This paper presents an exploratory narrative review of machine learning (ML) applications in Public Participation Geographic Information Systems (PPGIS)-related studies. Based on a structured Scopus search and screening process, the review examines how ML techniques are used to process and interpret collaborative spatial data and identifies associated challenges. The findings show that supervised learning methods, including random forest, boosted regression trees, artificial neural networks, and deep learning, are most commonly applied, while unsupervised methods are used to detect spatial and perceptual patterns. These approaches can strengthen PPGIS by supporting the analysis of complex citizen-generated datasets and linking participatory inputs to spatial decision-support processes. However, concerns remain regarding data representativeness, participant bias, spatial uncertainty, privacy, and model interpretation. The paper highlights the need for more transparent, evidence-based, and explicable PPGIS frameworks, which would enable the use of ML to support participatory spatial decision-making.
Air quality is a major public health concern, while particulate matter (PM) measurements remain limited by the spatial coverage and high operational costs of reference monitoring stations. This scoping review synthesises the literature on image-based PM estimation from outdoor photos as a potential low-cost approach to increase PM concentration sampling coverage. Methods are classified by input data requirements and modelling strategy, with summaries of performance and implementation resources. Results suggest that deep learning models generally exhibit the strongest predictive capability, although cross-study comparisons remain indicative, as they are limited by the availability of comparable metrics and by differences in geographic application contexts. Machine learning and physics-based methods are generally less accurate but more interpretable and computationally efficient, while hybrid approaches offer a promising trade-off between accuracy and operability. Nevertheless, most applications remain below reference regulatory accuracy levels. Training datasets are often limited in size and coverage, benchmarks are lacking, and code and data are not always available. Future work should prioritise multi-site datasets, improved hybrid models, and open resources to support reproducible model assessment and reuse.
State-level Spatial Data Infrastructure (SDI) is increasingly important for geospatial data sharing, inter-agency coordination, and evidence-based decision-making. However, its financial value remains under-measured, particularly at the sub-national level. This paper evaluates the economic value of SDI implementation in Perak, Malaysia, using cost-benefit analysis and return-on-investment assessment across five SDI pillars and five agency clusters. A mixed-method approach combined focus group discussions, site visits, and questionnaire data. Annual costs and benefits were estimated to calculate net present value (NPV), benefit-cost ratio (BCR), and return on investment (ROI), with risk-adjusted benefit values included. The conservative scenario produced limited returns (NPV = -RM975,326; BCR = 0.56; ROI = -44
A comprehensive morphometric analysis was conducted for the Anambra–Imo Basin, an energy-rich inland sedimentary basin of significant environmental and economic importance. Despite prior morphometric characterizations, no validated flood vulnerability index exists for this basin, and parameter interaction effects in morphometric FVI models remain unquantified in the literature. Fifteen morphometric parameters spanning linear, areal, drainage texture, and relief dimensions were derived from Shuttle Radar Topography Mission (SRTM) data and integrated into a fuzzy-AHP Flood Vulnerability Index (FVI), with robustness assessed via Monte Carlo–Sobol' global sensitivity analysis and validation against historical flood data through receiver operating characteristic (ROC) curve analysis. The basin's elongated morphology, characterized by a form factor of 0.32, a mean bifurcation ratio of 3.73, low drainage density (0.57 km/km2), and a large rho coefficient (4.0), indicates attenuated flood response and substantial hydrological storage capacity. The basin returned a composite FVI of 0.327 (low vulnerability), with high-hazard zones confined to 8.3
This study evaluates semi-decadal variations in rice cultivation in the regions of Indo-Gangetic Plain (Punjab, Haryana and Western Uttar Pradesh) during 2003 to 2023. The proposed framework integrates time-series MODIS-derived optical data with machine learning classifiers—Support Vector Machine (SVM), Gradient Tree Boosting (GTB), and Random Forest (RF) for mapping rice acreage. Sentinel-1 SAR data, in conjunction with Sentinel-2-derived indices, were employed to generate high-resolution rice maps and compare them with optical-based classifications. Results reveal persistent rice dominance in Punjab and acreage expansion in Haryana and Western Uttar Pradesh. RF and GTB had strong agreement with Directorate of Economics and Statistics (DES) records (R2 ≥ 0.89), with RF having highest accuracy in 2023 (R2 = 0.95). The SAR-derived rice map showed strong spatial agreement with optical-based classifications and improved delineation of fragmented rice fields. The study demonstrates the potential of integrating optical-SAR data for reliable rice acreage mapping and agricultural planning.
