Poster Session

Multi-scale classification is an important tool for urban area mapping, because the objects in an urban scene may have very different spatial scales. In technical literature, this idea is usually performed by means of the same classifier (or ensemble of classifiers) at multiple resolutions, working on sets of features at multiple scales. Although this may be very efficient approach for peculiar situations, it is not generally applicable, as it does not adapt to (very) different scenes. Furthermore, different classification methodologies lead to different results and, consequently, the accuracies may vary for different applications. Accordingly, it can be useful to design a procedure based on the possibility to assemble specialized multi-scale classification chains automatically, according to the classes to be recognized and their peculiar scales or features. In this paper, we present a methodology that combines different processing chains (made by a feature selection and a classification step) and automatically adapts to the spatial (and spectral) properties of the classes available in a urban scene. To this aim, a hierarchical hybrid decision tree architecture is exploited. Typically, in homogeneous decision trees, at each node the same algorithm is used for separation between groups of classes. In our case, the most useful processing chain, composed by the most suitable feature set and the most efficient classifier, is selected per each node. This selection is performed by computing an intermediate accuracy assessment for each processing chain. Only the feature set and classifier pair that produces the best result is selected and assigned to the node. The procedure is then repeated removing the already identified class/classes until all the nodes are identified. In the end, the final classification map is obtained applying the designed framework to the whole dataset. The first prototype of this framework exploits four different sets of features and four different classifiers. The proposed method has been validated on two multi-spectral datasets covering the city of Pavia (Italy) and Xuzhou (China), and having different spatial resolutions. The results have been analyzed using statistical hypothesis tests (i.e. matched-pairs t-tests), showing the usefulness of the proposed method when compared to single processing chains.

Rome is at the same time a historical and modern city. On one hand its monuments and artistic heritage testify its strong identity. On the other hand, the "eternal city" is open as well to the future and to innovations. Historical identity and modernization are moving forward hand in hand, and this ambivalence is now the true key to urban development. The urban agglomeration of Rome is therefore one of the fastest-growing regions in the world, and this growth has unprecedented effects on sprawl and population dynamics.

Our  research was conducted to examine past and current effects of the urbanization process, occurred over the large Roman urban system, based on old maps and multi-source and multi-temporal optical remote sensing (RS) data, collected between 1900 and 2014. These changes were then validated via Geographic Information System (GIS) techniques, in a particular procedure applied to urban land/agricultural transformations. The overlay of modern RS data over historic maps within GIS, allowed the understanding of the spatial relationships of past phenomena. The proposed approach, based on geo-statistical methods, was used to calculate the index of innovative space (AP Index), useful for the monitoring of the urban sprawl phenomenon. Strong evidence of urban expansion over the north-eastern quarter of the city, accompanied by environmental degradation and loss of biodiversity, is provided. Urban infill developments are expected to emerge in the south-eastern areas too, and these might increase urban pressure as well. In conclusion, RS and GIS technologies together with ancillary data can be used to assist decision makers in preparing future plans to find out appropriate solutions to urbanization encroachment.

The major task of this remote sensing study is to produce a land-use map for the catchment area of Tanoor and Rasoon spring and thereby assist in the delineation of groundwater protection zones taking into consideration that Tanoor and Rasoon springs are of the main water resources in the area.

Knowing the exact land-use and urban area density helps locating potential contamination sources that may pollute the spring system and in identifying areas which are vulnerable to groundwater contamination and therefore in planning for future human activities in the area and to formulate suitable restrictions on land-use. In that way a better protection of drinking water resources from contamination can be achieved.
 

High resolution satellite imagery was used due to the karstic nature of the working area and its sensitivity to pollution. Furthermore, the wide range of topographic and land-use changes required this step where 0.5 meter and 2 meters resolution Pleiades images were purchased.

The geometric correction of the data was performed through the collecting GCPs. At eight geographic locations coordinates were taken by using a differential GPS. After post processing by an external consultant an accuracy of 0.02 to 0.03 m was obtained. The shift between images coordinates and field coordinates ranged from 8 to 11 meter. Therefore, a geometric correction was performed using Pci-geomatica ortho engine.

A land-use map has been produced through supervised classification after that an accuracy assessment report was done based on the ground truthing points, 20 points were taken and checked in the field (by a handheld GPS with an accuracy of up to 3 meters). And another 118 points where taking using Google earth as a high resolution data source, however the report resulted in an overall accuracy of 89.13%.

 Supervised classification was used to extract buildings as potential contamination sources. A NDVI map was produced using ERDAS 2014 to determine the nature and density of vegetation in the area.

The analysis of the data was carried out by using the remote sensing software ERDAS2014 and Pci-geomatica2012. The results were exported as shapefiles. The commercial Geographic Information System (GIS) ArcGIS was used to perform post processing and produce final maps.

Mexico City is well known for its subsidence as a result of excess water withdrawal for many years. Soil deformations in the city areas are classified into two categories: sudden and slow subsidence. Sudden collapse rarely make geo hazard problems, however, slow subsidence usually make huge economic and human related disasters. Cities that were built on unconsolidated clays, silts, peats, or sands are in the danger of sudden and/or slow subsidence. Extreme ground water extractions, flooding, tsunami, earthquakes , etc are exposing the buildings in the danger of subsidence and/or related hazards. Most often the buildings and streets add weights to the region and intensify the soil’s stress even more. Meanwhile, monitoring of the buildings extensions and their relationship with ongoing subsidence is crucial.

The existing subsidence in Mexico City’s metropolitan area because of the over-pumping has been studied in this paper. Maximum of nine meters of subsidence in an area as big as  has been observed in the Mexico City’s metropolitan area. The main subsidence is happening because of the water extraction, and consequently compaction of the alluvial sediments. In the subside area, inter granular pressure of aquifers, decreases, and depletion of water at depth is the cause of the observed subsidence.

We studied this subsidence by means of Interferometric Synthetic Aperture Radar (InSAR), Continuous Global Positioning Systems (CGPS), and optical remote sensing data. Fifty two ENVISAT-ASAR, nine GPS stations, and one Landsat ETM+ image from Mexico City area have been analyzed to get better understand of the subsidence rates and its effects on Mexico City’s commune. InSAR methods like differential interferometry and Persistent Scatter Interferometry (PSI) carried out, to monitor the existing subsidence. Our InSAR data covers temporal baseline between 2002 until June 2010, and our GPS data cover temporal baseline from 1998 until 2012.

Radar permanent scatterers were extracted with use of three well known methodologies of solving the integer ambiguities of phases in the radar images: bootstrapping, ambiguity function, and integer least square. It worth noting that the newly operated satellites like Sentinel-1 and Radarsat-2 are showing similar patterns of subsidence, stressing the fact that the rate/pattern of progressive subsidence in the study area, is unchanged. For risk assessment, support vector machine (SVM) methodology based on Landsat ETM+ image is carried out to classify Mexico City’s populated density area, which further helped us to compare the subsidence rates (from PSI) with populated buildings. Maximum of 378.2 mm annually (from bootstrapping approach) change in Line Of Sight (LOS) direction in radar data, is in good agreement with the previous and newly measured studies. Considering this rate of subsidence, in ten years, it would reach in total of 3.8 meters of displacement, which mainly in the rainy sessions, floods would make real problems. This study shows that, the fastest subsidence in the over mentioned temporal baseline, occurs in the highly populated area, and ongoing subsidence are not slowing down.

Keywords: Mexico City, subsidence, Synthetic Aperture Radar (SAR), GPS, interferometry, Persistent Scatterer Interferometry (PSI).

Since decades, the land surface temperature (LST) is a parameter widely considered in the urban area mapping from space. LST has been often retrieved and mapped to evaluate the surface urban heat island (SUHI) using different spaceborne platforms, such as AATSR, ASTER, MODIS and Landsat. Several factors need to be assessed in the LST retrieval from satellite thermal infrared data: sensor radiometric calibration, atmospheric correction, surface emissivity estimation. Particularly, in urban area mapping issue, the satellite sensor spatial resolution may be a limiting factor in detailing the fine scale spatial variability, especially in the presence of impervious surfaces and sharp transitions (e.g., buildings, roads, parking lots, riverside, restricted vegetated zones), such as in a urban texture. The growing demand of remote sensing maps with finer and finer spatial resolution to successfully monitor the SUHI effects at district level and to avoid temperature underestimation stimulates the development of downscaling techniques when the actual sensor measurements do not meet the spatial detail requirements. In this work we perform the downscaling of coarse resolution LST maps from MODIS and Landsat to finer resolutions with the aim to increase the information content of the original maps, using summer satellite images over Milan, Rome and Florence. The downscaling is the enhancement of the spatial resolution of the original pixel data using ancillary information at higher spatial resolution. Different physical and statistical downscaling approaches have been proposed in literature: in this work, a statistical LST downscaling approach regression-based using different spectral indices over heterogeneous urban landscape is proposed, and the reliability assessed. This analysis allows to select the spectral indices and their combinations providing the best results in the LST image sharpening. First, the downscaling was performed using the Landsat TM data over Milan and Rome, assuming the 120 m spatial resolution image as reference. Then, the same downscaling regressive schemes were applied on the contemporary coarse resolution LST MODIS image and verified with the reference LST Landsat map. A further downscaling assessment at finer resolution was carried out using  the LST retrieved from Landsat TM over the city of Florence: in this case the sharpened image was compared with a high-resolution thermal image provided by an airborne survey carried out on July 18, 2010. Two Landsat scenes were processed before and after the flight, with the aim to evaluate the impact of the Landsat TM thermal channel resolution (120 m) on the LST estimation over a heterogeneous urban texture. Again, thermal data were downscaled at 30 m with a statistical algorithm using a regression on different spectral indices. The proposed approaches and comparisons allows us to assess potentials and limits of the LST downscaling performed over an heterogeneous urban area. 

After the launch of the Copernicus Sentinels 1, 2 and 3 by the European Space Agency, the availability of free and open Earth Observation (EO) data streams provides totally new opportunities for innovative scientific and commercial geo-information services. With the given spatial resolution and revisiting times, the potential of Sentinels missions to support a wide range of environmental, regional and urban planning and monitoring applications is high. Recently, the ERA.Net-RUS project GEOURBAN developed a set of EO-based environmental indicators for urban planning and a software tool for their on-line evaluation. GEOURBAN mainly focused on the local city level; however, planning in peri-urban and rural areas is particularly important for Russia, given its huge territory and its high number of large cities and scattered settlements.  Standard EO-based spatial datasets, such as the European Urban Atlas have proven to be quite valuable for various urban and spatial planning applications. However, this data just exists for the large and medium European cities, but not for the Russian ones. To this end, the main objective of the SEN4RUS (exploiting Sentinels for supporting urban planning applications at city and regional levels in Russia) project, that was recently funded by ERA.Net-RUS Plus, is to take into account the specific requirements of spatial and urban planning in Russia to develop indicators that effectively and efficiently exploit the information content provided by Sentinels mass data streams in support of city and regional planning. SEN4RUS is based on GEOURBAN outcomes, therefore using the expertise and basic techniques developed in the context of GEOURBAN, SEN4RUS will design and implement EO-based services for planners and decision makers that are specifically tailored to the Russian requirements. A key instrument in this context is the further development of a Web-based Information System (WIS) capable of evaluating the EO-derived indicators and provide them in a form that allows easy access and direct implementation into planning procedures. Three Russian cities with different typologies and planning perspectives have been included as case studies: St. Petersburg, Omsk and Vladivostok. To engage the users in the project, a Community of Practice approach will be employed. The innovation of SEN4RUS lies in the development of robust techniques for information extraction and derivation of geo-information products from Sentinel satellite imagery in combination with an improved  WIS that is adapted to and optimized for the  Russian urban and regional planning system and can be easily understood and controlled by non-experts. Adaptation of the SEN4RUS WIS to forthcoming missions have also been planned, therefore a fully operational tool is expected in the future.

