Recent technological advances in remote sensing sensors and platforms, such as high-resolution satellite imagers or unmanned aerial vehicles (UAV), facilitate the availability of fine-grained earth ...observation data. Such data reveal vegetation canopies in high spatial detail. Efficient methods are needed to fully harness this unpreceded source of information for vegetation mapping. Deep learning algorithms such as Convolutional Neural Networks (CNN) are currently paving new avenues in the field of image analysis and computer vision. Using multiple datasets, we test a CNN-based segmentation approach (U-net) in combination with training data directly derived from visual interpretation of UAV-based high-resolution RGB imagery for fine-grained mapping of vegetation species and communities. We demonstrate that this approach indeed accurately segments and maps vegetation species and communities (at least 84% accuracy). The fact that we only used RGB imagery suggests that plant identification at very high spatial resolutions is facilitated through spatial patterns rather than spectral information. Accordingly, the presented approach is compatible with low-cost UAV systems that are easy to operate and thus applicable to a wide range of users.
Invasive plant species can pose major threats to biodiversity, ecosystem functioning and services. Satellite based remote sensing has evolved as an important technology to spatially map the ...occurrence of invasive species in space and time. With the new era of the Sentinel missions, Synthetic Aperture Radar (SAR) and multispectral data are now freely available and repeatedly acquired on a high spatial and temporal resolution for the entire globe. However, the high potential of such sensors for automatic mapping procedures cannot be fully harnessed without sufficient and appropriate reference data for model calibration. Reference data are commonly acquired in field surveys, which however, are often relatively expensive and affected by sampling and observer bias. Moreover, a direct transferability to the remote sensing perspective and scale is difficult. Accordingly, we firstly assess the potential of Unmanned Aerial Vehicles (UAV) for semi-automatic reference data acquisition on species cover of three woody invasive species Pinus radiata, Ulex europaeus and Acacia dealbata occurring in Chile. Secondly, we test the upscaling of the estimated species cover to the spatial scale of Sentinel-1 and Sentinel-2. The proposed workflow includes the visual sampling of respective canopies in UAV orthomosaics and the subsequent spatial extrapolations using MaxEnt with spectral (RGB, Hyperspectral), textural (2D) and canopy structural (3D) predictors derived from UAV-based photogrammetry. These UAV-based maps are then used to train random forest models with multitemporal Sentinel-1 and Sentinel-2 data to map the invasive species cover on large spatial scales. Our results show that the semi-automatic UAV-based mapping of the three invasive species results in accurate predictions. Depending on the predictor combination, the correlation was 0.70, 0.77 and 0.90 for Pinus radiatia, Ulex europaeus, Acacia dealbata, respectively. Among the three species, we observed clear differences in the model performance between the tested photogrammetric predictors and their combinations (spectral, 2D texture or 3D structure). For scaling up the UAV-based estimates to the satellite-scale, the Sentinel-2 data (multispectral) were more important than Sentinel-1 data (SAR). An independent validation revealed that the R2 of the upscaling accounted for 0.78 or higher for all species and RMSE lower than 12%. Our results hence demonstrate that UAV-based reference data acquisitions are a promising alternative to traditional field surveys if the target species are directly identifiable in the UAV data.
•We combine UAV and multitemporal Sentinel-1 and Sentinel-2 data to map three invasive species.•Spectral, textural and structural UAV-based predictors are compared.•Up-scaling from UAV to Sentinel-scale is very accurate.•UAV-based reference data is a valuable alternative to traditional field sampling.•Sentinel-2 (spectral) outperforms Sentinel-1 (SAR) for mapping invasive species.
•Tree species recognition is possible for both protected and managed stands.•We obtained relatively high results for all the stands.•Forest management regime influences tree species ...classification.•The higher stand mixture, the lower accuracy is obtained.
