Spatio-Temporal Modeling Of Built Environment Stock Based On Multi-Source Geographic Data And Its Application In Environmental Effects Analysis

The built environment stock,including buildings and infrastructure,serves as the material foundation for urban development and human well-being.The construction,operation,maintenance,and demolition of the built environment stock require a large amount of material and energy consumption.With the continued growth of urbanization and population over the next few decades,it is expected that a significant amount of materials will be further used for urban expansion.Therefore,understanding the built environment stock and associated material flows is essential for benchmarking and informing cities’ circular,low-carbon,and sustainable development.Previous studies on built environment stock generally adopted a top-down quantitative approach,but such studies have difficulty in analyzing the physical contents and spatial characteristics.In recent years,studies on built environment stock using bottom-up approach have been extended to longer time periods,larger spatial scales,and higher spatial resolutions,but progress of this studies were limited due to the large amount of labor required for data collection and processing.Previous studies on fine-grained mapping of built environment stocks often focus on an urban area without consideration of temporal dynamics,which hinders detailed understanding of the temporal and spatial dynamics of existing built environment and its related environmental effects.Therefore,this study takes the Danish city of Odense as an example,focusing on the temporal and spatial dynamic modeling of built environment stock,mining multi-source remote sensing and geographic information data to improve stock estimation,combining the historical maps to quantify the spatially and temporally refined built environment stocks,and finally discussing the role of spatio-temporal dynamic analysis of built environment stock in environmental impact analysis.The research objectives and important conclusions are:Developing a method for estimating the built environment stock based on multiple data sources and machine learning.The method combines high-resolution satellite imagery(Google Earth),spaceborne laser altimetry data(ICESat-2),and crowdsourced geographic information data(Open Street Map)to quantify the 61,616 building’s archetype(including height,usage type,and construction year)within Odense city,and estimate the stocks of the three built environment elements: buildings,railways,and roads.The results show that the average absolute error of predicting building heights using the derived features from image and geographic information data is 2.39 m,with an overall classification accuracy of0.90 and 0.75 for building usage type and construction year,respectively.The final stock prediction based on these building archetypes prediction are differ from the reference values by only 1.24%.The results demonstrate that combining multiple remote sensing and geographic information data has significant advantages in accurately describing building archetypes on a large scale,and can be used as an effective and low-cost method for quantifying fine-gained built environment stock.Combining historical maps to quantify the spatially and temporally refined stocks of buildings and infrastructure and developed a novel indexing method to track the construction,demolition,and renovation for each building across various historical snapshots,with a case study of Odense,Denmark,from 1810 to 2018.The results show that the built environment stock in Odense has increased from 5 million tons two centuries ago to 56 million tons in 2018,with buildings contributing the most,but its share has decreased from 85% in the early days to 68% today.The per capita building stock has increased from 80 t/cap in 1810 to 279 t/cap in 2018.In the 12 selected historical snapshots,the original 596,509 building records can be effectively traced and dissolved into 112,820 through the indexing method.These results demonstrate the practical value of historical geographic data in the large-scale reconstruction of the evolution process of built environment stock,which can provide important scientific basis for public and governments in urban spatial planning,related resource/waste management,and climate strategy formulation.Analyzing the building potential waste and accumulated embodied emissions in Odense based on the spatio-temporal dataset of built environment stock.The results show that buildings constructed before 1950 had a lifespan ranging from 117 to 464 years,which reflects the influence of external factors such as social and economic backgrounds on the built environment.Based on this building lifespan estimation,the estimated building waste is range from 1.48 million tons to 5.32 million tons in next 30 years.In calculation of embodied emissions from building-use concrete in Odense since 1960,the predicted value derived from spatio-temporal dataset is of 36% higher than that from one-snapshot dataset.These results demonstrate the effectiveness of spatio-temporal dataset in analyzing environmental effect.Spatio-temporal stock dataset can be used in spatial planning,building design,material use,and other aspects,that providing scientific reference for accurately quantifying the historical carbon emissions cost of cities and allocating future carbon emission space rationally.