Major earthquakes can disrupt regional production, living conditions and ecological functions, reshaping sustainable development pathways long after the event. Quantitatively tracing the evolution of earthquake-affected areas from the pre-disaster to the post-disaster period can deepen our understanding of the long-term impacts of earthquakes and provide support for guiding post-disaster reconstruction toward sustainable development. Using the ten counties most severely affected by the 2008 Wenchuan earthquake, we construct a sustainable development index (SDI) within the production–living–ecology (PLE) framework based on annual county-level data from 2003 to 2023. We integrate recovery trajectory clustering, PELT change-point detection, a triangular coordination model and RF-SHAP analysis to examine long-term recovery, system reconfiguration and key explanatory factors. Results show that SDI declined sharply after the earthquake, then followed four differentiated recovery pathways aligned with the plateau–mountain–hill–plain topographic gradient. Southeastern and piedmont-transition counties recovered faster and maintained higher SDI levels, whereas northwestern counties had strong ecological endowments but delayed production recovery and slower transformation. Recovery was stage-dependent: living functions recovered first, production functions accelerated in the middle and later stages, and ecological functions improved steadily; all county types eventually moved toward stronger PLE coordination. RF-SHAP identifies industrial upgrading as the dominant explanatory factor, with fiscal self-sufficiency and non-agricultural employment also important. Overall, long-term post-earthquake recovery is not a simple return to pre-disaster conditions but a systemic reorganization process in which production recovery, living improvement, ecological protection, and risk governance are jointly transformed into sustainable development capacity during post-disaster reconstruction.
Snow-induced tree falls can damage overhead telecommunications infrastructure, yet inspecting widely distributed assets at high spatial resolution is resource-intensive. We propose a multi-scale framework that screens broad areas while limiting missed damage and then ranks local damage susceptibility. We analyzed 391 damage records collected in Tottori Prefecture from December 2019 to March 2024. Logistic regression was used at 1 km, followed by XGBoost screening at 250 m and XGBoost ranking at 20 m using forest-structure and neighboring-cell features. Nested spatial cross-validation assessed stage-wise performance; alternative strategies and leave-one-winter-out validation assessed robustness. Pooled held-out recall was 0.981 at 1 km and 0.938 at 250 m. Retrospectively, the final 20-m candidate area was reduced by 50.8%, retaining 325 of 357 damaged meshes (91.0%). Within this area, the top-ranked 5%, 10%, and 20% contained 14.2%, 24.0%, and 40.9% of damaged meshes, respectively, and the Very High class had 2.83 times the average damage rate. Across 763,174 meshes evaluated using common outer folds, the HardGate strategy reduced the candidate area by 37.48% while retaining 93.97% of damaged meshes. Although it achieved the highest numerical capture rates among three strategies, paired spatial-block bootstrap confidence intervals indicated no statistical superiority. Leave-one-winter-out validation retained an average of 94.6% of held-out damaged meshes while reducing the candidate area by 44.9%. The framework cannot deterministically predict individual events but can support transparent prioritization of limited inspection resources.
In 2022, the Copernicus Program launched the European Ground Motion Service (EGMS), providing open-access ground displacement products across Europe. This development created the need for methodologies to evaluate these datasets and integrate them into risk assessment frameworks. This study proposes an operational screening workflow for differential land subsidence risk based on EGMS Interferometric Synthetic Aperture Radar (InSAR) displacement measurements. The methodology was developed and tested in Emilia-Romagna region (Italy), an area affected historically by land subsidence since the 1950s, with maximum rates reaching 10 cm/year. EGMS data collected for 2018-2022 were used to estimate hazard levels associated with differential land subsidence. Exposure and vulnerability were assessed using datasets from the World Settlement Footprint (WSF) Evolution and the Global Human Settlement Layer (GHSL), together with a regional dataset from Emilia-Romagna Region to characterize building construction periods, types, and heights. Hazard, exposure, and vulnerability were integrated using a heuristic risk matrix to derive four qualitative risk classes, from low (R1) to very high (R4). The highest-risk areas are concentrated in Bologna and Ravenna. Although most of the region falls within the moderate-risk class (R2), the high- and very high-risk classes (R3, R4) account for approximately 1.6 km2 of the total buildings footprint. The resulting map characterizes the potential spatial distribution of adverse effects associated with differential land subsidence rather than probabilities of structural damage. The workflow provides an operational screening tool for regional-scale subsidence risk assessment, supporting planners and decision-makers in identifying priority areas for field investigations, structural assessments, and long-term monitoring.
Wildfire risk in high-latitude areas (above 50°N) presents significant challenges due to fuel accumulation, low humidity, remoteness, and resource constraints. As wildfire impacts on diverse human communities intensify, expectations grow for governments to recognise and address their needs by supporting wildfire risk reduction. Simultaneously, communities are encouraged to engage in wildfire preparedness and assume greater individual responsibility. However, these dynamics may generate disparities in perceived roles, responsibilities and expectations between governments and communities. This paper applies social contract theory to argue that disasters such as wildfires provide opportunities for political discussion to renegotiate power relations between governance levels and communities. A conceptual approach is proposed to examine tensions between governments and communities, including governance structures, resource distribution, collaboration approaches, and socio-cultural inclusion in wildfire risk management. Using Yellowknife, Northwest Territories (NWT), Canada, as a case study – grounded in the experiences of the 2023 wildfire evacuation – we examine how stakeholders from different levels of government, non-governmental organisations, self-governing organisations, private-sector organisations and other community groups participated in and influenced the development of wildfire risk reduction. Data were collected through semi-structured interviews with representatives from multiple government levels, NGOs, and key community stakeholders to explore how they work in preparing for future wildfire risks. This approach, combined with empirical findings, reveals tensions and conflicts in the transformation towards decentralised wildfire risk management and underscores the need for an inclusive, fair and long-term-orientated approach to managing wildfire risks.
Flood risk communication requires methods that translate technically robust flood information into spatial representations that can be interpreted in relation to buildings, roads, and other urban assets. Existing 3D visualisation approaches are often technically sophisticated but are not consistently demonstrated for routine GIS practice, while other approaches emphasise public engagement without documenting a GIS-based workflow. This study develops a documented GIS-based workflow for three-dimensional flood visualisation in an urban fluvial setting. The framework integrates LiDAR-derived terrain and building information with two forms of flood-depth data: indicative terrain-driven hydrologic surfaces generated using ArcHydro and authoritative hydraulic model outputs supplied by Auckland Council. These datasets are translated into neighbourhood- and property-scale 3D visualisations incorporating building exposure, road impacts, severity-based asset information, and potentially safe areas. The workflow is demonstrated for Te Auaunga, Auckland, New Zealand, and evaluated by nine professionals with experience in flood risk management, GIS, or related fields. Expert responses indicated high perceived understandability and stability (mean = 4.56 for both), together with favourable assessments of plausibility and practical usefulness. Hydrologic and hydraulic outputs showed different visual characteristics, with terrain-driven surfaces producing smoother inundation patterns and hydraulic outputs retaining greater local spatial variability. The findings support the feasibility of integrating heterogeneous flood-depth datasets and urban assets within a standard GIS environment, offering a practical complement to existing flood-information platforms. The present assessment represents an initial professional evaluation of the framework; direct evaluation with non-technical community users is required before conclusions can be drawn regarding public risk perception.