Information on tree species composition is crucial in forest management and can be obtained using remote sensing. While the topic has been addressed frequently over the last years, the remote sensing-based identification of tree species across wide and complex forest areas is still sparse in the literature. Our study presents a tree species classification of a large fraction of the Białowieża Forest in Poland covering 62 000 ha and being subject to diverse management regimes. Key objectives were to obtain an accurate tree species map and to examine if the prevalent management strategy influences the classification results. Tree species classification was conducted based on airborne hyperspectral HySpex data. We applied an iterative Support Vector Machine classification and obtained a thematic map of 7 individual tree species (birch, oak, hornbeam, lime, alder, pine, spruce) and an additional class containing other broadleaves. Generally, the more heterogeneous the area was, the more errors we observed in the classification results. Managed forests were classified more accurately than reserves. Our findings indicate that mapping dominant tree species with airborne hyperspectral data can be accomplished also over large areas and that forest management and its effects on forest structure has an influence on classification accuracies and should be actively considered when progressing towards operational mapping of tree species composition.
Aim
The spectral variability hypothesis (SVH) suggests a link between spectral variation and plant biodiversity. The underlying assumptions are that higher spectral variation in canopy reflectance ...(depending on scale) is caused by either (1) variation in habitats or linked vegetation types or plant communities with their specific optical community traits or (2) variation in the species themselves and their specific optical traits.
Methods
The SVH was examined in several empirical remote‐sensing case studies, which often report some correlation between spectral variation and biodiversity‐related variables (mostly plant species counts); however, the strength of the observed correlations varies between studies. In contrast, studies focussing on understanding the causal relationship between (plant) species counts and spectral variation remain scarce. Here, we discuss these causal relationships and support our perspectives through simulations and experimental data.
Results
We reveal that in many situations the spectral variation caused by species or functional traits is subtle in comparison to other factors such as seasonality and physiological status. Moreover, the degree of contrast in reflectance has little to do with the number but rather with the identity of the species or communities involved. Hence, spectral variability should not be expressed based on contrast but rather based on metrics expressing manifoldness. While we describe cases where a certain link between spectral variation and plant species diversity can be expected, we believe that as a scientific hypothesis (which suggests a general validity of this assumed relationship) the SVH is flawed and requires refinement.
Conclusions
To this end we call for more research examining the drivers of spectral variation in vegetation canopies and their link to plant species diversity and biodiversity in general. Such research will allow critically assessing under which conditions spectral variation is a useful indicator for biodiversity monitoring and how it could be integrated into monitoring networks.
The spectral variability hypothesis (SVH) suggests a link between spectral variation and (plant) species counts. Here, we reveal that in many situations this assumed link is weak, for example due to the influence of phenology or plant status on spectral variability. Further, we show that the chosen metric to calculate spectral variability also has a notable influence on the link.
•Sentinel-2 data were used to estimate woody cover in an extensive semi-arid area.•Only freely available data were used.•The workflow was fully implemented in open-source environment.•Our woody cover ...maps are notably more plausible than available global products.
Woody canopy cover is an essential variable to characterize and monitor vegetation health, carbon accumulation and land–atmosphere exchange processes. Remote sensing-based global woody and forest cover maps are available, yet with varying qualities. In arid and semi-arid areas, existing global products often underestimate the presence of woody cover due to the sparse woody cover and bright soil background. Case studies on smaller regions have shown that a combination of collected field data and medium-to-high resolution free satellite data (e.g., Landsat / Sentinel-2) can provide woody cover estimates with practically-sufficient accuracies. However, most earlier studies focused on comparably small regions and relied on costly field data. Here, we present a fully remote sensing-based work-flow to derive woody cover estimates over an area covering more than 0.5 million km2. The work-flow is showcased over the Zagros Mountains, a semi-arid mountain range covering western Iran, the northeast of Iraq and some smaller fraction of southeast Turkey. We use the Google Earth Engine to create homogeneous Sentinel-2 mosaics of the region using data from several years. These data are combined with reference woody cover values derived by a semi-automatic procedure from Google® and Bing® very high resolution (VHR) imagery. Several random forest (RF) models at different spatial grains were trained and at each grain validated with iterative splits of the reference data into training and validation sets (100 repetitions). Best results (considering the trade-off between model performance and spatial detail) were obtained for the model with 40 m spatial grain which showed stable relationships between the VHR-derived reference data and the Sentinel-2 based estimates of woody cover density. The model resulted in median values of coefficient of determination (R2) and RMSE of 0.67 and 0.11, respectively. Our work-flow is potentially also applicable to other arid and semi-arid regions and can contribute to improve currently available global woody cover products, which often perform poorly in semi-arid and arid regions. Comparisons between our woody cover products with common global woody or forest-cover products indicate a clear superiority of our approach. In future studies, these results may be further improved by taking into account regional differences in the drivers of woody-cover patterns along the environmental gradient of the Zagros area.
