Urban Environment Quality Indicators: Application to Solar Radiation and Morphological Analysis on Built Area

Urban Environment Quality Indicators: Application to Solar Radiation and Morphological Analysis on Built Area CLÁUDIO CARNEIRO 1, EUGENIO MORELLO 2, GILLES DESTHIEUX 3, FRANÇOIS GOLAY 1 1 Geographical Information Systems Laboratory (LASIG), Ecole Polytechnique Fédérale de Lausanne SWITZERLAND claudio.carneiro@epfl.ch; francois.golay@epfl.ch 2 Laboratorio di Simulazione Urbana, DIAP, Politecnico di Milano, Milano ITALY eugenio.morello@polimi.it 3 Haute Ecole du Paysage, d'ingénierie et d'architecture (HEPIA), University of Applied Sciences Western SWITZERLAND gilles.desthieux@leea.ch Abstract: The representation and visualization of more or less detailed 3-D urban scenes can be done using different techniques, from those more conventional (for example, photogrammetry) to the most recent ones (for example, laser scanning). However, the use and application of this kind of data for the study of urban environment quality (UEQ) remains unsettled; indeed, the correct definition of indicators related to this field is highly necessary for the analysis and planning of urban developments. Moreover, the state of the art of 3-D Geographical Information Systems (GIS) available in the market still lacks the possibility of applying complex spatial analysis operations to the urban scale. Hence, the applied methodology is directly related to the extraction of these indicators, which are calculated according to the available 2-D and 3-D geo-referenced urban data. The work here presented is part of a larger jointly project between many researchers, experts and end-users around Europe which integrates cross-disciplinary competences, like remote sensing, GIS, image processing, energy, environment, architecture and urban design. In order to extract and validate the use of the proposed indicators, some case studies have been applied to several pilot zones of the cities of Geneva, Lausanne and Lisbon and are here presented. Key-Words: LiDAR, digital surface models (DSM), 2.5-D urban surface models (2.5-DUSM), image processing, urban indicators, building morphology, solar radiation, visibility analysis 1 Introduction Several different types of analysis concerning urban morphology, such as solar accessibility, heat transfer and visibility analysis are introduced in this work with the objective of giving a wide spectrum of possible research that can be achieved using 2-D and 3-D geo-referenced urban data: LiDAR, 2-D vectorial building footprints, here called building outlines and, when available, the 2-D projection of roof lines existing in 3-D models of cities. Depending on the initial data source, different levels of accuracy can be reached. For example, LiDAR data gives a very detailed mapping of roof sections (orientation and slope) that would not be possible using building footprints from cadastral data. The proposed tools are innovative solutions at the urban scale, while existing tools on the market are generally implemented at the scale of architecture (the analysis of buildings as single objects) or at the land use scale (GIS plug-ins for the environmental analysis). Nevertheless, in order to investigate the urban form, an accurate reconstruction of the urban fabric reveals itself to be essential. For instance, the presented work cannot ignore a careful morphological analysis at its basis in order to avoid that results could be significantly affected; hence, the image processing techniques and model reconstruction. The ultimate scope of this research is to provide a set of tools finalized to the environmental assessment of cities, thus providing valuable feedback to urban designers and planners. Considering this target, it is very important to scrutinize the right modalities to present results, both in terms of quantifiable indicators and visual representations. On the one hand, quantitative indicators must be significant at the scale of the neighbourhood or at the scale of the city, and should be used for comparative studies among different design schemes; on the other hand, visualizations have to be immediately comprehensive also to the wider audience, thus trying to capture and translate the indicators on the maps in the most efficient way. ISSN: 1792-6130 141 ISBN: 978-960-474-246-2

