducted at the pixel level and no speckle filter was applied. The study concludes that the operational delineation of forest cover using PALSAR data is feasible. By applying a segmentation-based classification, an accuracy of 93% was obtained.