166 research outputs found
Sort by
Localization in the mapping particle filter
Full text linkData assimilation involves sequential inference in geophysical systems with nonlinear dynamics and observational operators. Non-parametric filters are a promising approach for data assimilation because they are able to represent non-Gaussian densities. The mapping particle filter is an iterative ensemble method that incorporates the Stein Variational Gradient Descent (SVGD) to produce a particle flow transforming state vectors from prior to posterior densities. At every pseudo-time step, the Kullback-Leibler divergence between the intermediate density and the target posterior is evaluated and minimized. However, for applications in geophysical systems, challenges persist in high dimensions, where sample covariance underestimation leads to filter divergence. This work proposes two localization methods, one in which a local kernel function is defined and the particle flow is global. The second method, given a localization radius, physically partitions the state vector and performs local mappings at each grid point. The performance of the proposed Local Mapping Particle Filters (LMPFs) is assessed in synthetic experiments. Observations are produced with a two-scale Lorenz system, while a one-scale Lorenz model is used as surrogate, introducing model error in the inference. The methods are evaluated with both full and partial observations, as well as with different linear and non-linear observational operators. The LMPFs with Gaussian mixtures in the prior density perform similarly to Gaussian filters such as the Ensemble Transform Kalman Filter (ETKF) and the Local Ensemble Transform Kalman Filter (LETKF) in most cases, and in some scenarios, they provide competitive performance in terms of analysis accuracy.</p
Review article: Deep learning for potential landslide identification: data, models, applications, challenges, and opportunities
Full text linkAs global climate change and human activities escalate, the frequency and severity of landslide hazards have been increasing. Early identification, as an important prerequisite for monitoring, evaluation, and prevention, has become increasingly critical. Deep learning, as a powerful tool for data interpretation, has demonstrated remarkable potential in advancing landslide identification, particularly through the automated analysis of remote sensing, geological, and topographic data. This review systematically examines and synthesizes over 400 studies, with a primary focus on literature from the last six years (2020–2025), alongside key foundational works. It provides a comprehensive overview of recent advancements in the utilization of deep learning for potential landslide identification. First, the sources and characteristics of landslide-related data are summarized, including satellite observation data, airborne remote sensing data, and ground-based observation data. Next, commonly used deep learning models are classified based on their roles in potential landslide identification, such as image analysis and time series analysis. Then, the role of deep learning in identifying rainfall-induced landslides, earthquake-induced landslides, human activity-induced landslides, and multi-factor-induced landslides is summarized. Although deep learning has achieved considerable success in landslide identification, it still faces several challenges, including data imbalance, limited model generalization, and the inherent complexity of landslide mechanisms. Finally, future research directions in this field are discussed. It is suggested that integrating knowledge-driven and data-driven approaches for potential landslide identification will further enhance the applicability of deep learning, offering broad prospects for future research and practice.</p
An Automated Data Validation Approach for Power Distribution Networks using Grid Partitioning and Multi-faceted Quality Scoring
Full text linkAccurate Geographic Information System (GIS) data is fundamental to the reliable management, planning, and operation of modern power distribution networks. Conventional validation methods, however, often rely on network-wide rule-based checks or manual inspections, which are inefficient at identifying and localizing errors within vast, heterogeneous infrastructures. These approaches frequently fail to detect complex spatial or topological inconsistencies, leading to significant operational challenges and costly data remediation efforts. To address these limitations, a novel, automated validation pipeline has been developed with a modular, two-stage approach. The first stage, Smart Grid Partitioning, spatially divides the network into manageable cells using either fixed-size grids or a density-aware dynamic partitioning. This dynamic mode employs a bottom-up, clustering-inspired algorithm that adapts grid sizes to the local intensity of network equipment, effectively resolving issues of data sparsity and overload. The second stage, AI-Assisted Grid Validation, calculates a comprehensive Correctness Score for each resulting grid. This score provides a quantitative measure of data quality by synthesizing four weighted factors: (1) configurable rule-based attribute checks, (2) connectivity file conformance, (3) topological integrity assessed via advanced network trace functions, and (4) a series of representative graph-theoretic metrics. By generating an intuitive, color-coded map of data health, our framework allows utility providers to precisely localize data quality issues and prioritize remediation efforts. This targeted approach significantly enhances the efficiency of data maintenance, improves the integrity of foundational GIS data for critical power infrastructure, and streamlines integration with essential platforms like SCADA, OMS, and DMS