2 Related work Since this work is very interdisciplinary, we had inevitably to refer to the state of the art of more research fields, such as model reconstruction, morphological properties of objects, solar radiation and visibility analysis, to cite just some of them. Previous literature on the interpolation of LiDAR point clouds is not rare. The advantages and disadvantages of numerous interpolation methods, such as triangle-based linear interpolation, nearest neighbour interpolation and kriging interpolation were presented by [1]. A method to interpolate and construct a more accurate 2.5-Digital Surface Model (2.5-DSM), here called 2.5-D Urban Surface Model (2.5-DUSM), that incorporates terrain and buildings heights, based on LiDAR and GIS buildings data, has already been proposed by [2]. The effects of LiDAR data density on the accuracy of digital elevation models and examination to what extent a set of LiDAR data can be reduced yet maintaining adequate accuracy for DEM generation were studied by [3]. The investigation of solar radiation in architecture is not new and there are currently several tools that calculate radiation performance of buildings very accurately. Most are based on Computer Aided- Design in the architectural domain and consist in simulating solar access, such as the RADIANCE lighting simulation model [4]. However, [5] stressed the need to couple such CAD tools with 3-D GIS so as to include data processing and spatial analysis and to provide automated analysis on urban areas. Initially, the pioneers in the use of image processing techniques in order to analyse environmental indicators of digital urban models were a group of researchers at the University of Cambridge [6]. A first method that allows for deriving morphological properties of city blocks using an urban landscape model, constructed from a large LiDAR dataset, was presented by [7]. Using different topographical data, [8] presented a new methodology describing the density of urban systems, letting to quantify the urban volume. Volumetric urban form indicators are also used for visibility analysis, such as the 3-D isovist, that is, the visible space from a vantage point. This indicator, first introduced as a 2-D parameter in architectural studies by [9] and derived from Gibson s perception theory ([10],[11]) have been used in the last decades for several purposes, such as analysing morphological properties of architectural and urban spaces ([12],[13],[14],[15]), providing a quantifiable basis for Lynch's urban analysis ([16],[17],[18]), explaining the psychological impact of spaces on people [19] and assessing urban environment quality (UEQ) indicators related to the openness of views [20]. 3 Data sources used Due to its highest level of accuracy, the use, when available, of the 2-D projection of roof lines existing in 3-D city models and detailed 2-D vectorial building outlines is crucial in order to classify LiDAR points contained within each building and to improve the final result of the different 2.5-DUSM interpolated and constructed. According to the algorithm initially presented by [21], raw LiDAR data points corresponding to terrain are classified. An example of raw LiDAR data of a pilot zone of the city of Geneva classified by type of use is presented in Figure 1. Figure 1. Example of raw LiDAR data classified by type of use for a neighbourhood (business area) of the city of Geneva, Switzerland. The use of a hybrid approach from raw LiDAR data contained within the 2-D projection of roof lines existing in 3-D city models and vectorial building outlines, allows to interpolate four independent 2.5- DUSM (which can also include, or not, vegetation, mainly trees higher than 3 meters), respectively: 1. normalized 2.5-DUSM of roof lines; 2. 2.5-DUSM of roof lines; 3. normalized 2.5-DUSM of building outlines; 4. 2.5-DUSM of building outlines. Figure 2 shows an example of the data used and general structure related to the construction of each of these four 2.5-DUSM (in this case simplified, because trees are not considered). For more details about the construction of these four types of 2.5- DUSM please refer to [22] and [23]. ISSN: 1792-6130 142 ISBN: 978-960-474-246-2