•Pre- and postfire UAS data were used to explain Sentinel-2-based dNBR values.•Consumed canopy cover is the main driver of variability in the dNBR values.•Cast shadows from dead trees may have a ...notable influence on dNBR.•UAS data enables a better understanding the meaning of common satellite products such as dNBR.
Characterizing the spatial variability of the severity of wildfires is important to assess ecological and economic consequences and to coordinate mitigation strategies. Vegetation indices such as the differenced Normalized Burn Ratio (dNBR) have become a standard tool to assess burn or fire severity across larger areas and are being used operationally. Despite the frequent application of dNBR-like vegetation indices, it is not yet fully understood which variables exactly drive the variability in dNBR observed by multispectral satellites. One reason for this is the lack of high quality prefire information about vegetation structure and composition. Consequently, the influence of prefire vegetation composition and other potentially influential variables such as cast shadows has hardly been examined. Here, we use very high resolution Unmanned Aerial System (UAS) orthoimages collected briefly before and after the large wildfires in Central Chile in the fire season 2016/2017 to derive variables related to the pre- and postfire landscape composition and structure. The variables are used as predictors in Generalized Additive Models (GAM) to explain the spatial variability in dNBR and RdNBR pixel values as observed by Sentinel-2. Our models explain more than 80% and 75% of the variability in dNBR and RdNBR values, respectively, using a sparse set of five predictors. The results suggest that in our study area the largest fraction of variability in Sentinel-2 based dNBR and RdNBR values can be explained by variables related to the fraction of consumed canopy cover while the vegetation composition before the fire does not have a large influence on dNBR and RdNBR.
Our results further suggest that cast-shadows of snags and standing dead trees with remaining crown structure have a notable influence on the dNBR signal which may have been underestimated so far. We conclude that spatially continuous, very high spatial resolution data from UAS can be a valuable data source for an improved understanding of the exact meaning of common vegetation index products, operationally used for monitoring the environment.
•UAV-based RGB-orthoimages can be used as alternative to traditional fieldwork.•Textural information improves land cover classification of UAV orthoimages.•UAV & medium resolution Sentinel-2 data can ...be used to detect invasive shrub species.•Spatial optimization approach detects spatial offsets between UAV and satellite data.•Multi-temporal satellite images covering the flowering stage are highly beneficial.
Mapping the occurrence patterns of invasive plant species and understanding their invasion dynamics is a crucial requirement for preventing further spread to so far unaffected regions. An established approach to map invasive species across large areas is based on the combination of satellite or aerial remote sensing data with ground truth data from fieldwork. Unmanned aerial vehicles (UAV, also referred to as unmanned aerial systems (UAS)) may represent an interesting and low-cost alternative to labor-intensive fieldwork. Despite the increasing use of UAVs in the field of remote sensing in the last years, operational methods to combine UAV and satellite data are still sparse. Here, we present a new methodological framework to estimate the fractional coverage (FC%) of the invasive shrub species Ulex europaeus (common gorse) on Chiloé Island (south-central Chile), based on ultra-high-resolution UAV images and a medium resolution intra-annual time-series of Sentinel-2. Our framework is based on three steps: 1) Land cover classification of the UAV orthoimages, 2) reduce the spatial shift between UAV-based land cover classification maps and Sentinel-2 imagery and 3) identify optimal satellite acquisition dates for estimating the actual distribution of Ulex europaeus.
In Step 2 we translate the challenging co-registration task between two datasets with very different spatial resolutions into an (machine learning) optimization problem where the UAV-based land cover classification maps obtained in Step 1 are systematically shifted against the satellite images. Based on several Random Forest (RF) models, an optimal fit between varying land cover fractions and the spectral information of Sentinel-2 is identified to correct the spatial offset between both datasets.
