Conference Agenda

The pace of urbanization has been unprecedented. Today, 55 per cent of the world’s population live in cities and another 2.5 billion people is expected to move to urban areas by 2050 (UN, 2018). The UN 2030 Agenda for Sustainable Development gives a prominent role to monitoring the urbanization process. With its synoptic view and large area coverage at regular revisits, satellite remote sensing has been playing a crucial role in urbanization monitoring at regional and global scale. Several methodologies have been developed using Synthetic Aperture Radar (SAR) and/or multispectral imagery to map urban extent globally including the Global Urban Footprint (GUF) and the Global Human Settlement Layer (GHSL). These datasets provide a reliable global map of the urban areas, but they are characterized by low temporal resolution (i.e. every five years) which highlights the need of further research and method development

The objectives of this research are two folds, one is to develop a globally applicable and entirely automatic method to monitor urban footprints using Sentinel-1 SAR and Sentinel-2 MSI dense time series exploiting the Google Earth Engine (GEE) cloud platform, and other is to evaluate derived urban extent for the monitorin of the UN Urban SDG indicator 11.3.1 Ratio of land consumption rate to population growth rate. The innovative aspects of the developed method is to integrate Sentinel-1 and Sentinel-2 dense time series using a totally unsupervised approach. The estimation of the selected urban footprint is performed in several progressive steps. First, the area of interest is divided into mountainous and non-mountainous areas using an available DSM (i.e. SRTM or ALOS World 3D) to take into account the layover and foreshortening of SAR geometric distortions. Then, Sentinel-1 ascending and descending time series are processed in order to enhance the backscatter of stable urban areas and to compute the Sentinel-1 Urban mask using an automatic thresholding procedure. The latest step is to compute a probability urban map combining the Sentinel-1 Urban mask with the Sentinel-2 multi-spectral time series. All available Sentinel-2 images, acquired during the selected sensing period, will be used to compute a cloud-free Sentinel-2 image composite, subsequently we applied a segmentation algorithm to the Sentinel-2 composite, and for each object, we compute several multitemporal spectral indexes statistics (i.e. Min/Max NDVI, Median NDBI, Mean NDWI). Finally, we use a ruleset to estimate the probability urban map combining the Sentinel-1 Urban mask with the computed Sentinel-2 indexes statistics.

To ensure its global applicability, we tested the developed approach in several cities worldwide (i.e. Beijing, Lagos, Milan, Mumbai, New York, Rio and Stockholm) characterized by different urban density and morphologies. We computed the urban footprint in different periods to evaluate the temporal stability of the method and to produce urban footprint time series. The results show that through this method it is possible to obtain high accuracy (kappa higher than 0.85) with respect to the reference data acquired within the EO4Urban project in all cities and in different periods ^[5]. The developed method obtains equal or higher accuracy than GUF and GHSL data in the same area, and a visual comparison shows that the integration of the Sentinel-1 SAR and Sentinel-2 MSI data leads to achieving highly detailed information.

Based on the methodology defined by the UN Habitat, the urban extraction results are being used in the monitoring of Urban SDG indicator 11.3.1 Ratio of land consumption rate to population growth rate in the selected cities. The preliminary results show that timely and reliable urban extraction is essential for the definition of cities and for monitoring Urban SDG indicator 11.3.1.

In the last ten years, the harmonization of rapid urbanization and extremely prosperous economic activity with the air quality and urban land use is the most concerned issue in Chinese urban development policy [1][2]. Accordingly, an urgent and challenging task is to improve the knowledge and understanding of change patterns in human settlements for fast-urbanized megalopolis in South and East Asia. Thanks to the availability of time-series of heterogeneous remote sensing data, it is now possible to explore these changes decoupling those due to urban expansion and those due to increasing economic activities [3]. In this work, we combine multi-temporal the sentinel-1 SAR C-band sensor and Visible Infrared Imaging Radiometer Suite (VIIRS) nighttime sensor (also called the Day/Night Band, or DNB) to explore urban change patterns at the geographical scale of Chinese Megalopolis. The joint use of heterogeneous sensor allows discovering more spatial-temporal features and deeper relationships between urban construction and nighttime-based changes, which indirectly reflect the connections between urbanization and economic development.

