Ana Cornelia Badea

Accurate crop classification from satellite imagery can substantially enhance agricultural monitoring. Recent advances in deep learning have improved crop classification by employing models that take into account both spatial and temporal information contained in satellite image time series. However, the limited interpretability of these models remains a major challenge for their adoption in agricultural applications. This study presents a one-versus-rest classification framework for five crop types (wheat, maize, rapeseed, sunflower and watermelon and melon), in Ialomița County, Romania, using a CNN-LSTM (Convolutional Neural Network–Long Short-Term Memory) architecture and nine Sentinel-2 images acquired throughout the 2025 growing season. Each image sequence consisted of four spectral bands and NDVI (Normalized Difference Vegetation Index), enabling the exploitation of spectral, spatial and temporal information. Model performance was evaluated using accuracy, precision, recall, F1-score, confusion matrices and the area under the receiver operating characteristic curve (ROC-AUC). The models achieved high validation accuracy, with values of 96.01% for wheat, 94.33% for maize, 97.48% for rapeseed, 97.06% for sunflower and 97.06% for watermelon and melon classifications. Explainable AI (XAI) techniques such as Integrated Gradients and SHAP (SHapley Additive exPlanations) were employed to quantify the contribution of individual spectral bands and acquisition dates to the classification decision. The Integrated Gradients analysis showed that for some crops, NDVI was the most influential input feature, accounting for 50.78% of the feature importance in wheat classification, 39.30% in rapeseed classification and 40.01% in watermelon and melon classification, while for other crops, the NIR (Near-infrared) spectral band was the most influential feature, accounting for 29.89% of the feature importance in maize classification and 41.00% in sunflower classification. The results demonstrate that the proposed framework achieves high classification performance while providing meaningful explanations of model predictions, highlighting the importance of both vegetation characteristics and seasonal dynamics in crop classification.

The increasing availability of geospatial data and the rapid development of geospatial artificial intelligence (GeoAI) create new opportunities for modernizing cadastral and land administration systems. However, integrating probabilistic GeoAI outputs into systems structured according to the Land Administration Domain Model (LADM) remains conceptually, semantically and institutionally challenging. ISO 19152 provides standardized structures for land-administration objects, sources and lifecycle information, while ISO 19157 supports the description and evaluation of geographic data quality. A distinct issue nevertheless arises before these mechanisms can support authoritative cadastral information: the status and institutional treatment of model-dependent spatial observations. Using a conceptual framework-development approach based on a critical and integrative literature synthesis, this article proposes an AI-ready LADM framework organized into six interdependent layers: geospatial data acquisition, GeoAI inference, uncertainty representation, semantic mediation, candidate integration, and legal validation with feedback. The framework introduces external, non-authoritative candidate-evidence constructs that preserve source provenance, model information, uncertainty, possible LADM relevance and review history before any institutionally authorized cadastral action. It translates these constructs into a proposed procedural structure comprising source classification, uncertainty and risk assessment, conceptual semantic-mapping rules, validation procedures, audit trails and institutional roles. The contribution is domain-specific rather than a general claim about human oversight of AI. It distinguishes technical detection, candidate cadastral relevance and legally recognized cadastral status, while treating semantic mediation as one component of the broader process of epistemic translation. Scenario-based analysis illustrates the framework’s internal logic, but its effectiveness, usability and legal–institutional feasibility remain subject to subsequent expert assessment and jurisdiction-specific empirical evaluation.

The 15-minute city concept has emerged as a framework for promoting more sustainable, inclusive, and accessible urban environments by enabling residents to reach essential daily services within a short distance from their homes. This study evaluates potential spatial accessibility to mapped points of interest in the Berceni neighbourhood, a socialist-era residential district in southern Bucharest, Romania. Walking isochrones were generated through network analysis and subsequently indexed and aggregated using the H3 hierarchical spatial system at resolution 10. Accessibility was assessed for seven social functions using three weighting specifications: 1:1:1, 3:2:1, and 5:3:1. The final analysis included 520 mapped POIs and 8465 H3 resolution 10 cells. The results reveal broadly consistent spatial patterns across the three weighting specifications, although the weighting assumptions influence local scores, the extent of high-accessibility areas, and the relative ranking of individual H3 cells. Caring has the largest mapped opportunity count and broad spatial coverage, whereas working and governing are the least represented functions and display comparatively low potential accessibility in several parts of the neighbourhood. A function-targeted scenario involving the addition of a childcare centre in the south-western part of Berceni produces a local improvement in working accessibility but does not eliminate the wider functional deficit. The findings demonstrate the applicability of combining network-derived isochrones with H3-based spatial aggregation for neighbourhood-scale accessibility assessment.