Considering the spatial shifts of the UAV orthoimages and using optimally timed Sentinel-2 acquisitions led to a significant improvement for the estimation of the current distribution of Ulex europaeus. Furthermore, we found that the Sentinel-2 acquisition from November (flowering time of Ulex europaeus) was particularly important in distinguishing Ulex europaeus from other plant species. Our mapping results could support local efforts in controlling Ulex europaeus. Furthermore, the proposed workflow should be transferable to other use cases where individual target species that are visually detectable in UAV imagery are considered. These findings confirm and underline the great potential of UAV-based groundtruth data for detecting invasive species.
Abiotic ecosystem properties together with plant species interaction create differences in structural and physiological traits among plant species. Certain plant traits cause a spatial and temporal ...variation in canopy reflectance that enables the differentiation of plant functional types, using earth observation data. However, it often remains unclear which traits drive the differences in reflectance between plant functional types, since the spectral regions in which electromagnetic radiation is influenced by certain plant traits are often overlapping. The present study aims to assess the relative (statistical) contributions of plant traits to the separability of plant functional groups using their reflectance. We apply the radiative transfer model PROSAIL to simulate optical canopy reflectance of 38 herbaceous plant species based on field‐measured traits such as leaf area index, leaf inclination distribution, chlorophyll content, carotenoid content, water and dry matter content. These traits of the selected grassland species were measured in an outdoor plant experiment. The 38 species differed in growth form and strategy types according to Grime′s CSR model and hence represented a broad range of plant functioning. We determined the relative (statistical) contribution of each plant trait for separating plant functional groups via reflectance. Therein we used a separation into growth forms, that is graminoids and herbs, and into CSR strategy types. Our results show that the relative contribution of plant traits to differentiate between the examined plant functional types (PFT) groups using canopy reflectance depends on the PFT scheme applied. Plant traits describing the canopy structure were more important in this regard than leaf traits. Accordingly, leaf area index (LAI) and leaf inclination showed consistently high importance for separating the examined PFT groups. This indicates that the role of canopy structure for spectrally differentiating PFT might have been underestimated.
Optical earth observation data is considered an important instrument for the mapping and monitoring of plant functioning. Here, the contribution of plant traits for differentiating plant functional groups is investigated using in‐situ trait measurements and radiative transfer models (PROSAIL). The results indicate that traits describing the canopy structure play an important role in differentiating plant functioning and might have been underestimated in the past. We found important spectral information for the separation of plant functional groups in varying and narrow wavelength regions. This outcome underlines the high potential of hyperspectral earth observation data to monitor plant functioning.
Display omitted
•Indicators of fluvial disturbance are derived from Landsat and Sentinel-2 imagery.•Field sampling plots are classified by plant species and habitat characteristics.•Disturbance ...indicators significantly differ among habitat classes.
In riverine ecosystems, flood disturbances govern the distribution and assembly of vegetation communities. However, anticipated connections between disturbance regimes and habitats often cannot be numerically described as the quantification of disturbances remains challenging. This work presents a novel approach to numerically characterize disturbance patterns in a spatially and temporally explicit way. For this, a dense time series of flood extents was created applying a semi-automated water detection approach to multispectral optical satellite data (Landsat 4, 5, 7, and 8; Sentinel-2) to an 8-km stretch of the alpine Lech River in Tirol, Austria. Twelve multitemporal metrics referred to as Flood Indicators (FIs) were derived and compared to habitat classes formed by vegetation communities as observed in field campaigns in July 2020 and 2021.
The FI values indicated high disturbance intensities for areas with early successional habitat classes, whereas little to no disturbance were related to late successional classes. This is in accordance with the presumed successional pathway of riverine vegetation communities. We observed numerous significant differences in FI values among early successional classes, supporting the assumption that flood disturbance is a dominant factor shaping the corresponding habitats. In late successional stages, the FI values differed less among the vegetation classes, pointing that other processes are more important in shaping the vegetation community in these stages.
The observed relationship between vegetation communities and the satellite-derived disturbance measures underlines the potential of the presented approach to characterize riverine disturbance regimes. Given the global availability of satellite data, the approach should be easily transferable to other study areas.