Urban Remote Sensing. Группа авторов
conductivity such as steel), together with contribution from re‐radiation of currents induced by incidence electromagnetic fields, multipath effects, diffraction effects, and multiple reflections similar to the Fabry–Perot reflection interactions between the front side of a building and the backside of an adjacent building constructed on two sides of a road. Thus, radar backscatter contains information to represent 3D building volume at the Ku‐band frequency of 13.4 GHz as validated by Mathews et al. (2019). Since X‐band (~10 GHz) is close to Ku‐band, we hypothesize that X‐band SAR backscatter, such as measured by the current CSK, TSX, and TDX and by the future LS1 and LS2, can also capture 3D building volume in cities.
2.4.2.2.1 Initial SAR Findings
Backscatter data can be combined and averaged from multiple SAR scenes to account for both ascending and descending observations as well as for noise reduction. Each backscatter ensemble is made over the 2D areal extent of building footprints on the land surface, for which remote sensing products are available such as the Global Urban Footprint (Esch et al. 2013, 2017, 2018). Initial results for Hồ Chí Minh City and its surroundings derived from CSK SAR data at 3 m spatial resolution (Nghiem and Science Team 2019) could detect complex of large building in District 1 in the city center on west side of the Saigon River while a mostly bare field was found in Thủ Thiêm on the east side of the river where the land was cleared but buildings have yet to be constructed. Other work supports the use of these data for 2D build‐up (i.e. extent, density) monitoring and characterization (e.g. Strozzi and Wegmuller 1998; Thiele et al. 2007; Brenner and Roessing 2008; Schmidt et al. 2010; Esch et al. 2011; Leinenkugel et al. 2011; Wu et al. 2011; Marin et al. 2015; Sorichetta et al. 2020) as well as 3D feature characterization (e.g. Soergel et al. 2003; Brunner et al. 2010; Buckreuss et al. 2018; Qiu et al. 2018) – for example, TDX data were employed for urban footprint delineation (Taubenböck et al. 2012) and to examine forest vegetation height (Qi and Dubayah 2016). In evaluating the utility of a variety of geospatial datasets for urban build‐up at a continentalscale, Li et al. (2020) found SAR data to be the most useful for estimating building height, which further attests to the significance of SAR data for 3D urban analyses.
Complex SAR data (including both real and imaginary parts in complex‐variable numbers, or magnitude and phase in the phasor form) can be used for urban building detection using time‐paring complex coherence method (Paloscia et al. 2019) together with geometricmean multi‐temporal coherence method from multiple CSK SAR scenes over urban areas (Nghiem and Science Team 2019) showing building density on the land surface not only within the city but also over small villages and minor settlements along roads. The Tân Sơn Nhất International Airport with a large area of runways was also clearly identified. Nghiem and Science Team (2019) detected houses built on the other side of the Yên Phụ levee directly adjacent to the waterway of the Red River in Hanoi, Vietnam. While multiple SAR datasets can be used synergistically to have more extensive and frequent observations of urban areas, cross‐validation of SAR data will be necessary to obtain consistent results for long‐term monitoring of urban built‐up change.
2.4.3 RADAR FOR BUILT‐UP VOLUME: IMPLICATIONS
Excitingly, spaceborne radar‐based approaches introduce a potential paradigm shift within the urban remote sensing community because of the ability to liberate remote sensing scientists and their analyses from the constraints of 2D, and do so with use of readily available, free data (unlike most airborne lidar data). Balk et al. (2019) even demonstrated that DSM data can actually capture socioeconomic characteristics, patterns, and status along the decadal urbanization trend of the Great Saigon (including the mega Hồ Chí Minh City), while 2D methods from satellite images may be insensitive to or even misrepresent socioeconomic change due to urbanization.
