.. _results-and-discussion: Results and Discussion ====================== As has been mentioned, in our study, seven parameters of coastal vulnerability were identified, and they are analysed in this section along with the CVI results. Figure 4 indicates the vulnerability ranking for three parameters utilised to calculate the CVI in the coastal area of Kalamata, Greece. It is also noted that AI-powered demonstrator was not only tested in Kalamata but also applied to a range of diverse coastal environments to assess its adaptability and effectiveness in evaluating CVI in areas of differentiated environmental, geographic and social characteristics. Characteristic areas of interest included Evros, with its complex river estuary dynamics; Eresos, Lesvos, known for its wetlands and documented erosion; and Heraklion, Crete, where diverse coastal uses and environmental significance were key considerations. Additionally, the demonstrator analyzed the Achelous Estuary, a region marked by historical shoreline changes, and Northern Kyparissia, an area experiencing intense erosion and hosting critical coastal ecosystems. In Attica, the analysis covered Marathonas-Nea Makri, a region with significant infrastructure and archaeological value, and Palaia Fokaia in the Saronic Gulf, where tourism and erosion risks intersect. The system also evaluated Rio-Antirio, a unique location combining port activities and marine technical infrastructure, as well as Barcelona, further expanding the scope of the analysis. In particular, for Kalamata, the results show the land cover is characterised by a variety of land-cover classes, grassland and built-up areas being the dominant classes according to the Copernicus landcover. Also, coastal elevation results highlight that the north and central part of the coast are dominated by low altitude areas (< 2 m) as well as the eastern part of Kalamata beach, increased altitudes occur at a range of 5 – 25 m. In contrast, the mean wave height results show that the majority of the coastline is represented by dark green shading, indicating a mean wave height between 0.0 and 0.55 meters. This suggests that the area generally experiences relatively calm wave conditions, which may be beneficial for coastal resilience and shoreline stability however, only one grid point of the Copernicus data has been identified as the closest point to the shore, due to the coarse spatial resolution (approx. 25 km). .. image:: images/image5.jpg :alt: A map of the ocean AI-generated content may be incorrect. .. image:: images/image6.jpg :alt: A map with a green line AI-generated content may be incorrect. .. image:: images/image7.jpg :alt: A map of the ocean AI-generated content may be incorrect. **Figure 4:** Sub-criteria results, as extracted following the transects approach Also, the CVI results for the Kalamata coastline (see Figure 5) indicate varying levels of susceptibility to coastal hazards. The index ranges from 0.58 (very low vulnerability) to 10.21 (very high vulnerability), with a mean value of 4.51. The spatial distribution reveals that areas with very low to low vulnerability (green shades) are primarily concentrated near the eastern coastline, likely due to favorable geomorphological conditions and lower wave exposure. In contrast, moderate to high vulnerability zones (yellow to orange shades) are more prevalent along the central and western coastal stretches, indicating increased susceptibility to erosion, sea-level rise, or storm impacts. The presence of very high vulnerability (red) in certain sections suggests critical areas requiring targeted coastal protection measures. .. image:: images/image8.jpg :alt: A map of a sea AI-generated content may be incorrect. **Figure 5:** CVI results, as extracted following the transects approach The integration of LLaMA 3 within the Generative AI-powered demonstrator was crucial for processing and analyzing CVI data, as showcased in the CVI Calculator results of Figure 6. The demonstrator's architecture enabled seamless data retrieval and processing from various OGC-compliant services. As seen in the CVI computation process of Figure 3, LLaMA 3 successfully interpreted commands, processes geospatial data, and generated vulnerability maps and assessment report by classifying coastline characteristics. The results presented in the CVI process demonstrate how the AI-driven workflow efficiently classified and visualized numerous coastal parameters, offering real-time insights into vulnerability levels. By integrating Django as the backend, the system facilitated smooth user interaction and API communication, ensuring that CVI metrics are computed and visualized effectively via a user-friendly interface for interpreting the results as also, for setting different queries related to the inherent results accuracies. Thus, to address uncertainties in the CVI assessment, it was essential to analyze potential inaccuracies arising from both AI-generated responses and the resolution and quality of source datasets. Identifying key datasets that require higher resolution or improved consistency can significantly enhance the accuracy of vulnerability assessments. Additionally, integrating temporal analysis capabilities will allow users to estimate CVI trends over time (e.g., from 1995 to 2025) based on the available temporal coverage of Copernicus data. .. raw:: html
**Figure 6:** CVI results, via the AI-generated responses Suggestions for further improvements incorporate the post-processing capabilities should also be refined by allowing users to recalculate results using adjusted weights for datasets that hold greater analytical significance. Exploring the impact of weight variations on key outputs will help ensure the robustness and reliability of the CVI assessment. Engaging with end users through testing sessions will be crucial in gathering feedback on the demonstrator’s usability, accuracy, and relevance. Incorporating user insights will support continuous improvements in both methodology and interface design, making the tool more intuitive and effective for policymakers, planners, and researchers working on coastal resilience.