Collaborative Research: Interactions of Sustainable Urban Design with Gentrification Processes

世界各地的城市旨在促进可持续和有弹性的建筑环境，以公平地减少碳排放，缓解热岛效应，并提高城市宜居性。然而，这些举措可能会提高房价和生活成本，最终通过一种名为绿色中产阶级化的过程使长期居民流离失所。这项研究将通过检查和比较谷歌街景的历史和当前图像以及人口普查局的人口数据，来评估和预测与城市社区各种可持续发展举措相关的绿色中产阶级化。利用人工智能工具，研究小组将确定绿色中产阶级化的物理指标和社会人口统计指标，以分析中产阶级化进程与城市可持续发展倡议之间的关系。这些工具将利用费城作为案例研究和它在过去20年中实施的可持续发展倡议来开发。这些倡议包括绿地开发、城市农业、植树、节能改造、自行车道、公共交通和太阳能安装。这项研究将是朝着解决费城和其他城市环境中的重大社会挑战迈出的重要一步。城市政策制定者和规划者将更好地了解可持续发展政策和计划如何影响士绅化，以及如何减轻其影响和改善公平结果。此外，社区和公共机构将能够更好地分析、预测和解决可持续发展的负面后果，确定最脆弱的社区，并推进公平的可持续发展倡议。在理解城市可持续发展计划(即改善交通、绿地和住房)如何、何时以及哪些项目影响士绅化导致的流离失所方面，存在着严重的知识差距。在这项研究中，研究人员将开发新的模型和方法，这些模型和方法依赖于机器学习的最新进展，以及大量时空和社会人口数据的可用性。研究小组将在城市分析和以建筑环境为中心的预测分析的交叉点开发方法，以预测和绘制士绅化敏感性地图。该团队将把这些预测与城市建筑能源使用、绿地开发和交通系统的模型相结合，以确定由可持续发展计划推动的所有变种和生命周期阶段的士绅化过程。该研究项目将利用人工智能图像识别方法与机器学习算法、城市能源建模和社会人口数据相结合，取得以下三个成果：(I)利用机器学习(ML)算法开发应用于谷歌街景(GSV)图像数据的人工智能计算机视觉方法，以识别和分类绿色中产阶级化指标；(Ii)将社会人口和能源数据与部分(I)开发的GSV-ML模型整合，以评估绿色中产阶级化与可持续干预之间的关系。该集成模型将使用机器学习来量化不同城市绿化特征对社区中产阶级化敏感度的预测能力，并开发研究区域的中产阶级化初步预测；(Iii)基于社会公正和公平关注并防止绿色中产阶级化的可持续城市设计和政策的引出。最终，该项目侧重于预测绿化干预措施影响士绅化进程的方式，以推进更公平的可持续城市政策和计划。该合作项目由CBET/ENG环境可持续发展计划和BCS/SBE人类环境和地理科学计划共同资助。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Cities around the world aim to advance sustainable and resilient built environments that equitably reduce carbon emissions, mitigate heat island effects, and enhance urban livability. However, these initiatives can increase housing prices and the cost of living, ultimately displacing long-time residents through a process called green gentrification. This research will evaluate and predict green gentrification associated with various sustainability initiatives inurban neighborhoods by examining and comparing historical and current imagery from Google Street View and demographic data from the Census Bureau. Using Artificial Intelligence tools, the research team will identify the physical indicators and sociodemographic metrics of green gentrification to analyze gentrification processes vis-à-vis urban sustainability initiatives. These tools will be developed using the City of Philadelphia as a case study and the sustainability initiatives it has implemented over the last two decades. These initiatives include green space development, urban agriculture, tree planting, energy efficient retrofits, cycle lanes, public transit, and solar energy installations. The research will be an important step towards addressing significant societal challenges in Philadelphia and other urban contexts. Urban policymakers and planners will gain a better understanding of how sustainability policies and programs influence gentrification and how to mitigate its effects and improve equitable outcomes. Furthermore, communities and public institutions will be better able to analyze, predict, and address the negative consequences of sustainable development, identify the most vulnerable neighborhoods, and advance equitable sustainability initiatives.There is a critical knowledge gap in understanding how, when, and which urban sustainability programs (i.e., improvements to transit, greenspace, and housing) impact gentrification-led displacement. In this research, the investigators will develop new models and methods that rely on recent advances in Machine Learning and the availability of high-volume spatiotemporal and sociodemographic data. The research team will develop methods at the intersection of urban analytics and built environment-centered predictive analyses to forecast and map gentrification susceptibility. The team will integrate these forecasts with models of urban building energy use, greenspace development, and transit systems to identify gentrification processes, in all its variants and lifecycle stages, that are driven by sustainability programs. The research project will harness artificial intelligence image recognition methods with Machine Learning algorithms, urban energy modeling, and sociodemographic data with the following three outcomes: (i) Development of Artificial Intelligence computer vision methods applied to Google Street View (GSV) image data with a Machine Learning (ML) algorithm to identify and categorize indicators of green gentrification; (ii) Integration of sociodemographic and energy data with the GSV-ML model developed in part (i) to evaluate the relationship between green gentrification and sustainable interventions. This integrated model will use Machine Learning to quantify the predictive power of different urban greening features on neighborhood gentrification susceptibility and develop a tentative forecast of gentrification for the study area; (iii) Elicidation of sustainable urban design and policies that are underpinned by social justice and equity concerns and prevent green gentrification. Ultimately, this project focuses on predicting the ways in which greening interventions impact gentrification processes to advance more equitable sustainable urban policies and programs.This collaborative project is co-funded by the CBET/ENG Environmental Sustainability program and the BCS/SBE Human-Environmental and Geographical Sciences program.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.