Remote Sensing Imagery: A new frontier for Urban Analytics?

在紧张的经济气候下，对建筑环境如何随时间变化有最好的了解比以往任何时候都更重要，这样政策制定者就可以优先考虑并有效地适应。再加上气候变化，预计到2050年，全球居住在建筑环境中的人口将增加到68%（联合国2018年），全球城市化模式将是异质的（McGee 2010; Kourtit et al. 2014），因此，准确测量城市群内部和城市群之间的变化（包括物理和虚拟）对于智能城市的规划至关重要。衡量城市变化对于应对未来住房、交通、能源系统、教育和医疗保健方面的挑战至关重要。在城市分析领域出现了三种数据来源，即个人携带的移动传感器收集的数据、从网站提取的数据和以开放格式发布的政府数据（Arribas-Bel 2014）。我们认为，遥感信息（RSI）将成为城市研究中一个庞大而广泛使用的新信息来源。RSI包含了丰富的信息，这些信息推动了对社会经济关系的见解（Jean et al. 2016）。RSI在建筑环境中帮助自动化变化检测的潜力是巨大的；与主要的替代方案——实地测量相比，这将既节省时间又节省成本。虽然一些研究已经应用激光雷达来监测建筑环境的变化，但由于无法对建筑物等复杂特征进行分类，粗中等空间分辨率（15-60米）多光谱卫星图像（MSI）的历史应用受到限制（Hegazy和Kaloop 2015）。为了克服这一限制，一些研究将高时间分辨率MSI与高空间分辨率LiDAR数据融合在一起，以建立更好的整体见解（Yan et al. 2014; Wu et al. 2015）。此外，融合方法还具有航空图像水平精度高、激光雷达数据垂直精度高的优点（Alonso-Benito et al. 2016）。例如，Dong et al（2010）将LiDAR与Landsat Thematic Mapper结合起来计算小区域人口估计。此外，空间工业部门正在迅速发展，全球正朝着RSI商业化的方向发展：最近的技术进步实现了与激光雷达类似的空间分辨率，小于100万。该项目的目的是评估应用RSI分析城市随时间变化的潜力。具体来说，我们将比较当前的激光雷达技术、MSI技术或这些技术的融合技术是否是对建筑环境中建筑物和道路等复杂特征进行分类的最佳实践，以及它们如何随时间变化。建立详细的城市演化时空数据库对预测和制定政策具有重要意义。例如，确定城市地区内部和城市地区之间的社会经济和人口多样性。鉴于RSI的空间和时间分辨率不断提高，这样的研究具有前瞻性。

In a tight economic climate, it is more important than ever to have the best possible understanding of how the built environment changes over time, so policy makers can prioritise and adapt effectively. Coupled with climate change, the global population living in the built environment is projected to increase to 68% by 2050 (United Nations 2018) and global patterns of urbanization will be heterogeneous (McGee 2010; Kourtit et al. 2014), so accurate measurements of changes in and between urban agglomerations (both physical and virtual) is critical for the planning of Smart Cities. Measurements of urban change is critical to meet the future challenges of housing, transportation, energy systems, education and healthcare.There are three sources of data that have emerged in the field of urban analytics, which are data collected by mobile sensors carried by individuals, data extracted from internet sites and government data released in an open format (Arribas-Bel 2014). We propose that Remote Sensing Information (RSI) will emerge as a large and thoroughly-used new source of information for urban studies. RSI contains an abundance of information, which has driven insights into socio-economic relationships (Jean et al. 2016). The potential for RSI to help automate change detection in the built environment is massive; it would be both time and cost efficient compared to the main alternative, field surveying. Whilst several studies have applied LiDAR to monitor changes in the built environment, historic applications of coarse-medium spatial resolution (15-60m) Multispectral Satellite Imagery (MSI) have been limited due to the inability to classify complex features such as buildings (Hegazy and Kaloop 2015). To overcome this limitation, some studies have fused high temporal resolution MSI with high spatial resolution LiDAR data to build better overall insights (Yan et al. 2014; Wu et al. 2015). Furthermore, fusion methods have the merits of high horizontal accuracy from aerial images and high vertical accuracy from LIDAR data (Alonso-Benito et al. 2016). For example, Dong et al (2010) combined LiDAR with Landsat Thematic Mapper to calculate small-area population estimates. Further, the space industrial sector is rapidly growing and there is a global movement towards the commercialisation of RSI: recent technological advances achieve spatial resolutions similar to LiDAR, less than 1m.The aim of this project is to evaluate the potential of applying RSI for the analysis of changes to cities over time. Specifically, we will compare whether the current technologies of LiDAR, MSI or a fusion of these technologies is best practice for classifying complex features such as buildings and roads in the built environment and how they change over time. Building up a detailed geo- referenced space-time database on urban evolution is important for forecasting and policy. For example, identifying socio-economic and demographic diversity within and between urban areas. Such a study is forward looking, given the increasing spatial and temporal resolution of RSI.