Conference Agenda

This study aims at investigating the suitability of very-high resolution satellite imagery, e.g., Pleiades, WorldView-2 for classifying the dominant urban tree species in both public and private areas and estimating the tree height. Tree species differentiation is challenging in cities, as trees can be lined up or grouped in patch, with a wide range of plant species, high spectral similarity of vegetation types, and due to the complexity of the urban environment (buildings, shadows, open courtyards).

In the present study, we have implemented the so-called “object-based classification of urban tree species from very high-resolution satellite imagery” methodology we published in Sicard et al. (2023), that is an object-based classification using Random Forest classifier with different textural features extracted from tree canopy and grassland (lawn/turf) to identify and map dominant types of vegetation. Four spectral bands (blue, green, yellow, red) and four texture features (i.e., energy, entropy, inverse difference moment, Haralick correlation) were the most efficient attributes. Tree height estimation was performed using Pleaides tri-stereoscopic imageries and the photogrammetric tool MicMac of the French National Geographic Institut and the ForestTools package.

We applied the methodologies in two cities Aix-en-Provence (France, 50km²) and Florence (Italy, 80km²) where about 420,000 and 555,000 canopies were successfully classified in 22 and 20 dominant species with an overall accuracy of 84% and 83%, respectively. We found that about 85% of trees in both cities are in private lands. The highest classification accuracy was obtained for Pinus spp. and Platanus acerifolia in Aix-en-Provence, and for Celtis australis and Cupressus sempervirens in Florence. For the tree height, we obtained an absolute accuracy of 1.8m.

As cities expand, the pressure on natural ecosystems intensifies. Urban and peri-urban areas try to overcome with conflicting demands, which results in fragmented green infrastructure, and severed links between rural and urban zones. The European Union Biodiversity Strategy for 2030 recognizes this urgency and calls for action. The ambitious goal of "Greening urban and peri-urban areas" aims to reverse these trends. Similarly, the UN Sustainable Development Goals indicate the need for inclusive and accessible green urban areas (SDG 11.3).

The concept of Green and Blue Infrastructure (GBI) represents a strategic network of natural and semi-natural areas, thoughtfully integrated into urban and peri-urban landscapes. GBI weaves together green spaces, water bodies, and wildlife corridors to create a dynamic fabric that sustains life and mitigates environmental challenges.

By combining the implementation of GBI together with prioritising accessible green spaces, cities and regions can combat the dual crises of biodiversity loss and climate change. To support policymakers in the identification of these priority areas, a methodology was developed by the authors to map urban green infrastructure in the European Functional Urban Areas. This methodology used several High Resolution Layers (HRLs) and the Urban Atlas from the Copernicus Land Monitoring Service (CLMS) as well as complementary data from the Open Street Map (OSM), improving the thematic and spatial resolution provided by Urban Atlas. As a result, two datasets were produced corresponding to 2012 and 2018 which allows monitoring changes between the defined 8 GBI classes. It also allows quantifying the amount of GBI created within this period. Results are presented in transparent and easy to understand dashboards that allows users to analyse and download GBI data in Europe. The work is carried out in collaboration of the European Environment Agency and DG REGIO of the European Commission.

Compound dry and hot conditions have the potential to result in adverse socio-economic and ecological impacts in urban areas; they are usually associated with larger repercussions and risks than individual extremes. The thermal properties of vegetated urban parks enable them to function as climate modifiers, thereby increasing the adaptive capacity of cities. In this work, a climatology of compound drought and heat wave events (CDHW) is firstly derived for selected megacities worldwide (cities with population over 10 million residents). To achieve this, reanalysis data (ERA5-Land) and extreme indices (Excess Heat Factor for heatwaves and the Standardized Precipitation Evaporation Index for droughts) are used. Thermal patterns inside cities during CDHW events are derived using remotely-sensed Land Surface Temperatures (LST) (from Landsat 8/9 and MODIS Aqua/Terra satellites) and the co-occurrence between air temperature and LST anomalies is assessed. Then, the thermal regime of large-scale parks within megacities is evaluated under both typical conditions and during the identified CDHW extremes. Remotely sensed data underscore the role of urban green in mitigating the increasing heat and drought hazards confronting cities, driven by local and global climate change.