The compared edge detection algorithms are Canny edge detector and SUSAN edge detector. The results are visualized and analyzed in order to extract significant information about edge detection algorithms efficiency when these methods are applied to map the land infrastructure. Positive and negative aspects of the methods are examined in terms to assess their results as advantages and disadvantages. The used image in the paper is over Venice, Italy. The algorithms are implemented by QGIS Desktop 2.4.0.

The compared segmentation algorithms are best merge segmentation by Beaulieu and Goldberg, best merge segmentation by Tilton, normalized cut segmentation by Shi and Malik, tree merge segmentation by Felzenswalb and Huttenlocher. The tree merge segmentation by Felzenswalb and Huttenlocher is chosen based on its optimized features when the method is used to map land infrastructure. The result image from the tree merge segmentation by Felzenswalb and Huttenlocher is classified by Mahalanobis classification and Bhattacharyya classification. Experimental results are shown to illustrate the advantages and disadvantages of the methods. Future plans are created. The used image in the paper is over Venice, Italy. The algorithms are implemented by QGIS Desktop 2.4.0.

The Land Surface Albedo (LSA) is a critical physical variable which influences the Earth’s climate by affecting the energy transfer and distribution in the Earth-atmosphere system. Its role is highly significant in both global and local scales, since LSA measurements provide a quantitative means for better constraining global and regional scale climate modelling efforts. Similarly, in urban environments LSA is crucial for the estimation of the local scale radiation and energy budget. In the present study, the LSA was estimated in large urban areas globally, at 0.5 km × 0.5 km spatial resolution and on an 8–day basis, for the period 2001–2014. Products from the Moderate Resolution Imaging Spectroradiometer (MODIS), on board NASA’s Terra and Aqua satellites were used, including the directional-hemispherical surface reflectance (black-sky albedo) and the bi-hemispherical surface reflectance (white-sky albedo), both available at 0.5 km × 0.5 km, and the MODIS-derived Aerosol Optical Thickness (AOT), at 1° × 1° spatial resolution. Since LSA also depends on Solar Zenith Angle (SZA), 8-day mean LSA values were computed as averages of corresponding LSA values for representative SZAs. The estimated LSA was analyzed in terms of both spatial and seasonal characteristics, while LSA changes during the period examined were assessed based on a linear regression approach. The effects of the Normalized Difference Vegetation Index (NDVI) and rainfall trends on LSA changes were also assessed. Urban areas were masked using the Global Urban Footprint (GUF) layer, i.e. the DLR global map of built-up areas derived by means of TerraSAR-X and TanDEM-X data acquired between 2011 and 2013. All computations were performed using the Google Earth Engine platform and the data available in its catalog, which include all the above mentioned satellite data. The Google Earth Engine is a cloud system designed to enable petabyte-scale scientific analysis and visualization of geospatial datasets. Google Earth Engine’s consolidated environment, including the abovementioned data, co-located with thousands of computers for analysis, made possible the global-scale urban LSA estimation for the 14-year period and the corresponding statistical analysis. The results revealed substantial spatiotemporal variability of LSA, highlighting the great potential of Earth Observation data in combination with the power of cloud computing in supporting relevant studies. 

Particulate Matter (PM) concentration is used as an air quality indicator in urban areas; it is highly important for urban planning and epidemiological studies. Numerous research works, analyzing the impacts of PM on human health, have found associations with increased morbidity and mortality. Monitoring of PM concentrations is primarily based on ground measurements. Despite the fact that dense station networks exist, in large cities like London, in situ measurements do not provide detailed information on the spatial distribution of PM at local scale. This reason has prompted an ongoing effort for PM estimation using satellite observations. The present study evaluates alternative spatio-temporal approaches for quantitative estimation of daily mean PM concentrations. Both fine (PM2.5) and coarse (PM10) concentrations were estimated over the area of London (UK) for the 2002-2012 time period, using Aerosol Optical Thickness (AOT) derived from MERIS (Medium Resolution Imaging Spectrometer) / AATSR (Advanced Along-Track Scanning Radiometer) synergistic analysis at 1 km x 1 km resolution. High-resolution (100 m) local urban surface cover and morphology datasets were incorporated in the analysis in order to capture the effects of local scale emissions and sequestration. Spatial (2D) and spatio-temporal (3D) kriging were applied to in situ urban PM measurements to investigate their association with satellite-derived AOT while accounting for differences in spatial support. Linear mixed-effects models with day-specific and site-specific random intercepts and slopes were estimated to associate satellite derived products with kriged PM concentration and their predictive performance was evaluated. The developed method will be adapted to Sentinel 3 series, the first of which (Sentinel 3A) is expected to be launched in late 2015: the synergistic use of the Sea and Land Surface Temperature Radiometer (SLSTR) and the Ocean and Land Color Instrument (OLCI), on board Sentinel-3, is expected to provide improved AOT and thereby to increase its potential to support local scale studies related to urban planning and public health. The statistical models produced in the present study will contribute to the development of an operational tool capable of producing high-resolution PM concentration maps using Sentinel-3 observations. Since all covariates used in the predictive models are satellite-derived products, the methodology can be transferred to other urban areas, to estimate both PM10 and PM2.5, depending on the availability of in situ measurements to calibrate satellite observations. While the MERIS and AATSR 11-year data set is suitable for investigating past PM changes and trends in urban areas, the forthcoming SLSTR and OLCI sensors offer the possibility of downscaling, if combined with the high spatial resolution MSI (Multispectral Instrument) onboard Sentinel-2 series. The common spectral channels of MSI, OLCI and SLSTR provide unique capabilities for synergistic use of observations from these sensors. Such synergistic use is expected to lead to daily PM concentration maps of high spatial resolution, which are necessary in urban air quality studies.

This paper explores the possibilities of synergy of data from Copernicus Program satellites on orbit, e.g. Sentinel-1A and Sentinel-2A, for land cover monitoring purposes.

After the launch of Sentinel-1A on 3rd April 2014 and the recent launch of Sentinel-2A on past 23^th June 2015, we are able to start combining both data for land cover monitoring.

We had used the ascending track 117 of Sentinel-1A products acquired in interferometric wide mode in Single Look Complex format to exploit the interferometric capabilities of this sensor, together with the available Sentinel-2A products, available on Sentinels Scientific Data Hub, over the Veneto region (NE Italy). It has been selected this area since Veneto region is sensitive to flooding and hence, most accurate land cover maps are needed for a better management and control of areas with flood risk and the potential damage that new floods could cause.

We have produced several land cover maps and we have compared the different possibilities of creation of these products, derived from SAR, Multi-spectral and the fusion of both data types.

These are the first land cover maps with the highest resolution ever generated over Veneto using only free data from European Satellites, such are Sentinel-1A and Sentinel-2A, which will benefit a better economical estimation for being used in risk management and planning. 

According to the United Nations Development Programme, the 21th century is the first “urban century” [1]. In 2014, 3.9 billion people, corresponding to the 54% of the global population, lived in cities [2], and, in 2050, this percentage will grow to 75% [3]. Furthermore, most of this growth concerns developing countries, which have limited capacities to deal with this rapid change.

The growth, often chaotic, of urban agglomerates has a not negligible impact on the environment. In fact, cities are the main responsible for some of the many global problems such as waste production and air and water pollution. Thus, the need for technologies that allow for monitoring and planning this expansion, for predicting and mitigating its effect on natural resources, as well as the exposure of populations to man-induced and natural risks, is growing rapidly.

Satellite imagery constitute a powerful tool for planning the most correct expansion of cities giving a synoptic view at regional scale. The recent launch of Sentinel-1 and Sentinel-2 satellites, under the aegis of the ESA Copernicus programme, represent an unprecedented occasion for the remote sensing community, which finally has the possibility to access data at free of charge both at multispectral and microwave frequencies.

In this paper, we present an innovative method for mapping urban areas from RGB multitemporal SAR derived composites [4], [5]. However, treating this topic, a question arise: how to define an urban area [6] ?Today, no generally accepted definition of “urban land” exists. In fact, urban areas could be defined by their administrative borders, but they often do not reflect the development of a town. Sometimes the boundaries lie beyond the cities built-up area, including rural countryside. Sometimes they lie within the built-up area. Other approaches limit the urban area to the “built-up” area or define it in terms of the areas for which services and facilities are provided. In any case, the definition of urban area involves some arbitrary decisions in finding boundaries [7]. In fact, towns tend to merge physically and functionally with neighbouring towns and their hinterlands. Therefore there is no hard border for urban areas and in any definition an urban area embraces land cover types not typically urban, such as forest, parks or agricultural land [7].

From the standpoint of remote sensing, the only feature that can be associated certainly to an urban area is the “built environment” which, at the boundary of a city, constitutes a continuum with the rural area. In order to extract a border between the urban environment and the rural one, a dichotomy between these two features has to be created, modeling appropriately the existing urban-rural gradient [6].

We propose to solve this problem by exploiting an object-based processing performed on classified products obtained by SOM clustering of the input RGB SAR composites [5]. The obtained results, either on stripmap resolution COSMO-SkyMed derived products or on scansar resolution Sentinel-1 derived products, testified the reliability of our approach.

References

[1] Q. Weng and D. A. Quattrochi, Eds., Urban Remote Sensing. Boca Raton, FL: CRC Press, 2006.

[2] UNWater, “The United Nations World Water Development Report 2015: Water for a Sustainable World,” UNESCO, Paris, Tech. Rep., 2015.

[3] UNDESA (United Nations Department of Economic and Social Affairs), “World Population Prospects: The 2014 Revision, Highlights,” United Nations (UN), New York, Tech. Rep., 2014.

[4] D. Amitrano, G. Di Martino, A. Iodice, D. Riccio, and G. Ruello, “A New Framework for SAR Multitemporal Data RGB Representation: Rationale and Products,” IEEE Trans. Geosci. Remote Sens., vol. 53, no. 1, pp. 117–133, 2015.