The 2.5-DUSM of building outlines of a neighbourhood (Alvalade) of the city of Lisbon presented in Figure 3 is an output example of the model construction procedure using LiDAR data and 2-D building outlines. Finally, it is important to note that each of these four 2.5-DUSM is independently used for the extraction of several UEQ indicators, as presented in section 4. - Urban morphological properties of the built environment, aiming at providing area coverage (land uses) and density indicators [23]; - The assessment of shadowing conditions, the solar potential of the urban fabric and the estimation of solar power generation [24],[25]; - The energy demand (heating and electrical lighting) in the city environment [26]; - Visibility analysis (visual openness, wayfinding, visual impact assessment) [27]. Table 1 shows the general picture of the so far available tools, the input data needed, the extracted UEQ indicators and the obtainable type of visualisation. Table 1. Example of available tools for the extraction of urban environmental quality (UEQ) indicators. Figure 2. Data used and general structure related to the construction of the following digital urban models: (1) - normalized 2.5-DUSM of roof lines; (2) - 2.5-DUSM of roof lines; (3) - normalized 2.5-DUSM of building outlines; (4) -2.5-DUSM of building outlines. In the next two sections of this paper, two examples related to urban morphology and solar analysis are briefly presented. Figure 3. The 2.5-DUSM of a neighbourhood (Alvalade) of the city of Lisbon, derived from LiDAR data. 4. Extracting urban environment quality (UEQ) indicators 4.1 Description of tools The list of indicators below can help in getting a general overview of the available tools, summarizing the topics covered so far with this technique: 4.2 Morphological properties of buildings Considering the information described in the previous sections, morphological properties of buildings can be calculated [23]. Using the normalized 2.5-DUSM of roof lines, areas of roofs can be calculated and by means of the normalized 2.5-DUSM of building outlines, areas of facades and volumes can also be calculated. Other minor indicators can be then derived: - General morphological indicators: the total built floor area considering all storey; the mean height of buildings on the site, total area of roofs (for solar collectors purpose for instance); ISSN: 1792-6130 143 ISBN: 978-960-474-246-2

- Derived indicators of density: urban density, as follows: the built volume on the considered urban area (m 3 /m 2 ); the ground occupation index, i.e. the covered area to the urban area ratio (m 2 /m 2 ); the floor area ratio (FAR) (m 2 /m 2 ). As a sample, the surface to volume (S/V) ratio is presented. This indicator is used in biology as well in architecture as one of the most relevant shape related property of objects. The more compact (low values of S/V) a shape is, the lower is its thermal loss to the environment. For instance, the sphere is the shape that optimizes the S/V ratio (refer to the igloo), since it contains the maximum volume within the minimum external surface. Hence, high values of S/V usually belong to short buildings, whereas low values of S/V are characteristic for tall buildings with bigger volumes and relatively small exterior surfaces. The two images below visualize the S/V ratio in a 2-D (Figure 4, image above) and 3-D (Figure 4, image below) representation. From the comparison between the 2 images, we can state that the 3-D representation reveals itself to be very important in cases where the perception of the volume component cannot be ignored. In other words, the 2-D representation does not allow the verification of the correctness of the map since we do not have a visual confirmation of the actual volumes of the represented objects. 4.3 Solar radiation on building facades and building roofs These indicators address analysis on solar radiation incident on building roofs and building facades of the urban built environment. According to [28], the most accurate surfaces are created using a grid with a sampling size that relates as close as possible to the LiDAR point density during the acquisition phase. Therefore, for the different pilot zones here presented, a sampling size of 0.5 by 0.5 meters was used for a density of LiDAR points around 3 to 4 points per square meter (for example, in the pilot zones of the city of Geneva) and a sampling size of 1 by 1 meter was used for a density of LiDAR points around 1 point per square meter (for example, in the pilot zones of the cities of Lausanne and Lisbon). The technique used for the calculation of analysis outputs related to solar radiation is based on the image processing of the 2.5-DUSM here presented in section 3 and other input masks that are interpreted as raster images. These images result from the transformation of all the information