Search for: All records

Understanding the spatial dynamics of bike-sharing usage is critical for effective urban planning and mobility resource management. In this study, we propose an interpretable deep learning approach to uncover spatial relationships embedded in bike-sharing activities. Specifically, we develop a spatially-adapted SHapley Additive exPlanations (SHAP)-based method to quantify the spatial dependencies between locations in bike-sharing activities and apply it to interpret the predictions of a bike-sharing model. Extensive experiments upon Citi Bike data from New York City in December 2023 reveal that spatial influence does not strictly follow geographic proximity and is anisotropic. Additionally, non-member users exhibit weaker spatial dependencies in their bike usage behavior, resulting in lower short-term predictability compared to member users. Our studies shed deep insights into the spatial dynamics of bike-sharing systems and provide guidance for more effective service deployment and system design. 

Amid accelerating climate change, climate-related hazards—such as floods, wildfires, hurricanes, and sea-level rise—increasingly drive land transformations and pose growing risks to housing markets by affecting property valuations, insurance availability, mortgage performance, and broader financial stability. This review synthesizes recent progress in two distinct domains and their linkage: (1) assessing climate-related financial risks in housing markets, and (2) applying AI-driven remote sensing for hazard detection and land transformation monitoring. While both areas have advanced significantly, important limitations remain. Existing housing finance studies often rely on static models and coarse spatial data, lacking integration with real-time environmental information, thereby reducing their predictive power and policy relevance. In parallel, remote sensing studies using AI primarily focus on detecting physical hazards and land surface changes, yet rarely connect these spatial transformations to financial outcomes. To address these gaps, this review proposes an integrative framework that combines AI-enhanced remote sensing technologies with financial econometric modeling to improve the accuracy, timeliness, and policy relevance of climate-related risk assessment in housing markets. By bridging environmental hazard data—including land-based indicators of exposure and damage—with financial indicators, the framework enables more granular, dynamic, and equitable assessments than conventional approaches. Nonetheless, its implementation faces technical and institutional barriers, including spatial and temporal mismatches between datasets, fragmented regulatory and behavioral inputs, and the limitations of current single-task AI models, which often lack transparency. Overcoming these challenges will require innovation in AI modeling, improved data-sharing infrastructures, and stronger cross-disciplinary collaboration. 

The rapid population growth and unplanned urbanization within Kathmandu Metropolitan City (KMC) have induced land use and land cover (LULC) changes that have exacerbated problems of air pollution and the Urban Heat Island (UHI) effect. These issues, as well as potential mitigations and possible counteractions, are currently under investigation by numerous research communities, resulting in various solutions being put forward including the creation of Urban Green Spaces (UGS). Establishing UGS would increase carbon dioxide extraction, minimizing photochemical ozone formation and liberation, while simultaneously cooling the microclimate of an area such as KMC. Optimized implementation of UGS throughout KMC requires an understanding of and prioritization of locations based on degraded air quality and the UHI effect. Unfortunately, such studies in these areas appear to be severely lacking, which has acted as a catalyst for this study. This research includes prioritization on two different spatial units—(i) at the administrative ward level and (ii) 0.0025° fishnet level. The result identifies the high-need locations where UGS establishment is recommended to mitigate air pollution and the UHI effect. Information obtained also heightened the existing UGS’s current sparsity and deplorable conditions. Findings from this study indicate that the utilization of rooftops are potential locations for new UGS, and enhancement of the existing UGS would prove to be an efficient use of currently underutilized spaces. 

We are currently living in the era of big data. The volume of collected or archived geospatial data for land use and land cover (LULC) mapping including remotely sensed satellite imagery and auxiliary geospatial datasets is increasing. Innovative machine learning, deep learning algorithms, and cutting-edge cloud computing have also recently been developed. While new opportunities are provided by these geospatial big data and advanced computer technologies for LULC mapping, challenges also emerge for LULC mapping from using these geospatial big data. This article summarizes the review studies and research progress in remote sensing, machine learning, deep learning, and geospatial big data for LULC mapping since 2015. We identified the opportunities, challenges, and future directions of using geospatial big data for LULC mapping. More research needs to be performed for improved LULC mapping at large scales.