[5] D. Amitrano, F. Cecinati, G. Di Martino, A. Iodice, D. Riccio, and G. Ruello, “Sentinel-1 Multitemporal SAR Products,” in IEEE Int. Geosci. Remote Sens. Symp., 2015.

[6] J. R. Weeks, “Defining Urban Areas,” in Remote Sensing of Urban and Suburban Areas, T. Rashed and C. Jurgens, Eds. Berlin: Springer, 2010.

[7] C. Boehm and R. Schenkel, “Analysis of Spatial Patterns of Urban Areas Using High Resolution Polarimetric SAR,” in 1st EARSel Workshop of the SIG Urban Remote Sensing, 2006.

Assistance provided by international organisations to developping countries is often hindered by lack of accurate and up-to-date geoinformation. GIM recently got the opportunity to address this issue by applying its in-depth expertise in object-based image analysis (OBIA) to the development of advanced semi-automated workflows extracting relevant features from a massive amount of very high resolution EO data. Two success stories demonstrating the contribution of space data to a sustainable, resilient and inclusive development will be presented.

First case will show how GIM unlocked value from imagery to help stopping polio in Nigeria. In 2014, in the context of the Global Polio Eradication Initiative, GIM supported a large scale immunization effort by processing 100.000 km² of 50 cm resolution Pléiades imagery to map all kinds of human settlements (i.e. from single huts to large cities) with an unprecedented level of detail and detection rate hence allowing for locating and reaching every last child and ending polio. 12 TB of data were processed in 5 months time to extract half a million of buildings or small building groups, 20.000 villages and 1.500 cities that were then visited by the vaccination teams. Thanks to this unique urban mapping effort, the efficiency of the immunization campaigns could be greatly increased to the extent that no new polio case has been reported in this region since more than one year.

The second case story will illustrate how GIM, thanks to the support of ESA and the World Bank, could address the issue of lack of insights on urban sprawl and densification patterns and on urban poverty and its vulnerability to natural disasters in one of the fastest-growing agglomerations in Asia. Historical and current space data was first used to analyse the evolution of the urban extent and density in Metro Manila. The main achievement of this project was however to leverage the availability of very high resolution satellite imagery to map with an unprecedented level of detail all the informal settlements over entire Metro Manila area and parts of surrounding province (circa 750 km²). Moreover no minimum mapping unit was applied what allowed for retaining even the smallest slums. Often overlooked in the development of urban planning policies the latter represent however a very significant proportion of detected informal settlements. Finally, thanks to the power of OBIA methodology, GIM could reach the level of individual housing structures delineation within informal settlements. This in turn allowed for deriving a large set of relevant physical attributes such as average size of buildings, proportions of different building categories, density of buildings, presence of vegetation, etc. These attributes were then used in a model providing an objective classification of the slums into different types.
Thanks to these products derived from space data, it is now possible to understand better the development of the city over the last 25 years and to identify the key processes driving current expansion of the city. The delivered products provided a comprehensive and consistent overview on how the land has been consumed and allocated while highlighting specific hotspots and patterns thereby providing leads for further analysis in conjunction with information on e.g. prevailing urban planning policies. Moreover, the informal settlements mapping layer and its attributes provide a reliable input to finally answer the question of the magnitude of the urban poverty issue (e.g. how many slums there are, how are they distributed, are they preferentially linked to specific LULC classes, etc). Likewise the informal settlement typology provided a unique insight and reveals the magnitude of particular aspects of urban poverty such as the example of the so-called "pocket slums". These layers also open up many perspectives in terms of informing decisions regarding the implementation of specific measures and the assessment of their social impact. They can for example support the identification and prioritization of specific and spatially targeted actions such as protection against flood, areas for resettlement or informal settlements upgrading for example in the context of the implementation of the new Flood Management Plan.

The effect of the built geometry on surface temperature and effective emissivity in urban areas was studied. An effective emissivity was first modelled as a function of the geometry of the built-up space only. This is our urban emissivity model based on the sky view factor, indicated hereafter as UEM-SVF. Second, the effective emissivity was calculated using a micro-climate radiative model, i.e. the Temperatures of Urban Facets in 3-D model (Krayenhoff and Voogt, 2007), indicated hereafter as TUF-3D model. The two effective emissivity data sets were compared and evaluated, and the induced errors in surface temperature retrieval were also evaluated. Firstly, the effective emissivity derived from UEM-SVF was compared with the effective emissivity estimated from TUF-3D model. The effective emissivity from TUF-3D was calculated from the total upwelling longwave flux at the top of the urban canopy determined by the Stefan-Boltzmann flux at the complete surface temperature. Results showed that the effective emissivity derived from UEM-SVF was similar to that estimated using TUF-3D, under the assumption that all elements within a scene have uniform surface temperature and emissivity. The correlation coefficient (r²) was 0.99 and the root-mean-square errors (RMSEs) are 0.002, 0.004, 0.005, 0.004 respectively when the uniform material emissivity is assumed, ɛ = 0.8, 0.85, 0.90, 0.95 respectively. However, when the surface is heterogeneous, e.g. different materials and/or surface temperatures in a scene, the difference in two sets of effective emissivity was much larger. The degree of discrepancy is highly dependent on the degree of heterogeneity. When the surface temperature of all elements in a scene is assumed to be the same but the surface emissivity is different, e.g. the broad-band material emissivity of rooftops, roads and walls was different, e.g. 0.95 for roof, 0.90 for road and 0.85 for wall,    respectively, the RMSEs between the two effective emissivities was larger than 0.01. On the other hand, when the material emissivity of rooftops, walls and roads was taken equal to the spectral emissivity of these materials in the Landsat-8 thermal bands, which are rather similar, thus high correlations and small RMSEs were observed between the effective emissivity values obtained with the UEM-SVF and the TUF-3D model. The r² in both band 10 and band 11 was 0.99, and the RMSEs were 0.004 and 0.003 respectively. This result is mainly due to the material emissivity of rooftops, roads and walls being similar to each other, e.g. 0.95, 0.949 and 0.95 in band 10; 0.957, 0.958 and 0.957 in band 11.

 When component temperatures within the observed target are very different, the discrepancy in effective emissivity increases dramatically. From the high resolution airborne thermal image obtained in Hong Kong on August 6^th, 2013, the mean temperature of the rooftop is about 328.56 K, the mean temperature of road is 316.64 K, and the mean temperature of the wall is 304.94 K in a built-up area. If the effective emissivity is estimated using the total upwelling radiance at the top of the urban canopy calculated by TUF-3D model and the Stephan-Boltzmann emittance at the complete surface temperature, values larger than 1 were obtained in some cases. This occurs when the range of ratio of building height to street width was from 1 to 4. In these cases, the r² between the effective emissivity from TUF-3D and UEM-SVF was 0.98, but the RMSE was 0.064 and 0.065 in the Landsat bands 10 and 11. In winter, the difference of component temperataure is smaller than summer (from satellite thermal image, the rooftop temperature is about 297.31 k, and the road temperature is 294.77 k, the wall temperature was assumed as 292.5 K). The RMSE between the effective emissivity from TUF-3D and UEM-SVF was only 0.009 and 0.01 respectively. Results showed that the RMSE is highly related to the different component temperatures in a scene when the material emissivities are same or similar.

When the effective emissivity based on the TUF-3D model is calculated by the area weighted exitance of all surfaces and the Stephan-Boltzmann emittance at the complete surface temperature, it does not exceed 1. When using the UEM-SVF model, the effective emissivity will never be larger than 1 since it does not consider multiple scattering and reflection in heterogeneous mixed pixels with different component temperatures. The TUF-3D model revealed that the exitance of the canopy can be higher than the area weighted total emittance, because of the heterogeneity of component temperatures, material emissivities and geometry characteristics in the scene. This is because the surface with lower temperature can reflect and absorb the higher radiance from other surfaces with higher temperature. The radiance measured by a space-borne radiometer over an urban canopy increases with specific surface area at constant horizontal footprint because of multiple reflection and scattering and this may cause the measured radiance over urban canopy at constant horizontal footprint to be higher than the total area-weighted emittance.

Considering that image pixels in urban areas are heterogeneous and non-isothermal, the effective emissivity calculated using TUF-3D model and UEM-SVF model was evaluated. Errors induced on the urban surface temperature retrieval were also analyzed. A case study on urban areas in Kowloon Peninsula was carried out, and errors induced by effective emissivities and surface temperature retrieval caused by heterogeneous pixels was investigated.

The population of Africa is predicted to double over the next 40 years, driving exceptionally high urban expansion rates. Urbanization has profound social, environmental and epidemiological implications and makes spatial and quantitative estimations of urban change, population density and socio-economic characteristics valuable information for epidemiology and vulnerability assessment. Satellite remote sensing offers an effective solution for mapping settlements and monitoring urbanization at a range of spatial and temporal scales. However, their products are largely dependent on the quality and type of data available and on the detected landscape.

African cites unlike occidental urban planning present a high heterogeneity where the urban areas are not well defined and extensive growth is usual characterised by slums. Factors as heterogeneous building structures, strong cycle of vegetation density and man-made constructions from natural raw local materials are some of the difficulties that urban mapping using remote sensing techniques are challenged with.

Synthetic aperture radar (SAR) technology, being indifferent to weather and illumination conditions, ensures continuous urban monitoring and demonstrates its suitability for operational urban mapping using texture information extracted from SAR data. However, the relationship between the radar signal and the built environment is complex, varying with the configuration of the sensor, the nature of the target, the environmental conditions and the employed data processing techniques. Optical remote sensing on the other hand is dependent on the illumination and frequently limited because the cloud cover conditions however its strength lies in the better resolution and higher temporal availability. One way to overcome these shortcomings is by fusing optical and SAR data.

This research is part of a large project, MAUPP, which addresses the Modelling and forecasting of African Urban Population Patterns for vulnerability and health assessments. One of the objectives of the study is to produce an urban expansion model for African cities in three steps:

i) The development of automatic and effective method for the delineation of urban extent using optical and SAR data;

ii) The creation of temporal data base of urban extent and land cover for about 50 African cities in sub-Saharan areas;

iii) The generation of urban expansion models based on Boosted Regression Trees (BRT) method and using remote sensing, statistical and contextual data.

In this work we present the processing method for the delineation of the multi-temporal urban extent of African cities. As a case test, the capital of Burkina Faso Ouagadougou, is selected. This large city of 1.7 million habitants covers an area of 370 km square and contains heterogeneous built up areas and land-covers.

Temporal optical data (SPOT, Landsat-TM, Pleiades) and SAR data ( ERS-1/2, ENVISAT, SENTINEL-1) will be collected for the period 1994-2015. To ensure consistence in the SAR data observation conditions over time, signal polarization (VV), trajectory (descending orbit) and similar incident observation angle are required. Pre-processing procedures of SAR data include speckle reduction, co-registration and histogram adaptation. Optical data requires mainly georeferencing.

