To assess urban heat risk, a knowledge graph will be employed as a structured framework for data representation and analysis. This graph consists of nodes that encapsulate information considering to diverse concepts within the urban environment. These concepts include data aggregated at various administrative levels and points of interest strategically distributed throughout the territory, such as climate shelters.
At the granular level, the data encompasses a wide array of parameters originating from different sources. For instance, at the building level, it encompasses comprehensive details such as building characteristics (status, typology, main usages, accessibility, materials), cadastral data, real electricity and gas consumption, thermal energy demand, and availability of tourism establishments, among others. Similarly, at the census tract level, the data incorporates socioeconomic and sociodemographic information, coupled with geographically-aggregated meteorological and air quality data. Furthermore, at the postal code level, the data integrates detailed electricity load curves and mobility patterns.
The connections (edges) established between these distinct node typologies play a pivotal role in facilitating the development of advanced geospatial data analytics modules. These sophisticated modules enable the accurate estimation of a climate vulnerability index, providing valuable insights into the urban heat risk dynamics and potential mitigation strategies.
To integrate machine learning techniques within the knowledge graph, an endeavor is made to predict urban heat patterns, drawing insights from historical data and identifying underlying trends. Leveraging the knowledge graph's capacity to store and manage heterogenous data, in tandem with specialized neural network techniques, this integration offers a platform for advanced analytical reasoning, simulation, and accurate forecasting, effectively capturing the spatial and temporal components inherent in the data. Through the assimilation of diverse datasets, machine learning algorithms discern intricate relationships and patterns in urban heat dynamics, considering both spatial variations across the urban landscape and temporal trends over time. The knowledge graph's ability to link and contextualize spatially dispersed information enables comprehensive analysis of the complex interplay between environmental factors and urban heat patterns, while the temporal aspect ensures the model accounts for dynamic changes in urban development and climate conditions. By incorporating both spatial and temporal dimensions, this scientific approach has the potential to revolutionize urban heat risk assessment, empowering data-driven decision-making for effective urban planning, resource allocation, and climate adaptation strategies amid the challenges posed by increasing urban heat.
The BIGG ontology is the composition and reuse of several legacy ontologies regarding buildings, energy and environment, which includes the urban and territorial extension that is being developed in the framework of the Climate-Ready Barcelona project.
https://github.com/BeeGroup-cimne/BeeOntology.git