Spaceborne remote sensing can track ecosystems changes thanks to continuous and systematic coverage at short revisit intervals. Active remote sensing from synthetic aperture radar (SAR) sensors allows day and night imaging as they are not affected by cloud cover and solar illumination and can capture unique information about its targets. However, SAR observations are affected by the coupled effect of viewing geometry and terrain topography. The study aims to assess the impact of global digital elevation models (DEMs) on the normalization of Sentinel-1 backscattered intensity and interferometric coherence. For each DEM, we analyzed the difference between orbit tracks, the difference with results obtained with a high-resolution local DEM, and the impact on land cover classification. Tests were carried out at two sites located in mountainous regions in Romania and Spain using the SRTM (Shuttle Radar Topography Mission, 30 m), AW3D (ALOS (Advanced Land Observation Satellite) World 3D, 30 m), TanDEM-X (12.5, 30, 90 m), and Spain national ALS (aerial laser scanning) based DEM (5 m resolution). The TanDEM-X DEM was the global DEM most suitable for topographic normalization, since it provided the smallest differences between orbital tracks, up to 3.5 dB smaller than with other DEMs for peak landform, and 1.4–1.9 dB for pit and valley landforms.
Dehesas are high value agroecosystems that benefit from the effect tree cover has on pastures. Such effect occurs when tree cover is incomplete and homogeneous. Tree cover may be characterized from field data or through visual interpretation of remote sensing data, both time-consuming tasks. An alternative is the extraction of tree cover from aerial imagery using automated methods, on spectral derivate products (i.e. NDVI) or LiDAR point clouds. This study focuses on assessing and comparing methods for tree cover estimation from high resolution orthophotos and airborne laser scanning (ALS). RGB image processing based on thresholding of the 'Excess Green minus Excess Red' index with the Otsu method produced acceptable results (80%), lower than that obtained by thresholding the digital canopy model obtained from the ALS data (87%) or when combining RGB and LiDAR data (87.5%). The RGB information was found to be useful for tree delineation, although very vulnerable to confusion with the grass or shrubs. The ALS based extraction suffered for less confusion as it differentiated between trees and the remaining vegetation using the height. These results show that analysis of historical orthophotographs may be successfully used to evaluate the effects of management changes while LiDAR data may provide a substantial increase in the accuracy for the latter period. Combining RGB and Lidar data did not result in significant improvements over using LIDAR data alone.
Métodos para la estimación de la cabida cubierta a partir de ortofotografías de alta resolución y LiDAR aeroportado en dehesas españolasResumen: Las dehesas son agroecosistemas de alto valor que se benefician del efecto de la cobertura arbórea sobre el pasto. Este efecto facilitador aparece cuando la cobertura arbolada es incompleta y homogénea. La cobertura arbórea puede caracterizarse con datos de campo o mediante fotointerpretación de datos de teledetección, ambas tareas que requieren mucho tiempo. Una alternativa es extraer la cobertura arbórea a partir de imagen aérea, derivados espectrales (i.e. NDVI) o nubes de puntos LiDAR. Este estudio se centra en evaluar y comparar métodos para la estimación de cobertura arbolada a partir de ortofotografías de alta resolución y LiDAR aeroportado (ALS).To cite this article: Borlaf-Mena, I., Tanase, M.A., Gómez-Sal, A. 2019. Methods for tree cover extraction from high resolution orthophotos and airborne lidar scanning in Spanish dehesas. Revista de Teledetección, 53,[17][18][19][20][21][22][23][24][25][26][27][28][29][30][31][32]. https://doi.
This study tested the ability of Sentinel-1 C-band to separate forest from other common land use classes (i.e., urban, low vegetation and water) at two different sites. The first site is characterized by temperate forests and rough terrain while the second by tropical forest and near-flat terrain. We trained a support vector machine classifier using increasing feature sets starting from annual backscatter statistics (average, standard deviation) and adding long-term coherence (i.e., coherence estimate for two acquisitions with a large time difference), as well as short-term (six to twelve days) coherence statistics from annual time series. Classification accuracies using all feature sets was high (>92% overall accuracy). For temperate forests the overall accuracy improved by up to 5% when coherence features were added: long-term coherence reduced misclassification of forest as urban, whereas short-term coherence statistics reduced the misclassification of low vegetation as forest. Classification accuracy for tropical forests showed little differences across feature sets, as the annual backscatter statistics sufficed to separate forest from low vegetation, the other dominant land cover. Our results show the importance of coherence for forest classification over rough terrain, where forest omission error was reduced up to 11%.
While products generated at global levels provide easy access to information on forest growing stock volume (GSV), their use at regional to national levels is limited by temporal frequency, spatial resolution, or unknown local errors that may be overcome through locally calibrated products. This study assessed the need, and utility, of developing locally calibrated GSV products for the Romanian forests. To this end, we used national forest inventory (NFI) permanent sampling plots with largely concurrent SAR datasets acquired at C- and L-bands to train and validate a machine learning algorithm. Different configurations of independent variables were evaluated to assess potential synergies between C- and L-band. The results show that GSV estimation errors at C- and L-band were rather similar, relative root mean squared errors (RelRMSE) around 55% for forests averaging over 450 m3 ha−1, while synergies between the two wavelengths were limited. Locally calibrated models improved GSV estimation by 14% when compared to values obtained from global datasets. However, even the locally calibrated models showed particularly large errors over low GSV intervals. Aggregating the results over larger areas considerably reduced (down to 25%) the relative estimation errors.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.