The image processing and fusion include the following procedures; Using the grey level co-occurrence matrix (GLCM) we obtain eight different texture images averaged over four directions. Additionally an adapted intersection histogram distance index (HDI) using two selected textures image was found to be a good representation of urban density. To assess the built-up density of an urban area, the probabilities functions related with the non, low-medium and high density built-up area are presented. Such probabilities functions are obtained using a multi-lineal logistic regression from the 8 GLCM textures and three VHR optical training datasets. As for SAR, the GLCM is applied to extract textural features using the optical data. For land cover features several indices are applied including indices for extracting built-up area.

The urban area extracted from the SAR and optical products are combined using Support Vector Machine (SVM). This method has the generalization ability and excellent learning performance to solve limited sample learning, nonlinear and high dimension problems.

After integrating the entire dataset and processing we mapped and classified the different urban areas according to their built-up density with a 90% confidence

The proposed processing scheme produces new valuable input that will allow the improving of available urban growth models for Africa.

Climate change poses serious challenges to urban areas and their growing population by increasing occurrence of heat waves, drought, heavy precipitation, cyclones and extreme high sea level events. These changes affect physical infrastructure, water supply, energy provision, transport and industrial production, hence resulting in a variety of ripple effects across different sectors of the city life. In this context, Earth observation (EO) has proven to be an effective tool for supporting decision makers in facing climate change; nevertheless, gaps still exist between the current state-of-the-art and the users’ requirements. The FP7 DECUMANUS (DEvelopment and Consolidation of geo-spatial sUstainability services for adaptation and environmental and cliMAte chaNge Urban impactS) project aims at bridging this gap. In particular, DECUMANUS has a principal objective to develop and consolidate a set of sustainable services that allows city managers to incorporate EO-based geo-spatial products and geo-information services in their climate and environmental change strategies to support the sustainable management of the cities in Europe. Furthermore, the project is user defined and driven; indeed, it has full engagement with the partner cities of Antwerp, Helsinki, London, Milan, and Madrid, which are fully integrated in the project workplan by defining requirements, testing products and services, validating results and acting as ambassadors of these technologies for other cities.

The four categories of DECUMANUS service products consist of: i) an urban climate atlas using proven GMES products, ii) land monitoring services providing land consumption information and urban ecosystems assessment tools, iii) city energy efficiency to assess energy consumption and improve energy efficiency in cities, and iv) citizen health tools to alert the population to health risks arising from poor air quality and excessive temperatures in the urban area. For each category, two different types of services have been implemented, namely basic and premium. Basic services include products derived from freely-available EO data (suitable for district-level analyses at larger scale and lower spatial resolution), whereas premium services include products, indicators and models developed and specified on the basis of an active engagement with the planning communities and/or the use of in situ information (suitable for local-level analyses at fine scale and very high spatial resolution).

In the context of DECUMANUS, the German Aerospace Center (DLR) is responsible of the development of the basic and premium land-monitoring services identified by the project user community as most relevant and useful for supporting their climate-change adaptation and mitigation plans. On the one hand, implemented basic services include: i) unsupervised spatiotemporal urbanization mapping by means of ERS-1/2 SAR PRI and ASAR IMP imagery; ii) automatic imperviousness estimation by means of multitemporal Landsat data, and iii)settlement patterns analysis by means of spatial networks. On the other hand, the portfolio of premium services comprises: i) mapping of current and potential green roofs (which are becoming more and more important due to their capability of absorbing rainwater, providing thermal insulation and reducing air pollution); ii) automatic tree detection (i.e., a task which currently is generally carried out by photointerpretation or in situ surveys and requires plenty of time); and iii) automatic bare-soil/sand versus asphalt/concrete discrimination (whose categorization is of great importance since it would allow to identify between areas that generally appear similar in optical imagery but are actually associated with different water permeability). All premium-service products are derived by using VHR satellite/airbone VIS+IR optical imagery together with LIDAR/DSM height information.

Ongoing validation activities in cooperation with the project partner cities assess the accuracy of the corresponding EO-based products and confirm their great potential for supporting climate change mitigation strategies both at district and local level.

Imperviousness High Resolution Layer (IHRL) 2006 and 2009, i.e. a 100 m resolution raster with 0‑100 % pixel values representing share of artificial impervious (built-up) surfaces, was produced using an automatic algorithm based on calibrated NDVI for the area of most of Europe within the frame of GMES precursor activities (for 2006 ±1 year) and Geoland2 (for 2009 ±1 year) and it is distributed by EEA in the framework of the Copernicus land monitoring service.

Some previous studies (e.g. Hurbanek et al. 2010) of small model areas have shown that the IHRL overestimates the share of impervious surfaces in areas with relatively compact (urban) settlement pattern and underestimates it in areas with relatively dispersed (rural) settlement pattern. Countrywide (Ibid.) or European (Maucha et. al 2010) studies of this phenomenon are rather rare. While they usually conclude with a relatively reliable estimate of commission error (using a stratified random sample), they fall short of a reliable estimate of omission error. Estimating omission error with acceptable confidence for a small class requires large sample (Maucha 2011).

The objective of this contribution is to assess the thematic accuracy of IHRL in Slovakia in both 2006 and 2009 using a large random sample of 20 000 pixels. A reference value for each IHRL pixel in the sample is derived by counting the artificial impervious points from a set of total 100 points regularly spread in a 10 × 10 m square lattice overlaid on top of aerial orthophoto. Using the IHRL and reference values in the sample, estimate of commission and omission errors with a relatively acceptable confidence is produced for each of the three different successively produced versions of IHRL 2006, for IHRL 2009, and also for the change layer IHRL 2006‑2009.

Acknowledgement

This research was supported by the Slovak Scientific Grant Agency VEGA project no. 1/0275/13 “Production, verification and application of population and settlement spatial models based on European land monitoring services”.

References

Hurbánek, Pavol – Atkinson, Peter M. – Pazúr, Robert – Rosina, Konštantín – Chockalingam, Jeganathan (2010): Accuracy of built-up area mapping in Europe at varying scales and thresholds. In: Tate, Nicholas J. - Fisher, Peter F. (eds.): Accuracy 2010: Proceedings of the Ninth International Symposium on Spatial Accuracy Assessment in Natural Resources and Environmental Sciences, 20-23 July 2010, University of Leicester, International Spatial Accuracy Research Association (ISARA), pp. 385-388, http://www.spatial-accuracy.org/system/files/img-X07133558_0.pdf

Maucha, Gergely – Büttner, György – Kosztra, Barbara (2010): European Validation of GMES FTS Soil Sealing Enhancement Data. ETC LUSI and EEA, 35 p., http://www.eea.europa.eu/data-and-maps/data/eea-fast-track-service-precursor-on-land-monitoring-degree-of-soil-sealing/eea-ftsp-degree-of-soil-sealing-1/soilsealing_european_validation_finaldraf2t.pdf/download

Maucha, Gergely (2011): Validation of GMES HR layers with respect to change detection - considerations and proposed methodology. Joint meeting Geoland2 & EAGLE group & Ms.Monina, 09 - 11 March 2011, 17 p., http://sia.eionet.europa.eu/EAGLE/EAGLE_5rdMeeting_g2_MONINA_FFM/EL_Validation_of_HR_layers_finaldraft.pdf

Satellite Earth Observation systems grant acquisition  repeatability, a feature which can be very important in the context of a typical application, i.e. tracking the evolution of urban areas. With this goal in mind, a new, open-source technique has been developed in a framework of vulnerability estimation, monitoring and forecasting. The technique, developed in the context of some EU FP7 projects [1-3], relies on object-based change detection and post-adjustment of results based on common-sense rules.

The developed technique does indeed take advantage of the wide time span typically considered when urban monitoring is performed using Earth Observation data, especially when prediction of future expansion is the intended use of the extracted maps. Long timespans are also involved when dealing with "Big Heritage Data" [4] to reconstruct historical development in the addressed area.

Urban area extraction is carried out relying on a previously developed technique [5] based on moderate-resolution, multi-spectral data. The series of extractions are then provided as input to a second processing stage [6] where continuity in time and reasonable assumptions are used to fix the possible extraction errors from the first phase. Three different filters are included, intended to correct apparently wrong extractions, following common assumptions to favour, and some time enforce, a "regular" behaviour over time.

This paper will illustrate the technique and some test results that have been carried out in specific areas to tune the subsequent development work.

[1] Framework to integrate Space-based and in-situ
sENSing for dynamic vUlnerability and recovery
Monitoring (EU FP7 SENSUM Project), Retrieved on Aug.
2015 from: http://www.sensum-project.eu/

[2] Rapid Analysis and Spatialisation of Risk (EU FP 7
RASOR Project). Retrieved on Aug. 2015
from:http://www.rasorproject.eu

[3] Marmara Supersite (EU FP7 MARSITE Project).
Retrieved on Aug. 2015 from http://marsite.eu/

[4] D. De Vecchi, M. Harb, F. Dell'Acqua: "Refining
registration of large, multi-temporal stacks of medium-
resolution images: a novel, automated approach for 'Big
Heritage Data'". Proc. of BiDS 2014, October 2014,
Frascati, Italy.

[5] M. Harb, D. De Vecchi, F. Dell’Acqua: "Automatic,
hybrid-based, built-up area extraction from LANDSAT-5,
-7 and -8 datasets". Proc. of JURSE 2015, March 2015,
Lausanne, Switzerland.

[6] D. De Vecchi, D. A. Galeazzo, M. Harb, F. Dell’Acqua:
"Unsupervised Change Detection for urban expansion
monitoring: an object-based approach". Proc. of IGARSS
2015, July 2015, Milan, Italy

Abha and Khamis-Mushyet, the twin cities of Aseer province, Saudi Arabia is a rapidly urbanizing agglomeration. The cities thus experience various environmental challenges that are true for all the global urban centers. Urban thermal environment is one such issue that the both cities is facing. Being a semi-arid mountainous city, Abha and Khamis-Mushyet’s urban thermal environment behaves differentially in comparison to topical as well as temperate climate cold country cities. This study analyzes the seasonal variation in spatial patterns of urban thermal environment and also the effects of terrain on the surface temperature. Temporal variation in distribution and magnitude of urban surface temperature was also studied and analyzed with respect to land use land cover practices using Landsat 8 data of all the four seasons (winter, spring, summer, autumn) for year 2014–15. An anomaly based approach was attempted to quantify seasonal and annual urban thermal environment intensities. The results reveal that the seasonal spatial distribution of surface temperature was affected by land use/land cover (LULC) and topography. Maximum and minimum seasonal urban thermal intensities were observed in summer and winter. Average annual land surface temperature anomaly map aided in identifying urban thermal intensities vulnerable locations in the both cities. The high dense built-up and major commercial/industrial areas display higher surface temperature in comparison with surrounding lands and were found to be the most vulnerable locations. There is gradual decrease of land use/land cover (LULC) classes’ surface temperature with the increase in altitude.  Therefore, the seasonal spatial variation in surface temperature also reflected the effects of topography on LULC classes. These findings can help urban planners to understand the urban thermal environment and design the green cool parks to counteract thermal intensity effects.