attributes needed into masks: 2.5-DUSM of building outlines and roof lines (altitude values), slopes and orientation of roofs (which can be also obtained from a segmentation process described in [24]), roof prints and building and facade labels. Figure 4. Two maps illustrating the surface to volume ratios visualized in a 2-D (image above) and a 3-D (image below) representation for a neighbourhood (Chavannes) of the city of Lausanne. Given that roof outlines do not necessarily correspond to building outlines due to the existence of front-roofs in some cases, the 2.5-DUSM of building outlines is only used for the calculation of radiation in building facades and alternatively, the 2.5-DUSM of roof lines is used for the calculation of radiation in building roofs. It is important to note that when the 2-D projection of roof lines existing in 3-D city models is not available the construction of the 2.5-DUSM of roof lines is not possible. In this case, the 2.5-DUSM of building outlines has to be used also for the calculation of radiation in building roofs. With regards to solar radiation analysis, solar geometry formulae allow the derivation of both the beam and the diffuse components of hourly radiations, based on irradiance statistical values for a specific location, for every orientation and inclination of surface starting from the previous mentioned inputs. The shadow casting routine first introduced by [6] is applied to the input images or masks at a ISSN: 1792-6130 144 ISBN: 978-960-474-246-2

specific day of the year and hour of the day and is used to detect which pixels on roofs and facades are in shadow (cast from buildings or trees in the surrounding environment) and which collect direct sunlight [27]. In the particular case of facades, the urban model is sliced at every storey, so as to consider which part of the facade is affected by overshadowing. On this basis, we can assign the global incident solar radiation calculated in W or J /m² for various times scales (hour, aggregation by day, month or year). Figure 5 shows an example of basic representation of solar radiation analysis, from 9:00 to 16:00 of the 15 th of December, at the pixel unit scale for a pilot zone of the city of Lisbon - in this figure, each row of pictures represent independent information, presented as follows: 1. Shadow casting; 2. Solar irradiation on roofs (W/m2); 3. Solar irradiation on the second storey of vertical facades of buildings (W/m2); 4. Percentage of solar beam irradiation on the second storey of vertical facades of buildings; 5. admittance, particularly during the sunny hours. - Performances of the facades on a 2-D map: Figure 7 shows the mean irradiance (W/m2) falling on the second storey of vertical surfaces. Such a map can be displayed for every storey, which can be useful for focused analysis, but does not really give an overall view. Alternatively, the average value of radiation on the whole facade can be displayed by only one single map. - 3-D visualization: Figure 8 shows the average values per facade. But this representation needs several maps considering various view origins in order to visualize the values of all the facades of a building. Therefore the 2-D form with one single map is more synthetic and preferable (similar to the map shown in Figure 7), which was justified by an analysis undertaken with several end-users of the city of Geneva [29]. The advantage of 3-D visualization would be to refine the map by showing the results by storey and thus the vertical variation of radiation with the elevation. Figure 5. Hourly maps of shadow casting routing, irradiances (W/m 2 ) collected on roofs and facades of buildings, for a neighbourhood (pilot zone) of the city of Lisbon. The pixel-based simulation presented was made for the 15 th of December, from 9 AM until 16 PM (hourly analysis). The indicators from solar radiation analysis on facades can be represented in various forms: - Graphical profiles within 3-D maps (hybrid approach for geo-visualization of urban indicators). As an example, figure 6 shows the percentage of vertical surfaces subjected to direct irradiation with increasing height. It highlights very well how much the top parts of buildings benefit from higher solar Figure 6. Hybrid approach of geo-visualization for solar irradiation on urban areas: (1) - 3D visualization of a quantitative indicator of the average irradiance values per building facades (expressed in W/m2) on the 10 th of December at 12 PM for a neighbourhood (business area) of the city of Geneva; (2) graphical profile with statistical data that show the percentage of vertical surfaces subjected to direct radiation with increasing height [%] for the same day, at 10AM, 12PM and 2PM. ISSN: 1792-6130 145 ISBN: 978-960-474-246-2