Three megalopolis, namely the Jingjinji, the Yangtze River Delta and the Pearl River Delta have been selected, which correspond to the currently most developed and the most densely populated portions in P.R.China. First, Sentinel-1 SAR is used to extract urban extents that ensure the focus of our analysis is in built-up areas at the finest spatial resolution for freely available data sets. To handle big-size data over each Megalopolis, defined as a chain of roughly adjacent metropolitan areas, which may be somewhat separated or may merge into a continuous urban region, the critical preprocessing steps and computations are performed in Google Earth Engine (GEE). Then, data-driven unsupervised classification is used to explore change patterns according to a feature space joining the base and the change images. In this way, both the initial state and the temporal change pattern are considered. To ensure the reliability of unsupervised clustering, GMM, K-method, and DBSCAN are adaptively applied to the same feature space. At last, the 2-dimensional vector analysis are given to interpret the clustering results. Consider the resolution difference between the Nighttime light sensor and Sentinel-1 SAR, upscaling is applied to the SAR images to match VIIRS images at 500-meter resolution.

The vector analysis according to the extracted clusters shows that these clusters are interpretable as meaningful temporal and spatial patterns, although the interpretability of the extracted cluster patterns depends on the classifier performance to different feature spaces. According to the most stable clustering results, change patterns of urban construction and nighttime fundamental facilities are clearly differentiated between core urban and suburban areas, and very new development city zone are singled out and highlighted.ß

Ref:

[1] Wang, Shuxiao, et al. "Effectiveness of national air pollution control policies on the air quality in metropolitan areas of China." Journal of Environmental Sciences 26.1 (2014): 13-22.

[2] Wu, Yanyan, Shuyuan Li, and Shixiao Yu. "Monitoring urban expansion and its effects on land use and land cover changes in Guangzhou city, China." Environmental monitoring and assessment 188.1 (2016): 54.

[3] Frolking, Steve, et al. "A global fingerprint of macro-scale changes in urban structure from 1999 to 2009." Environmental Research Letters 8.2 (2013): 024004.

Urban areas have developed mainly against a socio-economic paradigm ignoring to a large extent the environmental impacts Of particular concern for cities in our days, is the lack of balance between the natural, built and socio-economic environments, leading among others, to the degradation of their thermal environment, in particular overheating. Heat islands/spots within the urban ecosystems have negative impact to human health especially for vulnerable groups, increase energy use for cooling and lead to poor city energy efficiency, intensify energy poverty, deteriorate air quality and result in socio-economic problems in general.

While in vegetated areas, evaportranspiration transfers most of the incoming radiation into latent heat, in built up areas sensible heat is generated, which leads to the strong heat load of urban areas. In addition buildings strongly affect the flow patterns of wind and heat, practically keeping the heat close to the ground.

Mitigation plans to counteract overheating are now developed by several cities around the world, in line to the Sustainable Development Goals of the United Nations and the new Covenant of Mayors for Climate and Energy, which recognizes the role of ecosystem-based mitigation in enhancing urban resilience and providing multiple benefits.

Plans also take advantage of Nature Based Solutions (NBS) in an effort to restore the urban ecosystems and achieve the needed balance between the natural, the built and the socio-economic environments.

The scope of this paper is to assess the factors in support of a Planning Support System (PSS) for the design of the appropriate, science and policy wise, NBS so as to restore urban ecosystems, counteract overheating and improve thermal resilience.