Fires and floods are among the natural hazards with the higher social impacts in the 21^st century, with economic cost of the order of billions of euros. When these occur in urban and periurban areas, the loss of human lives, the destruction of private and public properties, the degradation of health and quality of life, as well as the disruption of economic activities are among the impacts that cause. Floods that occur after the manifestation of fires, are extremely catastrophic, especially in peri-urban areas. The study of both hazards is based on the same background data and Earth Observation (EO) is a crucial information source, as from the same satellite imagery up-to-date fuel map can be derived in case of fire modeling in urban and peri-urban areas, while the parameterization of flood modeling in different scales (hydrological modeling in catchment basin level and hydraulic model in the urban area) need dedicated land cover/use information, updatable when needed and suitable for the specifications of the models. The investigation of both fire and flood hazards traditionally has been conducted separately even if the same data are needed. This approach overlook the “collect once – use for many purposes” model which when is adopted, result in the increase of the accuracy and economies, as these phenomena are tightly interrelated; fires exacerbate the flood risk and the preceding flood dramatically reduce the fire risk. In the framework of the LIFE+ project FLIRE an integrated Decision Support System (DSS) was developed for both floods and fires risk assessment and management by adopting the model of “once collect – use for many”, by using existence infrastructure and by incorporating extensively EO sources in different phases of the project. The FLIRE DSS is consists of three modules and seven applications unified under the FLIRE Server. The modules for fire management, flood management and weather forecasting have been implemented as web-services. The system has been designed as web-based solution which integrates the abovementioned tools. FLIRE adopt the distributed architecture of the components of the system while the DSS is accessible from the web (www.flire-dss.eu). The FLIRE server uses FTP and HTTP communication protocols and web service technologies. Visual basic, JavaScript, Google Maps API and Ajax have been used for the design and implementation of the FLIRE DSS. The user’s interface has been designed and developed based on the user's requirements, goals and needs. ESA’s Sentinels missions (1 & 2) is expected to have a crucial role for these steps due to the high temporal and spatial resolution, the high data quality, as well as the free data access policy.

‘Urban villages’ is the Chinese version of informal settlement. It is a unique phenomenon that comprises mainly low-rise and congested, often illegal buildings surrounded by new constructions and high-rise buildings. Due to a lack of an unambiguous definition allowing for a spatial delineation of such areas this article investigates a joint use of high-resolution optical and SAR satellite data through building extraction and 3D reconstruction of urban villages in Shenzhen, China. First, potential urban village footprints are extracted through a combined image fusion analysis of multispectral GaoFen-1 (GF-1) and high resolution TerraSAR-X radar (SAR) imagery. Then, building height estimation is performed on the basis of interferometry principles using interferometric X-band SAR (InSAR) from the Tandem-X mission. We generated Digital Surface Models (DSMs) from InSAR processing of two co-registered TanDEM-X image pairs for Shenzhen. While earlier attempts focused on data fusion techniques without the use of a DSM and yielded mediocre results, the integration of height information clearly improves the detection and mapping of urban villages. It can be demonstrated that urban villages and surrounding urban areas are clearly distinguishable through particular combinations of optical data, SAR data and height information. In particular, a rigid analysis identified three types of information as most suitable: 1) Normalized Difference Vegetation Index (NDVI), 2) contextual parameters such as edge and line density from GF-1 multi-spectral imagery, and 3) textural parameters such as Grey-Level Co-occurrence Matrix (GLCM) variables from TerraSAR-X imagery. The additional height information from InSAR clearly improves the detecting of taller buildings surrounding the urban villages. Likewise, the InSAR height information improves the delineating the low-rise and congested urban villages. In conclusion, the fusion of SAR and optical imagery can effectively reveal the footprint characteristics of urban villages. It is an effective means to reduce the effects of layover, shadow and dominant scattering at building location. The 3D building reconstruction model based on urban village footprint maps can reduce the continuous alteration of layover and shadow areas from high-rise buildings in the dense urban area. The proposed methodology based on the optimization of criteria referring to GF-1, TerraSAR-X and TanDEM-X characteristic efficiently integrates urban village footprint extraction and height calculations. Ultimately, the transferability and repeatability of this workflow is analyzed with the aim to establish a standardized monitoring process for urban villages.

When an earthquake occurs a rapid and accurate damage assessment of the hit urban area is essential. In this contest, valuable information can be obtained from Earth Observation (EO) data at Very High Resolution (VHR) which provide detailed information on single structures present in the scene, allowing to produce damage maps at single building level.

In this work we propose an object-oriented image analysis approach to detect damaged buildings from a pair of VHR optical images acquired before and after a seism.

We have tested our procedure using images taken by the QuickBird satellite before and after the earthquake that hit L’Aquila city (Italy) on April 6, 2009. Two layers of polygons reporting the damage ground survey of the whole urban area of L’Aquila, coming from two different sources, are used for validation purposes. The first one is related to the survey performed by the National Institute of Geophysics and Volcanology (INGV) Macroiseismic team, while the second one refers to ground survey carried out by the Italian Civil Protection Department (DPC). The geolocated version of the data set  has been provided by the Construction Technologies Institute of the Italian National Research Council (ITC-CNR).

As usual in any object oriented image analysis approach, the proposed methodology comprises two steps: the image segmentation, for obtaining a buildings map, and the object classification, for detecting damaged buildings.

In this work the segmentation of the pre-event optical image in objects corresponding to the buildings in the scene is performed using a pre-existing building map provided as GIS layer, but it can also be obtained by means of suitable segmentation/classification algorithms applied to a pre-event VHR optical image.

Once the objects are identified, a feature extraction step is carried out in order to build the classifier input space. Several features can be computed within the objects in order to identify those changed due to the earthquake. Here we consider: a) change metrics derived from the Information theory, such as Kullback-Leibler divergence and the Mutual Information; b) changes in the textural parameters derived from the grey level co-occurrence matrix; c) changes in the colour space, i.e. differences in the Hue, Saturation and Intensity parameters. Instead of using a moving window with fix size, all the aforementioned change features are evaluated at object level, by considering only the pixels within the building footprints.

The discrimination between collapsed or heavy damaged buildings and less damaged or undamaged buildings  is performed in the Bayesian framework. A non parametric approach is used as the statistics of the features cannot be easily predicted and a leave-one-out validation approach has been carried out to test the classification accuracy against ground truth. 

Although the classification performances using EO data are not excellent, similar uncertainties were observed between the ground datasets. Compared to the highly costly and time consuming ground surveys, the EO still appear a valid tool for a rapid response after an earthquake.

The work has  been funded by the EC-FP7 APhoRISM project (Research, Technological Development and Demonstration Activities, grant agreement n. 606738).

The Illegal dumps have a heavy impact on the landscape and the urban environment and are a source of danger for the health and safety of the inhabitants. Land monitoring and control bring to the recognition of abuses: the use of satellite remotely sensed data helps to improve the environmental management system. The availability of object-oriented techniques to process satellite imagery allows to detect such situations of illegality. 

The traditional image processing and image interpretation methods are usually based only on the information extracted from features intrinsic of single pixel: the object's physical properties, which are determined by the real world and the imaging situation-basically sensor and illumination. A limitation of this method is that it allows evaluating only a part of the information content of the images, without exploring the appearance as important as geometric-textural information. The application of Object Based Image Analysis on very high resolution data allows, with an automatic or semi-automatic process – with a minimal manual participation-a good classification also in presence of high and very high resolution data, where higher is the chance of error. The final classification, through a suitable hierarchy of classes that takes into account the relationships between the produced segmentation levels, may be highly accurate. Thus we introduce other rules for the location of the context, and the relation between the objects increases meaningfully the chance of automatic recognition of objects on the land surface. Object-based techniques allow an elaboration of satellite data to detect an uncontrolled storage of waste. By including a shapefile containing some detected test areas in the segmentation process, we proceed to the recognition of the landfill areas. In this work are also used some cadastral data for the multiresolution segmentation and the object-based classification. This research is still in progress in order to refine the techniques used but already an application of the same methods in other contexts is possible.

Water is essential to human life and all activities of man depend on a constant supply. In African countries human health is impacted daily by lack of access to water. During 2013, 43.6% of cases of worldwide cholera outbreaks were reported from Africa countries with no proper access to adequate water and sanitation resources. Similarly, the WHO estimates that 88% of waterborne diarrheal disease burden is attributable to unsafe water supply, sanitation and hygiene. N’Djamena City, the capital of Republic of Chad has a well laid out city but experiences regular outbreaks of cholera and various forms of waterborne diarrheal disease. Providing water and sanitation to N’Djamena and similar ever expanding urban and semi urban settlements has traditionally being a challenge as the planning process for water supply and sanitation is faced with the absence of updated, reliable maps. The application of new tools and data that can simplify and support water supply and sanitation planning is an important requirement.

The Water Observation Information System (WOIS) Urban Sanitation Planning and Support planning tool developed under ESA Tiger Net project has a model and a mapping component. The tool was used in mapping, estimating, current and future water demands and for the planning for the development of the water supply and distribution system in N’Djamena as a demonstration case. The mapping component is an urban land cover map derived from very high resolution satellite imagery, and the model includes a spatial link table so that urban land cover classes can be attributed with the water demand values available through the model. The model is populated with numbers based upon a combination of population and housing census data, city administration reports, literature and assumptions. The model include settings for analyzing quantitative changes in future urban water demand e.g. in response to population growth and technology scenarios i.e. household improvements and upward movement in sanitation ladder.

One of the main reasons because urban climate became an important issue consists in the rapid growth of the urban population worldwide. Thus, the urban population of the planet has grown from 746 million in 1950 to 3.9 billion in 2014. Based on the United Nations report “World Urbanization Prospects, the 2014 revision” (http://esa.un.org/unpd/wup/), 54 per cent of the world’s population lives in urban settlements, while by 2050, this is expected to increase to 66 per cent. Based on both the current trends of the urbanization process and the overall growth of the world’s population, by 2050 the urban population of the world will increase with another 2.5 billion people.

Urban climate studies use several types of modern technologies and methodologies. One of the most important approaches is based on the processing of satellite remotely sensed data. There are mainly three types of applications, which are using remote sensing for urban climate studies purposes: urban growth, land use / land cover, and surface urban heat island. All three of them are used within the “Urban Climate Study of Bucharest, Romania” project (http://urbanclimate.utcb.ro). This project is jointly developed by the Institute of Meteorology, Climatology and Remote Sensing from the University of Basel, Switzerland and the Urban Engineering and Regional Development Department from the Technical University of Civil Engineering Bucharest, Romania. Since land-climate relations are crucial to evaluate the state of the urban climate, whether it is urban growth, land use / land cover, or the surface urban heat island, at the basis of all these three types of applications is the land use/land cover classification and further interpretation of the remotely sensed images. Urban growth can be determined and characterised based upon the changes in the land use/land cover classes produced in a city over the time and the urban heat island can be estimated based upon the correlations between satellite derived surface temperature values and the incidence of the land-use types, or the remotely sensed derived indices of differentiation.

