Search

Protected areas, or national parks, are established to preserve natural ecosystems; their effectiveness on the territory needs to be evaluated. We propose considering a time series of the SDG 15.1.2 indicator, "Proportion of important sites for terrestrial and freshwater biodiversity that are covered by protected areas, by ecosystem type", to quantify the presence over time of grassland ecosystem in Murgia Alta (southern Italy), within the Natura 2000 and national park boundaries. Time series of remote sensing imagery, freely available, were considered for extracting, by Support Vector Machine classifiers, a time series of grassland cover mappings from 1990 to 2021. This latter was, then, used for computing a time series of the SDG 15.1.2 indicator. A high reduction (about 15,000 ha) of grassland presence from 1990 to 2004, the foundation years of the national park, followed by the increasing stability up to nowadays, was evaluated. Furthermore, grassland presence was evaluated in a 5-km buffer area, surrounding Natura 2000 boundary, revealing a continuous loss from 1990 up to now (about 500 ha) in the absence of protection actions. This study represents the first long-term analysis for the grassland ecosystem in Murgia Alta and the first effort to analyze a time series of the SDG 15.1.2 indicator. The findings can provide inputs to governments in monitoring the effectiveness of protection actions. [ABSTRACT FROM AUTHOR]

The present study aims to discriminate four semi-arid grassland habitats in a Mediterranean Natura 2000 site, Southern Italy, involving 6210/E1.263, 62A0/E1.55, 6220/E1.434 and X/E1.61-E1.C2-E1.C4 (according to Annex I of the European Habitat Directive/EUropean Nature Information System (EUNIS) taxonomies). For this purpose, an intra-annual time-series of 30 Sentinel-2 images, embedding phenology information, were investigated for 2018. The methodology adopted was based on a two-stage workflow employing a Support Vector Machine classifier. In the first stage only four Sentinel-2 multi-season images were analyzed, to provide an updated land cover map from where the grassland layer was extracted. The layer obtained was then used for masking the input features to the second stage. The latter stage discriminated the four grassland habitats by analyzing several input features configurations. These included multiple spectral indices selected from the time-series and the Digital Terrain Model. The results obtained from the different input configurations selected were compared to evaluate if the phenology information from time-series could improve grassland habitats discrimination. The highest F1 values (95.25% and 80.27%) were achieved for 6210/E1.263 and 6220/E1.434, respectively, whereas the results remained stable (97,33%) for 62A0/E1.55 and quite low (75,97%) for X/E1.61-E1.C2-E1.C4. However, since for all the four habitats analyzed no single configuration resulted effective, a Majority Vote algorithm was applied to achieve a reduction in classification uncertainty. [ABSTRACT FROM AUTHOR]

Grassland ecosystems can provide a variety of services for humans, such as carbon storage, food production, crop pollination and pest regulation. However, grasslands are today one of the most endangered ecosystems due to land use change, agricultural intensification, land abandonment as well as climate change. The present study explores the performance of a knowledge-driven GEOgraphic-Object—based Image Analysis (GEOBIA) learning scheme to classify Very High Resolution (VHR) images for natural grassland ecosystem mapping. The classification was applied to a Natura 2000 protected area in Southern Italy. The Food and Agricultural Organization Land Cover Classification System (FAO-LCCS) hierarchical scheme was instantiated in the learning phase of the algorithm. Four multi-temporal WorldView-2 (WV-2) images were classified by combining plant phenology and agricultural practices rules with prior-image spectral knowledge. Drawing on this knowledge, spectral bands and entropy features from one single date (Post Peak of Biomass) were firstly used for multiple-scale image segmentation into Small Objects (SO) and Large Objects (LO). Thereafter, SO were labelled by considering spectral and context-sensitive features from the whole multi-seasonal data set available together with ancillary data. Lastly, the labelled SO were overlaid to LO segments and, in turn, the latter were labelled by adopting FAO-LCCS criteria about the SOs presence dominance in each LO. Ground reference samples were used only for validating the SO and LO output maps. The knowledge driven GEOBIA classifier for SO classification obtained an OA value of 97.35% with an error of 0.04. For LO classification the value was 75.09% with an error of 0.70. At SO scale, grasslands ecosystem was classified with 92.6%, 99.9% and 96.1% of User's, Producer's Accuracy and F1-score, respectively. The findings reported indicate that the knowledge-driven approach not only can be applied for (semi)natural grasslands ecosystem mapping in vast and not accessible areas but can also reduce the costs of ground truth data acquisition. The approach used may provide different level of details (small and large objects in the scene) but also indicates how to design and validate local conservation policies. [ABSTRACT FROM AUTHOR]

The present study focuses on the use of Convolutional Neural Networks (CNN or ConvNet) to classify a multi-seasonal dataset of Sentinel-2 images to discriminate four grassland habitats in the "Murgia Alta" protected site. To this end, we compared two approaches differing only by the first layer machinery, which, in one case, is instantiated as a fully-connected layer and, in the other case, results in a ConvNet equipped with kernels covering the whole input (wide-kernel ConvNet). A patchwise approach, tessellating training reference data in square patches, was adopted. Besides assessing the effectiveness of ConvNets with patched multispectral data, we analyzed how the information needed for classification spreads to patterns over convex sets of pixels. Our results show that: (a) with an F1-score of around 97% (5 × 5 patch size), ConvNets provides an excellent tool for patch-based pattern recognition with multispectral input data without requiring special feature extraction; (b) the information spreads over the limit of a single pixel: the performance of the network increases until 5 × 5 patch sizes are used and then ConvNet performance starts decreasing. [ABSTRACT FROM AUTHOR]