Urban green spaces are an important component in enhancing urban livability. Therefore, there is a need for sustainable management of these spaces, which can be acquired by using Earth Observation (EO) technology, to maintain an ecological balance, while mitigating the impact of urban heat islands. By harnessing high-resolution EO data, the S4UG project aimed to provide city managers and planners with advanced tools for informed decision-making in urban growth modeling and ecological assessments, particularly focusing on green and blue infrastructures, urban planning, and urban ecology. In this sense, stakeholders from twelve European cities were involved in defining the needs that arise in green infrastructure mapping in terms of resolution, frequency, and accuracy of the data. While Sentinel-2 imagery is suitable for monitoring urban spaces, due to covering vast areas frequently, their spatial resolution is not always enough to detect smaller green areas, specific to urban environments.

To overcome the limitations of Sentinel-2's image resolution, our research conducted during the S4UG project, focused on using super-resolution (SR) techniques for detailed vegetation extent mapping in urban environments. This application involved using images from Sentinel-2 satellites, which have a resolution of 10 meters, and from SPOT 7 satellites with a 1.5-meter resolution. The SR techniques were used to create synthetic imagery with a finer resolution of 2.5 meters. This improved resolution offered more accurate mapping and monitoring capabilities, highlighting its usefulness through services that provided detailed up-to-date land use/land cover maps and vegetation health monitoring, correlation between vegetation health with environmental factors over time and analyzes of green spaces accessibility for inhabitants. These capabilities proved valuable for stakeholders involved in the project, who require detailed spatial data to make informed decisions, showing that the exploitation of SR techniques at application level has the potential to enhance the added value of the provided solutions.

Urban planning is traditionally centred on human perspective, overlooking the needs of animals and plants that in turn help fostering liveability in cities. To address this bias, the concept of green urbanism emerges as a key driver for sustainable urban development. Nevertheless, the description of these habitats is usually not holistic, lacking the common accent to describe inter-specie interaction.

In response, we propose a novel methodology for characterising urban habitats showcased in an application to three distinct urban case studies: the Functional Urban Areas (FUA) of Vienna (Austria), Munich (Germany), and Genoa (Italy). Our approach transcends anthropocentrism by adopting an agnostic perspective—one that embraces the diverse living entities within co-habitative landscapes.

The classification is composed of four 10-meter rasters for each city, each with a separate set of variables focusing on different aspects of urban metabolisms, such as urban morphology at local and landscape scale, anthropic imprint, and biophysical conditions. The workflow scrapes from open data with global validity, resampling at a 10-meter resolution with a 100-meter moving window algorithm. After preprocessing, the pipeline ends with a combination of MiniBatch and Elkan KMeans algorithms.

The resulting rasters bring insights into distinct urban habitat classes and can provide meaningful observations alone or combined for both urban planners and natural scientists. The classification goal is to integrate with models for optimization and multi-criteria decision making. The role is not being an omniscient knowledge base, but a multi-disciplinary common understanding of urban habitats. The tool enables two methodologies: selecting a site for architectural design or ecological research, given a threshold of targets for liveability and conservation, and comparing different sites from different regions. The tool is developed with scalability in mind, and the next step will be to expand to all European FUA.

Cities are facing too many challenges. Urban vegetation, in particular trees, are essential as they provide services in terms of air pollution mitigation, freshness, biodiversity, and citizens’ well-being. The main aim of the European project AIRFRESH “Air pollution removal by urban forests for a better human well-being” is to quantify the environmental benefits provided by urban trees at city-scale in two front-runner cities: Florence (Italy) and Aix-en-Provence (France). To avoid a large underestimation of the quantification of benefits, a consistent inventory of vegetation within private and public areas is needed. Based on an object-based classification from very-high resolution satellite imagery (WorldView-2), we have detected and classified about 550,000 and 414,000 trees, and grassland, in the two study areas. Then, we applied the AIRTREE multi-layer canopy model and WRF-Chem model for the year 2019. We have also considered species-specific parameters, such as tree morphology (height, diameter at breast height, and crown leaf area), leaf habit (deciduous/evergreen) and eco-physiological responses to environmental factors. Finally, we have mapped the annual removal capacity of each urban tree for the most harmful air pollutants in cities, i.e., tropospheric ozone (O₃), nitrogen dioxide (NO₂), and particulate matter (PM_2.5, PM₁₀) in addition to carbon dioxide (CO₂). Deciduous broadleaves have high capacity to uptake O₃ and NO₂ (Platanus x acerifolia: 1.25 and 0.40 g m^-2 per year). Compared to deciduous broadleaves, evergreen conifers, such as Cupressus spp., showed higher performance in PM removal (11.5 and 2.5 g m^-2 per year for PM₁₀ and PM_2.5). We identified tree species with the highest CO₂ uptake capacity with values up to 2.5 kg m^-2 per year for Cedrus atlantica.