The paper presents the use of remote sensing technology applied to multi-temporal satellite imagery for studying urban growth, land use/land cover, and surface heat island in Bucharest, within the “Urban Climate Study of Bucharest, Romania” project.

Satellite data and further geo-information data are used for landscape ecological evaluations, e.g. to predict structural diversity in landscape, to derive quantitative data on open space fragmentation and on interlink of biotope structures. Satellite images are just as much used to identify compensational areas for planning of building land in conurbations or to quantify landscape metrics by means of derived medium and high resolution satellite parameters in order to calculate neighbourhood relations of objects.  Within the last two decades landscape structure indices or metrics have been implemented on remote sensing image data for different mapping scales.

Nature, in particular in the suburban cultural landscape is described regarding indicators such as structure (line or planar expansion, cutting, island areas, etc.), dynamics (entry of the modification processes) and texture (neighbourhood relations to other land use forms). This is based on the identification and computation of static and dynamic indicators that help providing a synthetic assessment of suburban landscapes. The indicators will also allow the comparison of the environment’s condition in different conurbations. A methodological approach is presented applied to different parts of Europe in growing as well as shrinking urban regions, after which monitoring and evaluation of a landscape diversity in suburban landscapes are feasible on the basis of medium and high resolution satellite data.

The present research focus on one specific dimension of environmental quality: land use and land cover, which experience changes over time within urban areas. Land change refers to adaptation in land cover or land use and is measured by comparing these two mentioned ones at two or more points in time.

Obviously, the term land change could refer to developments or several driving forces, which can cause change. One of these is the population dimension, such as the age, structure, mortality, morbidity and migration, which also affect consumption of vegetation and construction of buildings and infrastructure. Population events that increase consumption of vegetation and construction of buildings will change land use, a specific example being the fertility and birth rate. Along the well-known demographic events, a high impact on land cover changes has the human activities, which have modified land in various ways and intensities. Towns may suffer urbanization, suburbanization, residential areas can be converted into commercial areas, the neighborhoods in the city centers can go into slums, and all this driving to land degradation as a result of an extreme form of land-cover change based on involuntary development of Earth resources.

The objective of this research stands for linking and validating existing changes in land cover/land use extracted from Earth Observation satellite data with statistical socio-demographic data, using Pearson correlation coefficient. Acquired results and statements could aid to the development of a regional urban policy, as a key step for maintaining urban space and territoriality updated with a valuable significance in providing the needed direction and course of action to support sustainable urban development.

In the light of the constant change that urban areas undergo, it is of vital importance to be able to map the current state of the urban land cover in timely and accurate manner to support sustainable urban development.  The objective of this research is to evaluate multi-temporal Sentinel-1A SAR data for urban land cover mapping.  This study is part of the EO4Urban project funded by the ESA DUE INNOVATOR III program.

In the previous research of KTH Geoinformatics, we have examined various SAR data for urban land cover mapping including ENVISAT ASAR, RADARSAT SAR and TerraSAR-X data. Multitemporal Sentinel-1A data in both ascending and descending orbits acquired over Beijing and Stockholm during the 2015 vegetation season are selected for this mapping task. These two cities are significantly different in their structure and urbanization rate as well as the surrounding environments, thus provide excellent test scenarios with plenty of reference data.  The major classes are covering high-density built-up areas, low-density built-up areasroads, airport, urban green space, roads, airport, agricultural fields, bare soil, forest and water.

The methodology includes image preprocessing, segmentation, classification and accuracy assessment.  The core of the analysis is performed with our in-house developed software KTH-SEG, an object-based image analysis tool based on an edge-aware region growing and merging segmentator as well as a support vector machine for post-segmentation classification. In this research, multi-resolution segmentation are being performed with different object scales as some classes might be better assessed with a smaller scale while others require a larger scale in order to be properly mapped.

Preliminary results show that multitemporal Sentinel-1A data have the potential to produce detailed urban land cover maps, The classification accuracy will be assessed both in terms of classic measures such as the kappa value and overall accuracies and more recent methods like the location agreement and quantity agreement. The accuracy may vary depending on the quantity and variability in the Sentinel-1A images used for the classification.

A long exposure to asbestos is connected to the emergence of harmful diseases that affect the respiratory system. In particular, the friable asbestos is the most dangerous element, because its fibres easily disperse and have a particular aptitude to penetrate into the respiratory system. Despite the risks mentioned above, the number of roofs of buildings containing asbestos is still high in many countries. In Italy there are still many buildings characterized by roofs of a mixture of cement and asbestos, once common for the features of the material, as characteristics of mechanical resistance and its low cost.

An environmental monitoring for detecting the presence of this material in urban areas is generally made by means of expensive dedicated air travels, since typically the airborne payload is a hyperspectral sensor.

Unlike the air missions, satellites continuously orbit around the Earth. In particular, some commercial optical satellites in sun-synchronous orbit that monitor at very high geometric resolution are particularly suitable for this type of search.

This paper proposes the use of optical satellite data that, through the analysis object-based, have proved to be effective for detecting asbestos roofs.

Legislation requires intervention for the replacement of these roofs, and among the possible replacements is the smart substitution of asbestos roofs with PV roofs for producing energy.

What is proposed could so become a common and much less expensive practice than the current, which involves the use of airborne sensors.

The object based techniques for classifying satellite data allow to identifying buildings with roofs made by materials containing asbestos. In particular, in this paper from the start satellite imagery is integrated with cadastral data in vector format (shapefile) to maintain through the various procedures of segmentation and classification, the cadastral information relating to property. The object-based method consists in a Nearest Neighbor classification tool following a multi-resolution segmentation of the whole scene.

The high spatial and spectral diversity of man-made structures makes the mapping of urban surfaces with the use of Earth Observation data one the most challenging tasks of remote sensing field. Spectral unmixing techniques, although designed for and mainly used with hyperspectral data, can be proven useful for use with spectral data as well, to assess sub-pixel information. For urban areas, the large spectral variability of urban structures imposes the use of multiple endmember spectral mixture analysis techniques, which are very demanding in terms of computation time and impossible to implement in local computers.

In this study, an artificial neural network is used to inverse the pixel spectral mixture in medium resolution imagery (30 m). A series of Landsat images over Rome, Italy, were used to map the urban surface in a sub-pixel level. A spectral library was built, consisting of endmember spectra, collected from the images, and mixed synthetic spectra. Linear as well as non-linear mixture was assumed. The spectral library was then used to train a neural network and the resulting surface cover maps were compared against similar maps produced from higher resolution land cover information. The estimated versus the reference surface cover abundance images proved a good agreement.

Among the advantages of using a neural network is its ability to capture non-linearities in the spectral mixture and its low computational demand. Such techniques, affordable to implement, can be useful and support various application for urban studies and monitoring. The use of spectral unmixing techniques based on neural networks using medium resolution multispectral imagery, can be proven extremely useful for support of operational applications of the Copernicus Sentinels, because of their quick and accurate performance. With the recent launch of Sentinel-2, the method outlined in this study will be adjusted for and applied to Sentinel-2 imagery after the commissioning phase.

As described by [1], a “Satellite Image Time Series (SITS) is a set of satellite images taken from the same scene at different times. A SITS makes use of different satellite sources to obtain a larger data series with short time interval between two images… Sensors with high spatial and temporal resolutions make the observation of precise spatio-temporal structures in dynamic scenes more accessible. Temporal components integrated with spectral and spatial dimensions allow the identification of complex patterns concerning applications connected with environmental monitoring and analysis of land-cover dynamics.”

When we analyse the development of urban areas, it becomes clear that satellite image time series are highly valuable data sources that can be exploited to describe - besides vegetation cycles and land use changes - the dynamics of urban settlements and their infrastructure. Typical examples are given in [2] and [3].

For instance, one can focus on the extraction and analysis of long-term satellite image time series, and on applications in urban development monitoring [2]. Here, a Satellite Image Time Series comprised of more than 100 data sets, covering a time span of over 25 years is extracted from the Landsat data archives, in order to extract the annual built-up surface. The evolution of the built-up areas is then compared against population dynamics for the studied area.

Our presentation will focus on the currently attainable results from satellite image time series when being applied to urban scenes: Modern high resolution optical and SAR sensors with good signal-to-noise characteristics open new perspectives for local image classification and quantitative change analysis, while low resolution sensor data are often available over many years and provide more insight into long-term processes. Advanced analysis algorithms allow the identification of typical pixel changes and their confidence levels. Finally, data fusion represents a new perspective for urban mapping.

References:

[1] https://en.wikipedia.org/wiki/Satellite_Image_Time_Series

[2] T.  Costachioiu, R. Constantinescu, and M. Datcu, Multitemporal Satellite Image Time Series analysis of urban development in Bucharest and Ilfov areas, in Proc. 10^th International Conference on Communications (COMM), Bucharest, Romania, 2014.

[3] F. Petitjean, A. Puissant, P. Gançarski, Monitoring urban sprawl from Satellite Image Time Series, in Proc. IGARSS, Munich, Germany, 2012.

Conventional satellite ground segments collect, process, store, administer, and distribute satellite data without caring for the semantic content of the datasets. In the era of Big Data this concept precludes the efficient exploitation of satellite data as large numbers of irrelevant datasets have to be transferred from their archives to locally separate data analysis sites. On the other hand, if a satellite ground segment would already provide semantically annotated datasets, the number of dataset transfers could be limited to the actually relevant data products and users could concentrate on their proper applications.

In order to remedy this situation, the ESA-funded Earth Observation image Librarian (EOLib) project serves to setup the next-generation of Image Information Mining (IIM) systems, implementing novel techniques for image content exploration. Image Information Mining is a new field of study and methodology for automating the mining (extraction) of information from EO archives that can lead to content discovery and the creation of actionable intelligence (exploitation). IIM is more than just an extension of data mining principles and computer vision to satellite images. IIM for EO data also includes the joint mining of accompanying data describing the images, such as headers, metadata, etc. Therefore, IIM is an interdisciplinary approach to automating remote sensing analysis that draws on signal/image analysis, pattern recognition, artificial intelligence, machine learning, information theory, databases, semantics, ontologies, and knowledge management.

EOLib will exploit information about Earth Observation (EO) product contents which is usually hidden in raster data, image time series and metadata, thus enabling content-based search in very large archives of high resolution EO data. EOLib will be interfaced to and operated in the DLR Multi-Mission Payload Ground Segment (PGS) of the DLR Remote Sensing Data Center, representing at the same time a general new concept for the operations of Ground Segment infrastructures.