Rapid urban growth in Southeast Asia has increased the need for reliable walkability assessment, yet existing methods often rely on proprietary tools and do not scale well across large, fragmented cities. Results are also sensitive to how spatial boundaries are defined (the so-called modifiable areal unit problem problem). This paper introduces an open-source Python repository for large-scale walkability analysis using hexagonal grids and network isochrones. The approach is tested in Jakarta, Manila, Ho Chi Minh City, and Phnom Penh using thirteen indicators at multiple spatial resolutions. Results show that network indicators remain stable across scales, while amenity access and topography vary strongly. The repository provides a reproducible and transparent framework for scale-aware walkability analysis in rapidly changing urban contexts.
This study evaluates the AU20 Mobile Terrestrial LiDAR system’s performance at operational speeds of 60 km/h on a 15.2 km highway section. Addressing research gaps in high-speed data collection and long-distance base station stability, the methodology employs Post-Processed Kinematic (PPK) positioning validated against 2147 points. Results demonstrate a 99.23
This study examines the imprint of cyclone Dana on chlorophyll-a (chl-a) variability in the coastal waters of the northwestern Bay of Bengal. In situ observations and satellite/model-derived biophysical parameters were combined to characterize the upper-ocean responses. Dana exhibited a low translational speed during its intense phase, which amplified the Ekman Pumping Velocity, promoting upwelling of cooler, nutrient-rich sub-surface water. This fertilization produced approximately a two-fold increase in chl-a concentrations after the cyclone, accompanied by a sea surface temperature drop of about 2.2 °C indicative of vertical mixing and upwelling. Phytoplankton abundance rose in the post-cyclone phase, although diatoms dominated throughout. A synthesis of previous Bay of Bengal cyclones confirmed a multifold post-cyclone rise in chl-a. These findings highlight the influence of low translational speed in modulating cyclone-driven biophysical coupling within tropical coastal ecosystems.
This paper presents a framework for automatically generating interpretable LoD2 indoor 3D models from unstructured point clouds. The method treats indoor modeling as a cartographic generalization problem, introducing a 3D corpus of geospatial feature classes that represents indoor environments as parsimonious compositions of geometric primitives organized by spatial grammar and compatible with IndoorGML. A sparse convolutional neural network simplifies, and aggregates point clouds by learning recurring geometric primitives and their spatial relationships, reducing data complexity while preserving essential structural information. Experiments on terrestrial LiDAR, BLK360, and RGB-D datasets achieved up to 97
Floods are among the most recurrent hydrometeorological hazards in South Asia. The changing monsoon dynamics and reservoir operations influence downstream flooding. This study addresses three key aims: (i) to quantify the spatial extent of inundation during the 2025 flood event in the Sutlej Basin using multi-temporal Sentinel-1 Synthetic Aperture Radar (SAR) data, (ii) to assess land-cover exposure to flooding, and (iii) to evaluate the combined influence of extreme precipitation and reservoir operations on transboundary flood propagation. The flood extent derived from VH-polarised Sentinel-1 images indicated a total inundation of 2693 km² (2.3
This study developed a hedonic price model for Xaythany District, Vientiane, Lao PDR, utilizing key road-related variables for land valuation. To account for unobserved locational dynamics in this data-scarce context, residuals were spatially interpolated to generate a Locational Value Response Surface (LVRS), subsequently integrated into the mass valuation framework. Although road variables significantly influence land values, significant residual spatial autocorrelation revealed omitted locational factors. Incorporating the LVRS substantially reduced this autocorrelation and improved model performance across key indicators, including root mean squared error (RMSE) and mean absolute percentage error (MAPE). Notably, the coefficient of dispersion (COD) decreased from 27.199