Global agriculture is undergoing a period of profound transformation, driven by the need to increase food production in a context characterized by climate change, the degradation of natural resources and increasing pressure on agricultural ecosystems

Blockchain technology is very often presented as a promising solution for modernizing property registration, due to its characteristics of immutability, auditability, and traceability. However, the evolution of research in recent years shows that the initial technological enthusiasm has been followed by a more critical phase, in which the focus has shifted from abstract promises to the concrete conditions of legal, institutional, and operational integration. This article pursues two main objectives: (i) to present the current state of research regarding the use of blockchain in property registration systems and (ii) to identify the suitability of implementing such an infrastructure in Romania, by reference to the legal framework governing real estate registration, data protection requirements, and the current stage of digitalization of the National Agency of Cadastre and Land Registration (ANCPI). The study is qualitative, interdisciplinary, and evaluative in nature and is based on a set of sources such as scientific literature, institutional reports, case studies, and official normative sources. The analysis shows that international literature is dominated by prototypes and architectural models, while real-world implementations remain limited and, in all relevant cases, rely on permissioned and hybrid models, rather than fully decentralized public ledgers. For Romania, the article shows that the main barriers are not strictly technical, but legal–institutional: the legal recognition of distributed entries, the relationship with the function of the land book, the role of ANCPI, and compliance with GDPR. The conclusion of the research is that blockchain may be useful for cadastral records in Romania only within a complementary, phased, and institutionally controlled architecture, rather than as an immediate replacement for the current cadastral registration system.

The monitoring and conservation of built heritage is a major challenge for the scientific community, given the continuous degradation caused by natural, anthropogenic and climatic factors. The generation of high-resolution 3D documentation is important in the diagnosis of deterioration in historic buildings and the planning of conservation and restoration efforts. The present study proposes an integrated, multi-source workflow combining terrestrial laser scanning (TLS), unmanned aerial vehicle (UAV) photogrammetry, and 3D camera interior scanning. This workflow was employed to document and evaluate the Casa Rusănescu monument in Craiova, Romania. The following processes were incorporated: coordinated acquisition, processing, alignment, evaluation of geometric consistency and deviation-based diagnosis. The diagnosis process include measuring the distance between data clouds and analyzing surface roughness, curvature, planarity and linearity. The workflow was designed to be applicable in real urban conditions, ensuring the coverage of façades, interiors and roof structures. The final, combined dataset contained over 235 million points and includes both interior and exterior geometries. This process helped identify various types of damage, such as cracks, exfoliation, plaster detachment, moisture-related changes, and geometric deformations. An additional AI-assisted validation step (Twinspect) was used to cross-check the degradation indicators derived from point-cloud analyses. The findings suggest that using multiple sensors improves spatial completeness, enhances anomaly detection, and establishes a reliable baseline prior to restoration interventions and long-term monitoring. This methodology facilitates the development of digital twins and GIS-based risk assessments, thereby providing a scalable solution for heritage preservation.

The automation of public services represents a key area of development at the national level, with the main goal of facilitating citizens’ access to comprehensive, integrated and high-quality services in the shortest possible time. National strategies emphasize the need to integrate open geospatial data and artificial intelligence into information, transparency and decision-making processes. The evolution of artificial intelligence, particularly large language models (LLMs), has led to the development of virtual assistants capable of understanding user requirements and providing answers in natural, easy-to-understand language. This paper presents directions for the development and use of large-language-model-based virtual assistants, focusing on their ability to understand and interact with the geospatial domain through an LLM API. Geospatial modeling contributes significantly to the automation of public services, but access to this technology is often limited by technical expertise or dedicated software programs. The development of AI-based virtual assistants removes these barriers, facilitating access, reducing time and ensuring transparency and accuracy of information. The proposed approach is implemented using a commercial large language model API, integrated with domain-specific geospatial functions and authoritative spatial databases. This study highlights practical examples of virtual assistants capable of understanding the geospatial field and contributing to the optimization and automation of public services in the country. In addition, the paper presents comparative analyses, challenges encountered and potential directions for future research.

Urban Digital Twins (UDTs) represent a recent application of Digital Twins (DTs), with the objective of replicating cities and providing a framework for urban planning. The utilization of UDTs provides a structured approach for the modeling and analysis of urban environments, incorporating a range of geospatial data presented in both two-dimensional (2D) and three-dimensional (3D) formats. This article details the process of processing, modeling, and integrating urban geospatial data into a Digital Twin. Two integrations for end-user platforms were demonstrated using a Geographic Information System (GIS) and an Unreal Engine (UE5) game platform. GIS-based dashboard systems provide professionals with the tools necessary to monitor, analyze, and create scenarios, thereby promoting collaboration between authorities and citizens. Game engines have the potential to play a pivotal role in the development of future UDTs by facilitating the creation of immersive experiences that aid users in comprehending their environment and promoting citizen engagement.

Peri-urban areas in Eastern Europe are undergoing rapid land transformation driven by suburban housing expansion and infrastructure development, yet the processes through which vegetation is progressively degraded and built-up areas intensify remain insufficiently documented. This study analyses vegetation loss and urban expansion in the peri-urban belt of Timișoara, Western Romania, between 2020 and 2025 using Sentinel-2 multispectral imagery, two key spectral indices (NDVI and NDBI), and a Random Forest (RF) classifier. The results reveal a gradual, multi-stage transformation trajectory, where dense vegetation transitions first into sparse vegetation and bare soil before consolidating into built-up surfaces, rather than being replaced abruptly. Substantial vegetation decline is accompanied by notable increases in built-up land, with strong spatial differences between communes depending on development pressure. The integration of RF classification with spectral index analysis allows these transitions to be validated and interpreted more reliably, helping distinguish structural suburbanisation from short-term spectral variability. Overall, the study demonstrates the value of combining NDVI, NDBI and AI-supported land-cover classification to capture nuanced peri-urban transformation dynamics and provides actionable insights for spatial planning and sustainable land management in rapidly growing metropolitan regions.