This system is particularly suitable for (semi-)automated urban mapping as the EO product information content represents actionable information for local information mining, including the semantic annotation of image patches. The product information content is a result of feature extraction and auto-annotation and is archived together with an EO product as a distinct product component. Annotations are part of a controlled vocabulary and can be structured in a thesaurus or a graph of semantic concepts.

The product annotation is uploaded into a semantic catalogue that can be queried by the users. The catalogue service allows query by annotation and query by example, giving a typical positive or negative representation of the expected or unexpected result. The product information is also used for the local product content inspection.

References:

http://wiki.services.eoportal.org/tiki-index.php?page=EOLib

Urban sprawl is a Europe-wide serious problem, not only due to total area taken, but also due to its spatial distribution patterns (leading often to landscape fragmentation) and the composition of land taken (mostly agriculture and natural areas are converted to artificial areas). Finally, also the utilization of urbanized area (no. of inhabitants, no. of jobs) has to be taken into account to assess sustainability of urban development. Land is a finite resource and therefore urban planners need to select land to be used for further development more wisely applying the concept of land recycling. There is a need for a user-oriented service facilitating the identification of suitable sites for redevelopment based on objective criteria and providing flexible insight into these trade-offs on a strategic level for specific areas, still keeping in mind the general policy context. To achieve this goal, the European Urban Atlas alongside the High Resolution Layer on Imperviousness degree together with other related global data sets such as the Global Human Settlement layer and the Global Urban Footprint, represent unique data sets as a basis for developing EO based information services for providing detailed information to policy makers and practitioners on potential land to be (re)developed within existing urban areas thus contributing to reducing urban sprawl.

This abstract focuses on the recent results obtained from the exploitation of these data sets, and especially from EO image analysis, within the “URBan land recycling Information services for Sustainable cities” (URBIS) project funded by the European Union within the 2007-2013 Competitiveness and Innovation Framework Program. Within URBIS, a methodology for exploiting the aforementioned data sets to create and temporally update an inventory of potential development areas (PDAs) on the European territory is defined and experimentally validated with three pilot sites (Amiens, France; Osnabrueck, Germany; and Ostrava, Czech Republic).

As a major step in this methodology, pattern recognition, image analysis, and data fusion methods are being applied to SPOT-5 HRG imagery, which is included in the Urban Atlas, to characterize land cover at high spatial resolution (2.5 to 5 m), and are being integrated with the mapping results provided by the Urban Atlas at a coarser spatial scale. Methodological approaches include state-of-the-art texture and spatial feature extraction (e.g., histograms of oriented gradients, semivariogram), pixelwise (support vector machine, random forest), Markovian, and region-based classification, multitemporal statistical modeling, and domain adaptation techniques. Thematic products of interest to the mapping of vegetated and non-vegetated gaps in urban areas and of imperviousness at 5-m resolution are also derived.

Preliminary experimental results with data from the three aforementioned pilot sites were characterized by quite high accuracy in the discrimination of buildings, roads, vegetated areas, bare ground, and water bodies, regardless of the small number of spectral channels of SPOT-5 HRG. The impact of acquisition time and seasonality on the accuracy was also investigated and found relevant especially with regard to the discrimination of vegetated covers. The experimental results also confirmed the potential of the integration of advanced remote sensing image analysis techniques and of openly available thematic layers for the purpose of characterizing imperviousness, and more generally, thematic information of interest to urban monitoring and planning, at high spatial resolution. The potential of Sentinel-2 data will also be discussed in the framework of the considered application to urban areas and taking into account the open data availability foreseen for the Sentinel missions.

Today more than 50% of the world population lives in urban settlements that are continuously growing their infrastructures and build-up areas, often without appropriate planning, with significant changes in land cover/use and consequent changes in biophysical parameters related to their radiation budget, hydrological cycle and to microclimate. Densely urbanized areas, with a low percentage of green vegetation, are highly exposed to the rise in temperature extremal occurrences phenomena like Heat Waves (HW) which are increasing in terms of frequency and intensity, also in temperate regions, due to ongoing Climate Change (CC). Their negative effects may combine with those of the UHI (Urban Heat Island), a local phenomenon where air temperatures in the compact built up cores of towns increase more than those in the surrounding rural areas, with significant impact on the quality of urban environment, on citizens health and energy consumption, as it has occurred in the summer of 2003 on France and Italian central-northern areas.  In this context this work aims at designing and developing a methodology based on aero-spatial remote sensing (EO) at medium-high resolution and on most recent GIS techniques, for the extensive characterization of the urban fabric response to these climatic aspects related to the temperature (within the general framework of supporting local and national strategies and policies of mitigation and  adaptation to CC). Due to its extension and variety of built-up typologies, the municipality of Rome was selected as test area for the methodology development and validation. First of all, we started by operating through photointerpretation of cartography at detailed scale (CTR 1: 5000) on a reference area consisting of a transect of about 5x20 km, extending from the downtown to the suburbs (including all the built-up classes of interest). The reference built-up vulnerability classes found inside the transect were exploited as training areas to classify the entire Rome territory. To this end, the satellite EO multispectral data, provided by the new sensor Landsat 8 OLI (with perspective to be duly integrated with those acquired by the ESA Sentinel 2 system), were used within a "supervised" classification automatic procedure, based on data mining and “object-classification” techniques. The classification results were then exploited for developing a calibration method, based on a typical UHI temperature distribution derived from MODIS satellite sensor data (summer 2003), to obtain an analytical expression of the vulnerability model, firstly introduced on a semi-empirical basis.

The first European map of settlements (ESM), representing area covered by built-up structures at 100m of resolution (one pixel representing 1 hectare) was published recently produced and published using the Global Human Settlement Layer (GHSL) methodology developed by the Joint Research Centre of the European Commission.

Thanks to the high detail of ESM, architects and urban planners can compare and study the construction of any city across Europe. ESM has been used by policy makers to design and evaluate best practices on territorial cohesion.

While the strength of ESM lies on the geographic extent, resolution and quantitative information ESM reports on the manmade environment, hidden between the buildings there is another component: data about the un-built environment. Some of the data implicitly present in ESM like information about the presence or absence of green has heterogeneous influences and impacts on our cities, and is therefore important to the characterization of the built environment. This data has influence on recreation, ecology, public health, property values, or other socio-economic factors affecting human settlements. This data is important to people and their quality of life, to the environment and its sustainability, to the economy, and its growth potential.

Hidden in the built-up, between the buildings there is open/un-built space, from a planning point of view also known as public space. Although ESM does not distinguish between private and public open space, (understood as organized recreation and green areas), when auxiliary data is introduced, un-built space quantified by ESM may reveal opportunities for the valorization and consideration of open space for the general public even when this is private. In the case of open green space, a type of urban open space, for certain environmental considerations (like the heat island effect in cities, or the human footprint) it is not relevant if the open green space is privately or publicly owned. The presence or absence of green urban space is important for certain analyses. Currently, apart from the European Urban Atlas, which maps recreation areas and public gardens as open green space, there is scarcity of data for the mapping of urban green across Europe by comprehensive and objective data. This poster shows how ESM detects open space and how urban green is classified in an urban context.

The ESA Research and Service Support (RSS) service provides resources to support Earth Observation (EO) data exploitation. The purpose of the service is twofold: on one hand, to ease EO data processing by bringing user’s algorithms to data, and on the other hand, to foster the development of new applications and services aimed to deliver value-added products. RSS users are Principal Investigators (PI), scientists, and service providers. In this paper we introduce the RSS service model and present the resources made available to enable and support urban monitoring from space.

The environments made available by RSS can be divided in two main service classes: (i) Cloud Toolbox service and (ii) Grid Processing on Demand (G-POD) service.

The Cloud Toolbox service provides users with customized virtual machines with pre-installed software plus additional software/packages on request, and with flexible hardware resources (CPU, RAM, Hard Disk) responding to the actual user needs. The Cloud Toolbox is mainly intended for application development, software testing, post-processing, and for those activities requiring for a limited period of time processing resources which are not available on a common PC.

The G-POD service provides users with a high performance computing distributed system with (virtually) unlimited capacity. The environment uses the Grid computing paradigm to allow for automatic splitting and parallelisation of the processing jobs. The Grid Processing on Demand service is mainly intended for bulk processing, full mission reprocessing campaigns and all those activities taking full advantage from computing parallelisation.

Three examples of RSS support to urban monitoring from space research activities are presented in this paper: (i) pre-processing campaign of ASAR data for a DLR project focused on urban mapping on a global scale; (ii) provisioning of EO data and Cloud Toolboxes to support The Bartlett School of Architecture (UCL) in urban design projects; (iii) support to Delft University of Technology for a study on Cairo urban area expansion.

Moreover RSS makes available SAR interferometric algorithms (SBAS, DORIS/StaMPS) to support studies of subsidence in urban areas.

RSS delivers significant value to EO researchers in terms of time and resource savings, thus enabling enhanced scientific productivity.

In the recent years a couple of operational urban mapping activities for international financial institutions have been implemented by Gisat in South-East Asia and Pacific region. Mapping services have been delivered in frame of EOWORLD2 and its precursor EOWORLD, both joint initiatives of ESA and World Bank, and PUMA project, follow-up initiative of World Bank. Service cases have aimed to contribute to understanding the extensive urban growth in various metropolitan areas including e.g. Karachi, Mumbai, Dhaka, Colombo or Surabaya by assessment of urban land use development.

The service is based on two pillars. First, Earth observation data and techniques are utilized for extraction of both retrospective and up-to-date information on urban land use and subsequently, assessment and comparison of extracted information and their links to the standard statistical information are provided for users within dedicated web-based platform.

Land use status and changes are extracted by analysis of optical satellite imagery. Based on previous experience and throughout the course of the projects object-based image analysis techniques of detecting complex land use classes were developed and tuned up at multi-resolution data. Semi-automated workflows preceding manual enhancements support consistent operational implementation of the service for large urban areas. Depending on recency of requested retrospective land use high or very high resolution data are used as baseline for information extraction.

The platform for Urban Management and Analysis (PUMA) is web based geospatial software for exploring and analysing integrated spatial data. PUMA adapts open-source software and allows users with no prior GIS experience to access, explore, visualize, analyse and share local, regional and global urban spatial data from a variety of sources in an interactive and customizable way. It supports the objectives of Global Urban Growth Data initiative: it helps the World Bank and its clients to develop a shared understanding of the long-term spatial, economic and environmental implications of land use by assessment of harmonized, comparable urban reference datasets.

Services have been defined in a way to be operationally extendable in short or long term update time frame in order to serve as a base for further future monitoring. The service components have also been proven in frame of urban risk domain in Copernicus Emergency Management Service or ESA’s initiative EO for a Transforming Asia Pacific.