Realistically, the SeaWinds DSM approach provides moderate‐resolution data products (1 km pixels) only for the years 2000–2009 when QuikSCAT was operating. The scatterometer record can be extended with international satellite scatterometer data; however, a major issue is whether the data can be fully calibrated and made freely available. On the other hand, spaceborne radar data with spatial resolutions several orders of magnitude higher than that of a scatterometer, likely provided only by SAR systems, have potential to revolutionize urban built‐up analyses. Even with the availability of spaceborne lidar data from ICESat‐2 (Moussavi et al. 2014), GEDI (Qi and Dubayah 2016), and Sentinel 3 (ESA 2019), scientists may not have long‐term data records or high enough spatial resolution for intra‐urban analyses, which can be provided by the many SAR platforms now in operation, previously operated, or planned for the future.
2.5 A LOOK FORWARD
With a broader variety of data options now available to remote sensing scientists, comprehensive 3D analyses of urban environments are well within our grasp. This chapter outlines the importance of lidar and radar of many types as data sources to support such efforts. Although lidar remains the best option for building height estimation and built‐up volume extraction, a number of limitations persist related to its high cost and sporadic acquisition. Radar provides an innovative alternative to lidar with the potential to conduct similar analysis at a lower cost and higher repeatability (Gamba and Houshmand 2002; Sportouche et al. 2011; Mathews et al. 2019). Looking forward, SAR data offer the best data option due to the high spatial and temporal resolutions (Bagheri et al. 2018; Geiß et al. 2019) offered by several platforms currently and with more SARs planned. Such high spatial resolution enables intra‐urban analyses that are important for many fields beyond remote sensing. Specifically, estimating global population based on built‐up volumes including diurnal dynamics (Dong et al. 2010; Zhao et al. 2017), examining vegetation (i.e. greenspace) in urban areas and its environmental and anthropological benefits (Alonzo et al. 2014; Ellis and Mathews 2019), and making linkages with other data to explore urban pollution – air, water, etc. (Fang et al. 2015; Masetti et al. 2015; Jacobson et al. 2015, 2019). Significantly, previous (e.g. QuikSCAT), existing/current (e.g. CSM, TDX), and future (e.g. LS1, LS2) radar data sources will enable interdecadal analyses, which was emphasized repeatedly in the “Decadal Strategy for Earth Observation from Space” (NASEM 2018), at the intra‐urban scale – e.g. facilitating 3D characterization of urban typologies: analysis and refinement of urban surface material types such as building materials, impervious, and other surface types. Data fusion, cross‐calibration, and integration are and will continue to be major challenges faced within this area of research for both data product generation and validation.
Other data sources for 3D urban assessment such as Unoccupied Aircraft Systems (UAS), or drones, offer exciting opportunities for low‐cost, very high spatial resolution (imagery and/or lidar), and highly flexible temporal data acquisition (Mathews and Frazier 2017). Civil aviation authorities such as the US Federal Aviation Administration (FAA), though, do not permit UAS flight over highly populated urban areas (although regulations are not static and could change). Avenues do exist, however, to obtain permission to fly in urban areas but not without a number of obstacles.
2.6 CONCLUSION
Lidar and radar data provide opportunities for remote sensing scientists to model our world, specifically our urban environments, comprehensively in 3D. As this chapter has illustrated, lidar data enables quantification of urban built‐up volume as well as examination of its change over time. Likewise, radar data also facilitates 3D observation of urban environments with the potential to do so over larger areas and with increased temporal frequency. However, more work on advancing our 3D analysis methodologies and incorporation of various 3D data sources is needed moving forward. In sum, for a better understanding of urban form and its morphology (i.e. intensity and configuration of built‐up material), the vertical dimension is incredibly important (Wentz et al. 2018; Dong et al. 2019; Taubenböck et al. 2012, 2019). To improve upon our understanding of urban processes, including how urban built‐up volume influences environmental processes such as air circulation and urban heat island effect, we must think three‐dimensionally.
ACKNOWLEDGMENTS
This work partially funded by Michigan Space Grant