Regional land subsidence is an integrated systematic issue related to multidisciplinary and being of global focus, and has been being a serious threat to the urban infrastructure, high-speed railway and the utilization of underground space, and restricting the sustainable development of society. The study of the regional subsidence evolution in Beijing Plain is of great significance: it is necessary to reveal the regional land subsidence evolution pattern under the background of Integration of Beijing-Tianjin-Hebei and the South-to-North Water Diversion. Furthermore, it can help to realize the scientific regulation of regional subsidence and ensure the sustainable development of regional economy and society, which has a special significance and application prospect. Therefore the MT-InSAR method is used to obtain the regional ground subsidence time series information of the study area in three periods: Jun. 2003 ~ Aug. 2010, Oct. 2010 ~ Nov. 2015, and May. 2015 ~ Jun. 2018. Then equations are established based on the time-overlapping information to complete the fusion of multi-platform time series, the inconsistence between different reference points is solved, simultaneously. The results show that, the maximum subsidence values in Beijing Plain are 690.6 mm, 649.2 mm and 411.7 mm during the three periods, with maximum deformation rates of 100.6 mm/a, 130.0 mm/a and 142.3 mm/a, respectively. For the spatial distribution and the evolution of the land subsidence field, the weighted spatial kernel density analysis, profile analysis, trend-surface analysis and profile-gradient analysis are used to analyze the spatial-temporal evolution characteristics of the land subsidence field. In this case, land subsidence in Beijing Plain are thoroughly analyzed overall distribution characteristics and evolution process. Nine subsidence centers are identified and the subsidence centers are connecting to form a main subsidence area in the northern part of Beijing Plain. The spatial clustering degree of the subsidence in the Beijing Plain indicates an overall heterogeneity in spatial. Moreover, the northern subsidence areas spread along the Nankou-Sunhe fault, and is cut into several subsidence centers by active faults, indicating that the regional geological structure has obvious control effect on the spatial distribution of land subsidence areas. The evolution of the subsidence field: the northern subsidence areas spread along the northwest-southeast direction, and then expands to both the east and west sides. Then through the distribution of subsidence areas and groundwater funnel, the InSAR based time series and the monitoring well based groundwater level changes, the correlations in spatial and responses between land subsidence field and groundwater flow field are analyzed. The results show that the subsidence center in the northern Beijing Plain is consistent with the groundwater drop funnel in spatial, with a similar downward trend over the whole observation time. Through the analysis of well based results located in different areas, the long term groundwater exploitation in the northern subsidence area has led to the continuous decline of the water level, resulting in the inelastic and permanent compaction; while for the monitoring wells located outside the subsidence area, the subsidence time series show obvious elastic deformation characteristics as the groundwater level changes.

Cities worldwide experience enhanced heat stress, as a result of the impact of their surface properties and geometry on the surface energy balance. Land surface temperature (LST) from satellite thermal imagery can provide an overall view of extended urban areas, assisting in the identification of thermally vulnerable sites for mitigation responses. Several studies have demonstrated the relationship between LST and vegetation fraction within cities. Here, the Sentinel-3 SLSTR Level-2 LST product is used to examine the synergy between several two- and three-dimensional properties of the urban morphology (e.g. building height, canyon aspect ratio) and LST, for summer conditions. Results, demonstrate that at daytime open impervious urban areas experience higher temperatures than densely built neighborhoods. The contrasting thermal patterns of nighttime imagery reveal the surface heat island development cycle, and indicate the potential caveat of using solely LST as a map based planning tool, due to its dependence on satellite overpass timing. Furthermore, the surface temperature distribution is examined in the context of Local Climate Zones (LCZs); in general, a considerable differentiation among LCZs is found, although local circulations exhibit a stronger control on coastal zones

Benefit from steering antenna beam, TOPS (Terrain Observation by Progressive Scans), the default imaging model of Sentinel-1, can obtain wide coverage with azimuth-invariant SNR and at the same time avoid scalloping. Meanwhile, the introduced Doppler centroid difference higher than 5000 Hz requires a stringent co-registration accuracy to prevent phase discontinuities over burst boundaries. For multi-temporal analysis, it is even harder to achieve the required accuracy. The residual mis-registration can bias the extracted geophysical inversion parameters such as surface deformation. To improve the co-registration accuracy, a co-registration approach based on Floyd-Warshall algorithm and Enhanced Spectral Diversity is proposed. For further improvement in low coherence regions, we present a coherence estimator by combining two consecutive bursts SLC samples. The performance of the presented approach is validated by two low coherence scenes.

PolSAR data has become the significant data source in urban research. However, the widely used methods extract features at the expense of spatial resolution loss. A PolSAR feature optimization approach is addressed. The new method relies on the adaptive selection of homogeneous samples, both polarimetric and spectral characteristics are taken into account. Those homogeneous samples are first applied to suppress the speckle and to refine the feature estimators afterwards. The comparison with the state-of-the-arts using real-world datasets shows that more accurate PolSAR features with well-preserved edges can be obtained over textural regions.