Evaluating the Accuracy of Pedunculate Oak Tree Volume Estimates Using Static Terrestrial and Mobile Personal Laser Scanning
Full text linkTree volume estimation is fundamental to forest management and inventory, yet traditional methods rely on allometric equations that introduce significant uncertainties due to generalized relationships and measurement limitations. This study evaluates the accuracy of mobile personal laser scanning (PLS) technology for tree volume estimation in pedunculate oak (Quercus robur L.) forests through a controlled comparison framework. Field work was conducted in January 2025 under optimal leaf-off conditions in a lowland oak stand in central Croatia. Three morphologically typical mature oak trees were selected within a single plot to enable controlled comparison while minimizing environmental variability. Data was acquired using PLS Faro Orbis Scanner, emphasizing complete stem coverage from multiple azimuths to support robust SLAM trajectory estimation and minimize occlusion effects. Three principal volume estimation approaches were evaluated: (i) sectioning volume obtained after felling, serving as the operational reference; (ii) PLS Schumacher-Hall volume, computed from LiDAR derived DBH and total height using established allometric relationships; and (iii) PLS Trunk Volume, computed directly from point cloud data using LiDAR360's trunk slicing workflow. Following PLS data acquisition, target trees were felled and bucked into contiguous sections, with length and end diameters recorded for each section to compute reference volumes. The sectioning dataset was treated as an operational reference rather than absolute ground truth, acknowledging potential reconstruction errors due to field conditions and occasional stem breakage. The study reveals important trade offs between measurement accuracy, operational efficiency, and methodological complexity, with sectioning volume providing the most direct measurement approach by eliminating remote sensing processing uncertainties. The research establishes a robust methodological framework for evaluating PLS performance in oak forests while highlighting both significant potential and current limitations of mobile laser scanning for operational forest inventory applications
Law of Sines Based Approach for the Accurate Orientation Determination
Full text linkThanks to developing sensor technologies and autonomous systems, modelling buildings with environmental data and the digital twin approach are becoming increasingly important. This study examines the environmental perception, obstacle detection, and directional correction capabilities of an autonomous vehicle equipped with on board distance sensors integrated with building information modelling (BIM). The vehicle operating around a building detects obstacles through its on board sensors while maintaining its motion parallel to the building wall. Possible angle and direction deviations during movement are determined using the geometry of a triangle formed by three distance measurements taken from specific angles with ultrasonic sensors and the sine theorem. The resulting environmental map from the obtained data will form the basis for digital twin modelling. Instead of using expensive sensors to determine direction, this study uses a method that determines the vehicle's direction using geometry based on the sine theorem and three distance measurements taken from different angles. Field studies have not encountered any adverse events other than erroneous distance measurements due to sensor echo detection. The direction determination method developed in this study is expected to be particularly beneficial for mapping and creating digital twins of buildings. 
Investigating the Role of Geographic Information Systems in Advancing Positive Energy Districts
Full text linkBuildings represent a significant portion of global energy demand, with the European Union accounting for approximately 40% of total energy consumption and 36% of CO₂ emissions. In response to these pressing challenges, the concept of Positive Energy Districts (PEDs) has emerged as a forward-thinking framework. This concept incorporates elements of urban decarbonization and energy sharing mechanisms. The primary objectives of PEDs are to enhance energy efficiency and to advance environmental, social, and economic sustainability within urban settings. Several key projects, including MAKING-CITY, POCITYF, ATELIER, CityxChange, PROPEL, and PED4ALL, are being undertaken to facilitate the large-scale deployment and replication of PEDs.This study aims to examine the definitions and evolving scope of the PED concept, review significant European Union-funded initiatives, and highlight the vital role of Geographic Information Systems (GIS) in the planning and implementation of PEDs. GIS offers robust advantages in the management, analysis, and visualization of spatial data. Furthermore, a comparison of 2D and 3D GIS applications underscores that 3D-GIS enables more detailed and realistic analyses, facilitating accurate evaluations of solar potential, shading effects, and renewable energy capacity at both the building and district levels. The findings indicate that the integration of GIS, particularly through 3D-based analyses, significantly enhances the feasibility, scalability, and decision-making processes associated with PED initiatives