Automated registration of potential locations for solar energy production with Light Detection And Ranging (LiDAR) and small format photogrammetry

Elsevier Editorial System(tm) for Journal of Cleaner Production Manuscript Draft Manuscript Number: JCLEPRO-D-15-00934R1 Title: Automated registration of potential locations for solar energy production with LiDAR and small format photogrammetry Article Type: Original Research Paper Corresponding Author: Dr. Szilárd Szabó, Ph.D. Corresponding Author's Institution: University of Debrecen First Author: Szilárd Szabó, Ph.D. Order of Authors: Szilárd Szabó, Ph.D.; Péter Enyedi; Miklós Horváth; Zoltán Kovács; Péter Burai, Phd; Tamás Csoknyai, PhD; Gergely Szabó, Phd Abstract: Energy production and consumption is a key element in future development which is influenced both by the technical possibilities available and by decision makers. Sustainability issues are closely linked in with energy policy, given the desire to increase the proportion of renewable energy. According to the Horizon 2020 climate and energy package, EU member countries have to reduce the amount of greenhouse gases they emit by 20%, to increase the proportion of renewable energy to 20% and to improve energy efficiency by 20% by 2020. In this study we aim to assess the opportunities available to exploit solar radiation on roofs with LiDAR and photogrammetry techniques. The surveyed area was in Debrecen, the second largest city in Hungary. An aerial LIDAR survey was conducted with a density of 12 points/m2, over a 7×1.8 km wide band. We extracted the building and roof models of the buildings from the point cloud. Furthermore, we applied a low-cost drone (DJI Phantom with a GoPro camera) in a smaller area of the LIDAR survey and also created a 3D model: buildings and roof planes were identified with multiresolution segmentation of the digital surface models (DSM) and orthophoto coverages. Building heights and building geometry were also extracted and validated in field surveys. 50 buildings were chosen for the geodetic survey and the results of the accuracy assessment were extrapolated to other buildings; in addition to this, 100 building heights were measured. We focused primarily on the roofs, as these surfaces offer possible locations for thermal and photovoltaic equipment. We determined the slope and aspect of roof planes and calculated the incoming solar energy according to roof planes before comparing the results of the point cloud processing of LiDAR data and the segmentation of DSMs. Extracted roof geometries showed varying degrees of accuracy: the research proved that LiDAR-based roof-modelling is the best choice in residential areas, but the results of the drone survey did not differ significantly. Generally, both approaches can be applied, because the solar radiation values calculated were similar. The aerial techniques combined with the multiresolution processing demonstrated can provide a valuable tool to estimate potential solar energy. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 1 Automated registration of potential locations for solar energy production with LiDAR and small format photogrammetry Szilárd SZABÓ1, Péter ENYEDI2, Miklós HORVÁTH3, Zoltán KOVÁCS1, Péter BURAI2, Tamás CSOKNYAI3, Gergely SZABÓ1 1 Department of Physical Geography and Geoinformatics, University of Debrecen, Egyetem tér 1. 4032, Debrecen, Hungary 2 Research Institute of Remote Sensing and Rural Development, 3University of Debrecen, Károly Róbert College, Mátrai út 36. 3200, Gyöngyös, Hungary 3 Department of Building Service and Process Engineering, Budapest University of Technology and Economics Address for correspondence: Szilárd Szabó Department of Physical Geography and Geoinformatics, University of Debrecen, Egyetem tér. 1. 4032, Debrecen, Hungary, tel.:+36 52 512900/22326 (switchboard), fax: +36 52 512945, e-mail: [email protected] *Manuscript Click here to view linked References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 2 Abstract Energy production and consumption is a key element in future development which is influenced both by the technical possibilities available and by decision makers. Sustainability issues are closely linked in with energy policy, given the desire to increase the proportion of renewable energy. According to the Horizon 2020 climate and energy package, EU member countries have to reduce the amount of greenhouse gases they emit by 20%, to increase the proportion of renewable energy to 20% and to improve energy efficiency by 20% by 2020. In this study we aim to assess the opportunities available to exploit solar radiation on roofs with LiDAR and photogrammetry techniques. The surveyed area was in Debrecen, the second largest city in Hungary. An aerial LiDAR survey was conducted with a density of 12 points/m2, over a 7×1.8 km wide band. We extracted the building and roof models of the buildings from the point cloud. Furthermore, we applied a low-cost drone (DJI Phantom with a GoPro camera) in a smaller area of the LiDAR survey and also created a 3D model: buildings and roof planes were identified with multiresolution segmentation of the digital surface models (DSM) and orthophoto coverages. Building heights and building geometry were also extracted and validated in field surveys. 