ADVANCES in VISUALIZATION, IMAGING and SIMULATION available solar admittance and roof area. When roof sections are available (for example, in Geneva), yearly irradiation values (KWh/m2) are aggregated on each roof section to highlight in a synthetic representation which are suitable surfaces for installing solar collectors [24]. The 3-D map shown in Figure 9 is adequate to make a quick selection of suitable roof sections in a given urban area. Figure 9. 3-D visualization of annual solar irradiation (KWh/m2) by roof sections for a neighbourhood (Moillesulaz) of the city of Geneva. Figure 7. The mean solar irradiance collected by the second storey of each facade on the 10th of December at 12 PM, considering both beam and diffuse contributions for a neighbourhood (business area) of the city of Geneva. In large roof sections, the irradiation may be very heterogeneous due the local variation of overshadowing throughout the sections. In consequence, for more detailed analysis, a mix between raster and vector representation would enable to select suitable parts of roof sections; it consists of identifying clusters of pixels that share homogenous irradiation values (through classification techniques), as shown in Figure 10. Figure 8. 3-D visualization of average irradiance values per building facades (expressed in W/m2) on the 10th of December at 12 PM for a neighbourhood (business area) of the city of Geneva. The potential user would be in principle rather interested in assessing which section of the roofs would be suitable, or not, for solar panels collectors (for PV or thermal purpose) depending on the ISSN: 1792-6130 Figure 10. 2-D visualization of annual solar irradiation (KWh/m2) by suitable parts of roof sections for a neighbourhood (Moillesulaz) of the city of Geneva. 146 ISBN: 978-960-474-246-2

5 Conclusions This paper introduces several tools that use different data sources in order to analyse UEQ indicators of the built fabric. As emerged, both the accuracy and reliability of some applications are significantly affected by the quality and availability of the source information. The presented UEQ indicators are the basis for the development of further indicators addressed to various urban applications. The morphological analysis could lead to interesting indicators that may be used by urban planners in order to predict the environmental behaviour of different urban textures. The solar admittance indicator on buildings is rather useful in terms of urban energy planning and environmental policies devising at the level of community: inventory of well irradiated buildings, calculation of thermal and electrical potential for sun collectors, ratio to energy needs and global statistics. Future work should provide a common container as to manage different types of input data and to facilitate the computation of further indicators as those mentioned above. Moreover, the improvement of the interfaces among the different software used to read, analyse and reconstruct the models will certainly be fundamental to the increase of the usability of the proposed tools. References: [1] Zinger S., Nikolova M., Roux M., Maître H., 3-D resampling for airborne laser data of urban areas. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 34(3B), 2002, pp. 55-61. [2] Osaragi, T., Otani, I. Effects of ground surface relief in 3-D spatial analysis on residential environment, The European Information Society: Lecture notes in Geoinformation and Cartography, Edited by Sara Irina Fabrikant and Monica Wachowicz, Published by Springer Berlin Heidelberg, 2007, pp. 171-186. [3] Liu X., Zhang Z., LiDAR data reduction for efficient and high quality DEM generation. