Democratic Landscape Monitoring: Integrating Earth Observation, Landscape Ecology and Participatory GIS

This study presents an integrated framework for Landscape Character Assessment (LCA), comparing an expert-based GIS approach with a Generative Artificial Intelligence (GenAI)-assisted approach across two spatial scales. The aim is to explore how GenAI can support the delineation, interpretation, and communication of landscape character, while critically comparing its outputs with those produced through conventional expert-based geospatial analysis. The expert-based methodology follows a hierarchical structure, beginning with the delineation of Land Description Units (LDUs) based on geomorphological and land cover attributes, and progressing to the synthesis of Landscape Character Types (LCTs) through spatial aggregation and expert interpretation. The overall analysis integrates spatial datasets related to morphology, land cover, vegetation condition, and population density, capturing both natural and anthropogenic dimensions of the landscape. In parallel, a GenAI-assisted LCA methodology is applied using the same spatial datasets as input. Through a structured prompt-based workflow, GenAI supports the interpretation of spatial classifications, the formulation of landscape character descriptions, and the development of scenario-based narratives. The comparison is conducted at two spatial scales: a detailed local-scale analysis of the Akrotiri Peninsula, Chania/ Crete (Greece), and a broader regional-scale assessment covering the Prefecture of Chania. This multi-scale design allows the evaluation of how landscape classification and interpretation vary according to spatial resolution, territorial extent, and analytical scale. Overall, the framework demonstrates the potential of GenAI to enhance the efficiency, consistency, and communicability of LCA documentation, while underlining the continued importance of expert validation for local specificity, landscape meaning, and contextual interpretation. This research was funded by the European Union’s Horizon Europe MSCA Staff Exchanges 2022 (HORIZON-MSCA-2022-SE-01), project “Towards Democratic Landscape Observation Through Geoinformatics and Public Participation (TOPIO)”, under grant agreement No 101131109.

Mediterranean rural landscapes are shaped by long-term interactions between water management, agrarian practices, and local knowledge, generating hybrid environments where ecological functions, land-use patterns, and heritage values are tightly interwoven. This paper proposes a spatial decision-support framework that combines Historical GIS with Public Participation GIS (PPGIS) to construct a Heritage Potential Index capable of identifying areas of high heritage value and landscape fragility, with a specific focus on case studies in Italy and Greece.. Geo-historical layers (multitemporal land-use/land-cover change, historical hydrography, archaeological records, traditional rural features) are integrated with participatory and volunteered geographic information (community-mapped landmarks, perceived values and conflicts, PPGIS data, OSM-based contributions) within a GIS-based Multi-Criteria Decision Analysis (MCDA) workflow. Criteria and indicators are structured into ecological, historical, and socio-cultural dimensions; they are weighted through an expert–stakeholder procedure and aggregated via a weighted linear combination to derive a continuous Heritage Potential Index. The framework is designed for application across Mediterranean contexts, enabling the identification of “core” zones where historical continuity, ecological connectivity, and community meanings converge, as well as contested areas where land-use pressures and perceived values diverge. By explicitly integrating geo-historical reconstruction and participatory spatial knowledge, the paper contributes to democratic landscape monitoring and offers a transparent, replicable tool to support heritage- oriented planning in Mediterranean landscapes.

This study introduces an Artificial Intelligence (AI)-based framework for assessing the visual quality (“scenicness”) of rural landscapes by integrating geospatial data, landscape ecology principles, and participatory evaluation processes. The approach is grounded in the Information Processing Theory developed by Rachel Kaplan and Stephen Kaplan, which explains landscape preference through four key perceptual dimensions: coherence, complexity, legibility, and mystery. The methodology incorporates participatory data collected from local communities, stakeholders, and visitors, alongside image-based inputs, to support democratic landscape monitoring and inclusive landscape quality assessment. A dataset of images from diverse rural case studies is used to train Artificial Neural Networks (ANNs), specifically Multi-Layer Perceptron (MLP) models, to predict scenicness. The system employs a pipeline including semantic segmentation of landscape elements (e.g., vegetation, water, built environment), combined with feature extraction based on spatial composition, and a constrained optimization model to align outputs with human perception. The framework is demonstrated through a case study in Falasarna, Crete, Greece, a Mediterranean coastal landscape of high ecological and cultural value. Results show that the AI-based approach provides more accurate scenicness predictions compared to traditional regression models. The results were evaluated against human scenicness assessments, demonstrating strong agreement between the AI-based outputs and participants’ perceptions. By integrating cognitive theory with machine learning and participatory GIS principles, the proposed framework enables scalable, transparent, and user-informed landscape monitoring. Its application within digital twin environments supports real-time, human-centered evaluation, contributing to more inclusive landscape planning, sustainable tourism management, and evidence-based policy-making. This research was funded by the European Union’s Horizon Europe MSCA Staff Exchanges 2022 (HORIZON-MSCA-2022-SE-01), project “Towards Democratic Landscape Observation Through Geoinformatics and Public Participation (TOPIO)”, under grant agreement No 101131109. Kaplan, Stephen, Kaplan, & Rachel. (1981). Cognition and environment: functioning in an uncertain world. https://www.hse.ru/data/2019/03/04/1196348207/%5BRachel_Kaplan,_Stephen_Kaplan%5D_The_Experience_of_(b-ok.xyz).pdf