Continuous urbanization changes the face of our globe raising questions of sustainability, ecological functionality and living quality in peri-urban and urban regions. Remote sensing enables us to obtain timely and reliable information on the state of urban areas and their changing patterns. The overall objective of this study is to evaluate the contribution of fused ESA Sentinal-1A SAR and Sentinel-2A MSI data to the City Biodiversity Index (CBI), the only biodiversity index designed specifically for monitoring and evaluating biodiversity in cities thus enabling inter-urban comparisons. In this study, five of the 23 index components of the CBI are derived with the proposed method, namely proportion of natural areas in the city, connectivity measures, regulation of quantity of water, climate regulation in terms of carbon storage and cooling effect of vegetation and recreation and education (area of parks with natural areas) for the megacity of Beijing, P.R. China. For the connectivity measure index, landscape metrics are used based on the classification result. The combined use of Sentinel-1A SAR and Sentinel-2A MSI data is advantageous over single sensor classification approaches due to the complementary information each sensor provides. In this research, multitemporal Sentinel-1A SAR data over the vegetation season are selected. Sentinel-2A data will be selected when available. The SAR data is processed using the Range-Doppler Terrain correction to remove relief displacements and filtered to remove speckle by a Lee speckle filter. Thereupon, the filtered Sentinel-1A SAR and Sentinel-2 MSI images are co-registered. The image stack is then segmented and classified by an object-based SVM classifier in the KTH-SEG software package. The classification results are then analysed using landscape metrics to determine the degree of landscape connectivity. Consecutively, the five indices are given scores between 0 and 4 according to the CBI framework. The use of fused Sentinel-1A/2A data is expected to produce better urban land cover classification than SAR or optical data alone. The fusion of Sentinel SAR and MSI data has the potential to contribute to urban ecosystem studies in general and urban biodiversity research in particular through their availability and improved spatial and temporal resolutions.

 Keywords: Sentinel-1A/2A, Urban land cover, KTH-SEG, City Biodiversity Index (CBI), Landscape Metrics

 Highlights:

New application domain and methodology for fused Sentinel-1A and Sentinel-2A data

Demonstration of the contribution of Sentinel-1/2 data for the City Biodiversity Index

Evaluation of Beijing’s Biodiversity with five components of the CBI

Today, nearly half of the world's population lives in cities. In developing countries, people are deserting rural areas while population is rising rapidly. In less than 20 years from now, these two factors will combine to drive over two billion people into urban areas, which in some cases are already overcrowded.

This fast growing of some cities and relocations of commercial and residential areas have produced important changes in the industrial and urban sectors not always following sustainability criteria. As results most urban growth falls outside formal planning controls and many cities suffer poor urban services management, traffic, and congestion, loss of green areas, poor air quality, and noise.

The main advantages of satellite-based EO applications are to bring parties together, to spread and improve idea, to support the decision-making process, and the development and operation of smart services. This way, satellite-based EO products can be used (both quantitatively and qualitatively) to help Armenian local authorities in development assessing the growth of urban areas in order to manage their geoinformation needs.

The monitoring of the urban areas and their evolution in time is one of key application of Earth Observation. The aim of this paper is to present an overview of a methodology developed with the main goal to take advantages from the great quantity of data that are coming with free and open access. In particular, in order to derive a status map for urban areas, a combination of SAR and optical data is used, primarily using Sentinel-1 and Sentinel-2 data, but the methodology has been tested also using Landsat-8, RapidEye, SPOT and other data. Chosen a reference period in which it is possible to consider the urban areas substantially unmodified, all SAR and optical data available in that period in the area of interest are used. All the interferometric acquisitions of Sentinel-1 are combined one to each other in order to produce several coherence maps. To reduce the noise, naturally present in SAR data, all the coherence maps are averaged producing only one coherence map of the area in the reference period. The optical data are pre-processed calculating the Top of Atmosphere reflectances. Each Sentinel-2 acquisition is partitioned in three datasets according to the spatial resolution of its bands (three images). All data used, both SAR and optical, are then automatically coregistered. The following step is to individually classify these datasets by means of an object based supervised classification using the European Environmental Agency Imperviousness layer 2012 to train the machine learning algorithms. Finally, the resulting maps are then merged, pixel based, using bayesian rules in order to derive the final map.

The processing workflow described, is implemented in a web based and cloud based system, called DFC (Data Fusion Center), starting from the data search and ends up with final map through a step-by-step process, managed by an open and interoperable system. In fact, it has the capability to interact automatically with the Sentinel-1 and Sentinel-2 catalogs (and other open access catalogs) selecting the appropriate images for the area of interest. Most of the processing steps described above are automatic, but, at this moment, some human supervised steps are present in the production workflow. Both are managed by DFC running the appropriate component or alerting the operator which has in charge the particular processing step which has to be performed. The input files, all the intermediate files and the final maps are also managed automatically by the system in a secure way. Moreover, automatic reporting capabilities and easy visualization of input, intermediate and final data through a web-based interface, complete the characteristics of the system.

Irregular settlement of railways, either due to the loading of the trains or local ground deformation, impacts its structural stability and the safety of passengers on board. Conventional methods for structural monitoring of railway use in-situ measurements, from GPS, leveling or special survey trains. These methods are expensive and can only be applied on a limited scale, either in space or time. Moreover, they are usually only used at locations where structural deformation is suspected, requiring a-priori knowledge which may not be available everywhere. Using satellite InSAR, we are able to complement these conventional methods and monitor the kinematic behavior (deformation) of railways with millimetric precision, to detect irregular settlement.

Here we use a probabilistic method for InSAR time series post-processing for the automatic detection of anomalies (e.g. railway irregular settlements). It is based on statistical hypothesis testing and the B-method of testing. In this method, we first (1) build a library of canonical kinematic functions,  based on physically realistic behavior, such as linear, seasonal, temperature-related, step-wise discontinuities and exponential behavior. Then, (2) we find the best model per InSAR measurement point using multiple hypotheses testing. Particularly to detect irregular settlement of railways, the localized differential deformation between two nearby points (i.e., over ‘short arcs’) is more important for railway stability than the large deformation of certain point with respect to a far-away reference point. Therefore, we apply the testing methodology on short arcs. Finally, (3) we evaluate the quality of the estimated parameters, and classify the InSAR measurement points along the railway in terms of their temporal behavior. We conclude that irregular settlement of railways can be recognized. Since there are more than 100,000 InSAR measurement points for testing, we use the B-method of testing to increase  the computational efficiency and define the optimal testing settings such as the level of significance and the power of the test. Our method is applied to all railways in the Netherlands. The kinematic time series of InSAR measurement points are derived from 73 Radarsat-2 acquisitions between June 2010 and August  2015.

Many areas in the world are susceptible to sinkholes and the associated risk of a sudden collapse of the surface.  About 13% of the world’s land surface is covered by carbonate rocks which are sensitive to erosion by running water, leading to cavities and potentially sinkholes.  In urban areas, sinkhole risks are a direct threat to human lives.

Detecting sinkholes is notoriously difficult, as techniques such as ground-penetrating radar (GPR), electrical resistivity tomography (ERT), seismic methods, and microgravity, usually have a very localized range, and are difficult to deploy between buildings. Yet, recently it has been demonstrated that the detection of small depressions using radar interferometry can be indicative for imminent sinkhole collapse site identification. In an earlier study, we have demonstrated that a sinkhole occurring in the south of the Netherlands appeared to be observable as gradual  deformation years before the actual collapse.

 Building on this experience, we designed a warning system to detect locations where the spatio-temporal behavior of the surface, or objects on the surface, has the characteristic fingerprint of a subsurface cavity, or an imminent sinkhole. We apply a robust method of hypothesis testing based on time series of TerraSAR-X and Radarsat-2 SAR data. We report on the characteristics of the method, the ways to deal with false alarms, and the potential for operational deployment.

The objective of this study is to explore the pertinence of mapping urban greenness, and hence all low-lying and tall vegetation within urban and peri-urban areas using data acquired by the recently launched Pleiades 1A sensor. Of course green areas within out-lying or in-lying rural and/or agricultural areas are taken into account.

The target here is to develop services for land planners to provide mean with which to observe both in detail and extensively the green corridor or green infrastructure in their areas by they on public or private lands, to develop an overall view. The request that is crystallising is the provision of geo-information layer that can be generated at a reasonable cost, over large areas and still be used for 1:2 000 land planning processes. The land planning specifications stipulates the mapping of low-lying permanent vegetation surfaces including prairies and grass-strips, urban grasslands, scrublands, isolated large trees, and other landscape characteristics such as hedges, woodlands and forests. The mapping of these features require their distinction from croplands and permanent urban mineral surfaces and other more specific features such as paths. Furthermore, the work needs to take into context the roads and hydrology networks.

To meet this demand SERTIT is proposing a remote sensing and GIS solution using imagery acquired by the recent sensor Pleiades 1A at 50cm. Supported by the CNES, SERTIT received a good coverage of the Urban Community of Strasbourg (CUS) and its environs and embarked on an ambitious and time-constrained initial green infrastructure mapping of nearly 400km². Firstly, SERTIT insured a strict observance of geometrical specifications. Then, after radiometric calibration pixel based and object-orientated methods were prototyped incorporating image bands, neo-channels and textural indices to produce the vegetation layers. This poster will outline aspects of the chosen method and initial results showing the advantages of Pleiades in the precise mapping of vegetation in complex urban environments.  

Within the framework of the CNES ORFEO preparatory program and after withn eh Pleaides Users Commissioning phase (RTU Pleiades) , very high resolution optical data, initially Pleiades Like then Pleiades HR were acquired over two  African capitals, Dakar (Senegal), Ndjamena (Chad) in context of public safety (large gathering, protection of civilians in conflict) and / or public health (parasitic disease - malaria). The diachronic analysis of these data shows significant changes both in terms of local changes in land use, and urban sprawl.

From 2005 to 2012, Ndjamena, Chad's capital, has expanded its urban area by 169%. Considering urban densities level by level (dense, moderate, sparse), it appears a strong evolution of the dense urban class of nearly 146%. Urban class of moderate density shows stability in terms of extension, but its spatial distribution is not the same between the two periods.

During this same period, for Dakar Capital, for on which space is more stressed due to the location of the town on a peninsula, there is also a significant densification of existing urbanization. Class to moderate density grows and expands noticeably towards North and West, particularly in the districts of Pikine, Cambérène and Yoff. Between the two sites, observed changes / developments could be more problematic from a public health point of view. In Ndjamena, the expansion of the city is done on an unstructured way with the implantation and intensification of buildings within low lying areas, subject to seasonal fillings inducing in one hand an important flood risk but also, in another hand as being potential breeding sites of mosquitoes that could transmit malaria. In Dakar, the way of urbanization is not totally similar. Within low-lying areas (Pikine East Medina Gounass Thiaroye Guédiawaye) water retention basins were dug to drain the soil in these densely populated areas.

These works carried over two major capitals of West Africa illustrate well the powerful of  Pleiades HR data for urban mapping. The results also pint pointed the profound changes in terms of urbanization and expansion that known these cities in less than 10 years, emphasizing the need of regularly update the land cover maps for both questions monitoring territory that security policy and / or health.