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 37(B3b), 2008, pp. 173-178. [4] Compagnon R., Solar and daylight availability in the urban fabric. Energy and building, Vol. 36, 2004, pp. 321-328. [5] Batty M., Dodge M., Jiang B., Smith A., Geographical information systems and urban design. Stillwell, J., Geertman, S., Openshaw, S. (Eds.), Geographical Information and Planning, Springer, Berlin, 1999, pp. 43-65. [6] Ratti C., Richens P., Raster analysis of urban form. Environment and Planning B: Planning and Design, Vol. 31(2), 2004, pp. 297-309. [7] Yoshida H., Omae M., An approach for analysis of urban morphology: methods to derive morphological properties of city blocks by using an urban landscape model and their interpretations. Computers, Environment and Urban Systems, Vol. 29, 2005, pp. 223-247. [8] Koomen E., Bação F., Searching for the polycentric city: a spatio-temporal analysis of Dutch urban morphology. Proceedings of the 8th AGILE Conference on GIS, May 26th-28th, Lisbon, Portugal, 2005 (no pagination). [9] Benedikt M. L., To take hold of space: isovists and isovist fields. Environment and Planning B, Vol. 6, 1979, pp. 47-65. [10] Gibson J.J., The Perception of the Visual World, Houghton Mifflin Company, Boston, 1950. [11] Gibson J.J., The ecological approach to visual perception, Houghton Mifflin Company, Boston, 1979. [12] Batty M., Exploring isovist fields: space and shape in architectural and urban morphology. Environment and Planning B: Planning and Design, Vol. 28(1), 2001, pp. 123-150. [13] Turner A., Doxa M., O'Sullivan D., Penn A., From isovists to visibility graphs: a methodology for the analysis of architectural space. Environment and Planning B: Planning and Design, Vol. 28(1), 2001, pp. 103-121. [14] Teller J., A spherical metric for the fieldoriented analysis of complex urban open spaces. Environment and Planning B: Planning and Design, Vol. 30(3), 2003, pp. 339-356. [15] Turner A., Analysing the visual dynamics of spatial morphology. Environment and Planning B: Planning and Design. Vol. 30(5), 2003, pp. 657-676. [16] Lynch K., The image of the city. MIT Press, Cambridge MA, 1960. [17] Conroy Dalton R., Bafna S., The syntactical image of the city: a reciprocal definition of spatial syntaxes, 4 th International Space Syntax Symposium, London, 2003. [18] Morello E., Ratti C., A Digital Image of the City: 3-D isovists in Lynch s Urban Analysis. Environment and Planning B: Planning and Design, Vol. 36, 2009a, pp. 837-853. [19] Stamps A. E., Isovists, enclosure, and permeability theory. Environment and Planning B: Planning and Design, Vol. 32(5), 2005, pp. 735-762. ISSN: 1792-6130 147 ISBN: 978-960-474-246-2

[20] Fisher-Gewirtzman D., Wagner I. A., Spatial openness as a practical metric for evaluating builtup environments. Environment and Planning B: Planning and Design, Vol. 30(1), 2003, pp. 37-49. [21] Axelsson P., Processing of laser scanner data - algorithms and applications. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 54, 1999, pp. 138-147. [22] Carneiro C., Golay F., Silva V., Plazanet C., Park J.J., GIS and LiDAR data analysis for the integration of multidimensional indicators on urban morphogenesis multi-agent vector based geosimulation. Geocomputation and Urban Planning: Studies in Computational Intelligence, B. Murgante, G. Borruso and A. Lapucci (Eds.), Vol. 176, 2008b, pp. 187-208. [23] Carneiro C., Morello E., Voegtle T., Golay F., Digital urban morphometrics: automatic extraction and assessment of morphological properties of buildings. Transactions in GIS (accepted, but not published yet), 2010. [24] Carneiro C., Morello E., Desthieux G., Assessment of Solar Irradiance on the Urban Fabric for the Production of Renewable Energy using LiDAR Data and Image Processing Techniques. In: Sester M., Bernard L., Paelke V. (Eds.), Advances in GIS, Proceedings of the 12th AGILE Conference, Springer Berlin Heidelberg, 2009, pp. 83-112. [25] Carneiro C., Morello E., Ratti C., Golay F., Solar Radiation over the Urban Texture: LiDAR Data and Image Processing Techniques for Environmental Analysis at City Scale. Lectures notes in geoinformation and cartography: 3d Geo- Information Sciences, Part II, Zlatanova, S., Lee, J. (Eds.), Springer, Berlin, 2008c, pp. 319-340. [26] Ratti, C, Baker, N, and Steemers, K 2005, Energy consumption and urban texture, Energy and Buildings, 37, pp. 762 776. [27] Morello E., Ratti C., SunScapes: solar envelopes and the analysis of urban DEMs. Computers. Environment and Urban Systems, Vol. 33, Part 1, 2009b, pp. 26-34. [28] Behan A., On the matching accuracy of rasterised scanning laser altimeter data. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 33(B2), 2000, pp. 75-82. [29] Carneiro C., Communication and visualization of 3-D urban spatial data according to user requirements: case study of Geneva. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. 37, Part B2, 2008a, pp. 631-636. ISSN: 1792-6130 148 ISBN: 978-960-474-246-2