The fires of the summer of 2007 in Greece have been recorded as the most destructive in Europe in recent decades. The total area burned exceeded 250,000 hectares, of which 30,132 were located within the Natura protected areas 2000 network. Furthermore, for the first time, the damage report included not only burned forests and agricultural land, but also entire villages, civil infrastructure and the loss of 57 lives. This research thoroughly investigates and examines the short, middle, and long term impact of these catastrophic wildfires. It estimates the severity of the impact and the time needed for regeneration of the forests, regeneration implying that the forests returned to their previous extends and syntheses before the fires. The study is conducted using remote sensing data and methods, i.e., satellite images and products from NASA's MODIS satellites, namely the products MOD09A1 and MCD64A1. Seven different polygons (burned areas) were singled out as Regions of Interest (ROIs) and their evolution was monitored for the entire period of satellite data (2002-present) with the construction of the time series of the NDVI (Normalized Difference Vegetation Index) and NBR (Normalized Burn Ratio) ratios. The post-fire yearly evolution of these ratios was compared with the averaged pre-fire values leading to some first conclusions. It is noted that the NBR index is normally used as the difference pre-fire to post-fire (dNBR) but the particular point in time that is considered pre-fire can be varied to conform with the scope of the study. Additionally, the concurrent examination of the two indices can provide useful input for the timely growth of the forest vegetation. Given the uncertainties inherent in remote sensing methods the results were compared with other similar studies, especially those that included field examinations, and the comparisons were favorable.

One of the key elements of urban sustainability and resilience is ensuring that natural components in cities, such as green and blue spaces, are managed and protected. The rapid transformation of Varanasi and its built-up expansion have caused the loss and fragmentation of urban green and blue spaces, negatively affecting urban ecosystem health and threatening long-term sustainability. This study derived and mapped ecosystem health indicators from 1993 to 2026 for Varanasi city. Multi-temporal Landsat imagery is employed to derive urban green and blue spaces dynamics, and landscape fragmentation metrics are used to characterise structural changes. The Ecosystem Vigor Index (EVI) is derived from four spectral indices (Normalised Difference Vegetation Index, Normalised Difference Water Index, Normalised Difference Built-up Index, and Modified Bare Soil Index), and the Ecosystem Structure Index (ESI) is derived from landscape shape, connectivity, and heterogeneity indices using FRAGSTAT v4.2.1. The results showed noticeable decreases in green and blue spaces over the past three decades, along with increased fragmentation, particularly in the periphery of the city. These indicators can help assess ecosystem health, provide evidence-based data to improve it, and support effective governance of green and blue spaces in Varanasi city.

The coastal and marine zone of Cyprus is under increasing pressure from multiple anthropogenic activities, making the accurate spatial assessment of their impacts imperative. This study presents a comprehensive Cumulative Impact Assessment (CIA) methodology for Cypriot marine ecosystems, building upon the established Halpern model. The objective is the high-resolution mapping of environmental degradation hotspots by integrating advanced Geographic Information Systems (ArcGIS Pro) techniques with programming algorithms (Python, GeoPandas). Spatial analysis was conducted across three grid scales (1x1km, 500x500m, and 250x250m resolutions), examining the interactions between diverse anthropogenic pressures (e.g., shipping, port facilities, aquaculture, military operations) and sensitive marine habitats, with a focus on Posidonia meadows (Seagrass), Reefs, and MSFD Benthic Broad Habitat Types (BBHT). The geospatial data was subsequently fed into a custom Python script, which applied a specific vulnerability matrix to compute the final CIA index. The results highlight areas of maximum accumulated pressure in Cypriot waters, providing a realistic and quantified representation of the environmental status. To facilitate broad accessibility and practical decision-making, the final research outcomes have been visualized and integrated into an interactive web application. The proposed methodology offers a powerful, scientifically validated decision-support tool, directly applicable to Marine Spatial Planning (MSP) for the sustainable management of Cyprus's marine environment.