Search

The ash dieback pandemic, caused by the invasive fungus Hymenoscyphus fraxineus, represents one of Europe's biggest threats to preserving natural biodiversity. To ensure the suppression of forest damage caused by fungi, timely recognition of the symptoms of ash dieback and further continuous monitoring on an adequate spatial scale are essential. Visual crown damage assessment is currently the most common method used for identifying ash dieback, but it lacks the spatial and temporal coverage required for effective disease suppression. Remote sensing technologies, with the capabilities of fast and repetitive retrieval of information over a large spatial scale, could present efficient supplementary methods for ash damage detection and disease monitoring. In this study, we provided a synthesis of the existing remote sensing methods and applications that considers ash dieback disease, and we described the lifecycle of the disease using the major symptoms that remote sensing technologies can identify. Unfortunately, although effective methods of monitoring biotic damage through remote sensing have been developed, ash dieback has only been addressed in two research studies in the United Kingdom and Germany. These studies were based on single-date hyperspectral and very-high-resolution imagery in combination with machine learning, using previously specified ground-truth information regarding crown damage status. However, no study exists using high-resolution imagery such as Sentinel-2 or radar Sentinel-1, although some preliminary project results show that these coarser sources of information could be applicable for ash dieback detection and monitoring in cases of Fraxinus angustifolia, which forms pure, more homogenous stands in Southern Europe. [ABSTRACT FROM AUTHOR]

This paper shows the work carried out to obtain a methodology capable of monitoring the Common Agricultural Policy (CAP) aid line for the protection of steppe birds, which aims to improve the feeding and breeding conditions of these species and contribute to the improvement of their overall biodiversity population. Two methodologies were initially defined, one based on remote sensing (BirdsEO) and the other on Machine Learning (BirdsML). Both use Sentinel-1 and Sentinel-2 data as a basis. BirdsEO encountered certain impediments caused by the land's slope and the crop's height. Finally, the methodology based on Machine Learning offered the best results. It evaluated the performance of up to 7 different Machine Learning classifiers, the most optimal being RandomForest. Fourteen different datasets were generated, and the results they offered were evaluated, the most optimal being the one with more than 150 features, including a time series of 8 elements with Sentinel-1, Sentinel-2 data and derived products, among others. The generated model provided values higher than 97% in metrics such as accuracy, recall and Area under the ROC Curve, and 95% in precision and recall. The methodology is transformed into a tool that continuously monitors 100% of the area requesting aid, continuously over time, which contributes positively to optimizing the use of administrative resources and a fairer distribution of CAP funds. [ABSTRACT FROM AUTHOR]

Land-Use/Land-Cover (LULC) products are a common source of information and a key input for spatially explicit models of ecosystem service (ES) supply and demand. Global, continental, and regional, readily available, and free land-cover products generated through Earth Observation (EO) data, can be potentially used as relevant to ES mapping and assessment processes from regional to national scales. However, several limitations exist in these products, highlighting the need for timely land-cover extraction on demand, that could replace or complement existing products. This study focuses on the development of a classification workflow for fine-scale, object-based land cover mapping, employed on terrestrial ES mapping, within the Greek terrestrial territory. The processing was implemented in the Google Earth Engine cloud computing environment using 10 m spatial resolution Sentinel-1 and Sentinel-2 data. Furthermore, the relevance of different training data extraction strategies and temporal EO information for increasing the classification accuracy was also evaluated. The different classification schemes demonstrated differences in overall accuracy ranging from 0.88% to 4.94% with the most accurate classification scheme being the manual sampling/monthly feature classification achieving a 79.55% overall accuracy. The classification results suggest that existing LULC data must be cautiously considered for automated extraction of training samples, in the case of new supervised land cover classifications aiming also to discern complex vegetation classes. The code used in this study is available on GitHub and runs on the Google Earth Engine web platform. [ABSTRACT FROM AUTHOR]