Urbanization is one of the main factors to cause the land use change in the world, the urban area is rapidly expanding with the accelerated process of urbanization, so the knowledge of urban distribution can provide a reliable technical and decision-making basis for urban planning and development potential evaluation. Remote sensing technology has played an active role in the extraction of urban land, and the urban light at night can be used to analyze the city expansion trend and development potential through light intensity and change rate. In this study, cities in Southeast Asia were taken as the research objects, and landsat8 OLI remote sensing data was taken as the main data source. Firstly, spectral characteristics of different objects were analyzed, and the thresholds of normalized difference vegetation index (NDVI) and normalized difference water index (NDWI) were determined to remove the vegetation and water. The difference between the remaining objects were enlarged after the removal of vegetation and water, the normalized difference building index (NDBI) and the minimum distance classification method were adopted to eliminate the unplanted farmland and bare land. The results showed that this method was simple and capable to extract the impervious land step by step and the accuracy was above 85%. The slope of light change was calculated based on the light data of Defense Meteorological Satellite Program (DSMP) between 2000-2013, in order to analyze the development trend and potential of urban and its surrounding area. The results showed that the urban light index increased at an average annual rate of 0.75 in the built-up area and the surrounding area grew at a high rate of 1.03 per year in the past years from 2000 to 2013, which showed a high density and quick growth rate. It still maintained a strong growth trend and had great potential and space for development.

Over the past two decades, there has been substantial urban growth in Stockholm, Sweden. As a result of accelerating urbanization, Stockholm is now the fastest-growing capital in Europe and it is expected that the Greater Stockholm area will more than double in population by 2045, to 4.5 - 5 million people. The Swedish government has recently taken steps to ensure sustainable management of its green and blue resources in urban areas by requiring all counties to draw up regional plans for their green infrastructure. Using Sentinel-2 and SPOT 5 images, this research investigates the evolution of land cover change in Stockholm County between 2005 and 2015 with a particular focus on what impact urban growth has had on protected green areas, green infrastructure and urban ecosystem service provision.

One scene of Sentinel-2A MSI imagery from 2015 and ten scenes of SPOT 5 imagery from 2005 over Stockholm County were selected for this study. These images are classified into 10 land cover categories using an object-based SVM classifier with spectral, shape and texture features as inputs. The classifications are then used in calculations and comparisons to determine the impact of urban growth in Stockholm between 2005 and 2015, including generation of land cover change statistics, urban ecosystem service provision bundles which include spatial configuration information and evaluation of impact on legislatively protected areas as well as ecologically important habitat networks.

Preliminary results indicate that Urban areas increased by15% or approximately 116 km² while non-urban land cover, mainly agricultural areas and green structure, decreased by just under 4%. The increase in urban areas is just over 2% of the total county land area. More specifically, the results suggest that urban areas may soon overtake agricultural areas to become the second largest land use/cover category in the county landscape after forest. The largest increases in urban areas and significant losses of green structure occurred mainly in the northern and southern outskirts of the county in the rural-urban fringe, with the exception of two municipalities close to Stockholm city which also experienced significant urban growth.

In terms of ecosystem service provision, notable decreases occurred in temperature and global climate regulation, air purification, noise reduction and recreation, place values and social cohesion. Urban areas within a 200m buffer zone around the Swedish EPA’s nature reserves in Stockholm County increased by 16% over the decade, with several examples of new urban areas constructed along the boundary of nature reserves. Further research will include evaluation of important ecological networks in Stockholm county, such as its regional green wedges and broadleaved forest distribution, to see how these have been affected by the urban growth. The results of this study can assist policymakers and planners in their efforts to ensure sustainable urban development and natural resource management for the Stockholm region.

Over the past several decades, the unprecedented pace of urbanization and socioeconomic development in China has placed great anthropogenic pressures on inland surface water quality. In recent years, continuously increasing environmental investments have been undertaken to control pollutant discharge and improve inland water quality across the entire country. However, the quantitative response of water quality to both increasing human pressure and effort is less well understood, particularly for a large-scale region with diverse driving factors. In this study, we use satellite-derived nocturnal radiance signals as proxy measures for both the negative (using the product of lit area and population size) and positive (using the area-weighted magnitude of nighttime lights) effects of human activity to evaluate how water quality changes over time in response to anthropogenic disturbances. Our method clearly demonstrates the extended application of remotely sensed data of nighttime lights with dual actions, particularly in the absence of direct observations of socioeconomic variables due to their consistent, timely and spatially explicit proxy measures for diverse human activities.