Public bicycle sharing (PBS) systems play a critical role in strengthening first and last mile connectivity to metro networks, yet their implementation in Indian cities remains fragmented and weakly guided by evidence. This study proposes a rigorous data driven framework to prioritize metro stations for PBS deployment in Bengaluru. A total of 66 operational metro stations were evaluated using spatial, demographic, operational and environmental indicators. Principal Component Analysis was applied to confirm indicator independence, followed by hybrid weighting using the AHP and CRITIC. Station suitability was assessed through an ensemble of five multi criteria decision making methods comprising EDAS, MARCOS, COPRAS, WASPAS and CoCoSo, supported by consensus ranking, sensitivity analysis and spatial zoning. The results indicate that six principal components explain 79.73
Landslides are significant natural hazard in the Tehri Garhwal District in Uttarakhand, India, posing risks to human life, infrastructure, and environment. As the district ranks second among 147 districts based on the landslide index, accurate spatial modelling of landslide susceptibility is crucial for effective risk management. This study presents a comprehensive comparative assessment of machine learning (ML) and deep learning (DL) models for spatial landslide modelling. The study further examines the varying correlations between conditioning factors and the occurrence of landslides. The models were trained and tested using a 70:30 split, with 17 landslide conditioning factors and 1,600 landslide locations. Model performance was assessed using the Area Under the Receiver Operating Characteristic Curve, mean absolute error, and root mean square error. The novelty of this study lies in the integrated evaluation of conventional machine-learning and deep-learning models under a unified framework using the most recent and updated landslide inventory and conditioning datasets for the study area. The accuracy of the deep learning neural network models was the highest (90.31
Effective wildfire susceptibility mapping is essential for reducing ecological degradation and socio-economic losses in fire-prone landscapes. This study develops an uncertainty-aware modelling framework for the Hyrcanian forests of northern Iran by integrating Random Forest (RF), Support Vector Machine (SVM), and Multilayer Perceptron (MLP) with Dempster–Shafer theory. Fourteen environmental and anthropogenic predictors, together with satellite-derived fire occurrence data (2000–2024), were employed to model spatial fire susceptibility. Among individual models, MLP achieved the highest predictive performance (AUC = 87.25
This study examines eight crime types—murder, rape, robbery, aggravated assault, burglary, larceny, motor vehicle theft, and arson—across counties in the United States from 2000 to 2014 using ArcGIS-based spatial analysis across three five-year intervals. Thematic mapping revealed the highest crime intensities during 2000–2004, with most crime types—except rape—declining through 2014. Larceny, burglary, and aggravated assault consistently recorded the highest rates. Hotspot analysis identified spatial clustering across all crime types, with burglary and aggravated assault forming the most persistent hotspots. In contrast, rape exhibited a distinct spatiotemporal pattern with a marked decline in hotspot counties. Geographically, persistent hotspots were concentrated in the South and extended into the West and Southwest, whereas cold spots remained centered in the Northern Plains and Central Midwest, particularly the Dakotas, Nebraska, and Iowa. These findings provide valuable insights for policymakers by identifying regional vulnerabilities and informing targeted long-term crime intervention strategies.
The 15-minute city framework was developed for walking-centric cities and performs poorly in dense South Asian contexts where informal transport—auto-rickshaws and similar intermediate public transport (IPT) modes—is structurally integral to daily mobility. This paper extends the framework to Varanasi, India (1.2 million residents, 90 administrative wards) by quantifying ward-level accessibility to four public service facility categories across five transport scenarios using a two-step floating catchment area (2SFCA) method applied to road network and GTFS data. Ward rankings are derived via TOPSIS, weighted by environmental cost and local travel behaviour; the Next Proximity Index (NEXI) provides an independent proximity benchmark. Walking NEXI scores range from 27.4 to 59.2 across wards; car NEXI from 83.8 to 100. TOPSIS and NEXI rankings diverge systematically, confirming that geographic proximity and multimodal service reach are distinct accessibility dimensions with different policy implications. The 2SFCA-TOPSIS-NEXI framework is replicable across other South Asian cities using open data.
This study examines how artificial intelligence and machine learning are being used in townscape assessment and urban design research. A PRISMA-style Scopus review identified 72 studies published between 2007 and 2025. Structural Topic Modelling was used to identify thematic clusters, trace temporal patterns and align topics with nine townscape attributes. The final K = 8 model identified two mature topics: image-based urban modelling and LiDAR/point-cloud vegetation mapping. Urban clustering, landscape indicators, land-use/public-space analytics and street-view environmental analysis appeared as developing topics. Affective/generative design and urban network prediction remained emerging or fragmented. Topic–attribute mapping showed stronger coverage of form, vegetation, connectivity, land-use and public realm. In contrast, topography, water, skyline, enclosure and architectural style remained weakly operationalised. Robustness checks supported the broad stability of the topic structure. The study contributes TownscapeBench v0.1 as an operational pilot reporting standard for reproducible AI-enabled townscape assessment and future benchmarking.
Activity-based travel modeling seeks to explain travelers’ spatial and temporal decision-making processes. While many studies examine destination choice or activity duration separately, joint modeling of destination-duration decisions remains relatively limited, particularly when latent behavioral factors are considered. This study develops a cross-nested logit (CNL) model to jointly analyze shopping destination and activity-duration choices while incorporating latent attitudinal and lifestyle variables. The empirical analysis is based on a survey of 1810 shoppers collected at eight major shopping centers (four clothing and four grocery malls) in Tehran, Iran. The proposed modeling framework captures correlations among temporal and spatial alternatives while preserving the closed-form properties of generalized extreme value models. Results indicate that latent variables, socio-demographic characteristics, travel time, and transport mode availability significantly influence shopping destination-duration decisions. The findings further reveal behavioral differences between clothing and grocery shopping activities, particularly regarding shopping duration patterns and the role of lifestyle and attitudinal effects. The study contributes to the activity-based travel behavior literature by providing a joint behavioral framework for analyzing spatiotemporal shopping decisions under heterogeneous traveler preferences.
Remote sensing (RS) has evolved from aerial photography mounted on kites and pigeons to satellite-based platforms, and autonomous artificial intelligence-driven unmanned systems. In the era of Industry 5.0, or harmonizing smart machines with human influence, the role of RS remains unclear due to limited research on this domain. Hence, this paper traces the influence of Industry 5.0 in RS through a bibliometric and systematic review. A literature search covering 2014–2024 using selected keywords, Boolean, and wildcard operators across major academic databases yielded 23 eligible articles for analysis. Results indicate a growing publication trend, with Asia (n = 16; 69.57
Aegean and Mediterranean countries have been experiencing many forest fires which caused extensive ecological damage and posed significant risks to habitants in recent years due to extreme temperature anomalies. Post-fire analyses are essential for rapidly assessing wildfire impacts and informing future mitigation strategies. This study investigates the effects of the Çeşme wildfire in İzmir, Türkiye, which began on July 2, 2025, and lasted four days, using Sentinel-2 and Sentinel-5P satellite imagery. Changes in vegetation conditions were examined using normalized difference moisture index (NDMI), normalized burn ratio index (NBR), normalized difference vegetation index (NDVI), and green chlorophyll vegetation index on the Google Earth Engine cloud-based platform. Burned area extent was quantified using differenced normalized burn ratio (dNBR) while changes of nitrogen dioxide (NO2) and carbon monoxide (CO) gases were extracted from Sentinel-5P TROPOspheric Monitoring Instrument satellite data. The study results demonstrated that 96.21 km2 was burned around Çeşme. The statistical relationships of NBR with NDMI and NDVI were lower after the fire. The average NDVI decreased from 0.39 to 0.25 and the average NBR decreased from 0.19 to 0.09, indicating the damage caused by the fire to the vegetation. The study findings highlighted the huge amount of forest destroyed due to wildfire. It has been shown that NO2 and CO gases were released into the atmosphere at high rates during the wildfire. The study represents a powerful application of spatial information science in disaster management through the integration of various remote sensing indicators.