50 buildings were chosen for the geodetic survey and the results of the accuracy assessment were extrapolated to other buildings; in addition to this, 100 building heights were measured. We focused primarily on the roofs, as these surfaces offer possible locations for thermal and photovoltaic equipment. We determined the slope and aspect of roof planes and calculated the incoming solar energy according to roof planes before comparing the results of the point cloud processing of LiDAR data and the segmentation of DSMs. Extracted roof geometries showed varying degrees of accuracy: the research proved that LiDAR-based roof-modelling is the best choice in residential areas, but the results of the drone survey did not differ significantly. Generally, both approaches can be applied, because the solar radiation values calculated were similar. The aerial techniques combined with the multiresolution processing demonstrated can provide a valuable tool to estimate potential solar energy. Keywords: roof plane, solar irradiation, point cloud, multiresolution segmentation, drone 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 3 1. Introduction Renewable energy resources are becoming increasingly important in the structure of energy production. As non-renewable sources (such as petroleum or coal) are often considered polluters of the environment, greenhouse gas producers, or as posing a high risk (nuclear energy), it is crucial to find solutions to replace them with environment-friendly alternatives. At the same time, the EU introduced the Horizon 2020 Framework Program for Research and Innovation: the efficiency of energy should be increased by 20%, the proportion of renewable energy should be increased by 20%, and greenhouse gas emissions should be reduced by 20% (European Commission, 2014). Considering the private contribution by residents, an increase in the number of passive houses can represent a genuine milestone in efficiency (Kozma et al., 2013), while local energy production can decrease the GHGs and improve the proportion of renewable energy sources (Farkas, 2010; Lázár, 2011; Lewis, 2007). In this study, we focus on solar energy as a possible solution for private energy production. It is a solution which has both advantages and disadvantages. In the current economic environment, private properties are not supported to install photovoltaic (PV) solar systems in Hungary. Consequently, the high cost of installation is a serious disadvantage, but it is a solution which can offer complete or partial continuous energy for both institutions and households. Accordingly, remarkable efforts have been conducted to determine the solar potential of winemaking facilities (Smyth, 2012). Besides, there is no loss involved in the transportation of the energy. A limiting factor is that not all roofs are appropriate for installing solar panels, as this depends on the size, aspect and slope of the roof planes. Shadows generated by the roof elements, chimneys, antennas, or by the trees and pylons in the street can seriously reduce efficiency (Stevanovits, 2013). Roofs can be detected with remote sensing techniques (e.g. Nagyváradi et al., 2013); however, a simple identification is not sufficient to assess which roofs are suitable for the installation of PV panels, as methods must be employed that can reveal the roofs’ geometry. Photogrammetry and Light Detection And Ranging (LiDAR) are the two possible methods suitable for this task. While photogrammetry requires aerial photographs, and the outcome depends on the geometrical resolution and the quality of the images, LiDAR works with laser beams and the reflecting signs are recorded. Photogrammetry yields a digital surface model (DSM), while LiDAR, based on the emitted and backscattered signs with different returning times, provides a model both for the ground (digital terrain model, DTM) and the surface (digital surface model, DSM). In terms of roof detection, both techniques are suitable; we only need information about the surface of the objects (i.e. the roofs). The LiDAR technique was developed in the 1960s, but became popular only in the first decade of the 2000s. Recently, several studies have dealt with terrain and surface models derived from LiDAR point clouds. Highly detailed digital elevation models are the most popular application fields (e.g. Chassereau et al., 2011; Liu, 2008) in natural or urban environments (Ghuffar et al., 2013; Zlinszky et al., 2014) or to extract different elements of the surface, such as geomorphic forms (Dorninger et al., 2011), trees (Mücke et al., 2013), city buildings, or street furniture (Priestnall et al., 2000). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 4 Numerous publications have discussed the detection of buildings based on LiDAR. In research conducted by Yu et al., (2010) the accurate detection of city buildings was the aim, as in the case of Zhou and Neumann (2013). Filtering buildings was also the objective of the works of Mongus et al (2014) and Li et al., (2013). While Alexander et al., (2009) dealt with roof structure, Lukac et al., (2014) focused particularly on the potential solar radiation of built-up areas with LiDAR data. There are several research studies which have adopted a photogrammetric approach, too, ranging from the digital representation of the solar panels (Shortis et al., 2008) through solar potential estimation on a city-scale (Nex et al., 2013) to a complete survey of roof geometry (Lin and Zhang, 2014; Protic et al., 2012). LiDAR has a relevant advantage against photogrammetry as it provides data of the ground even it is covered by tree vegetation (Demir et al., 2008; Korpela et al., 2012). Both techniques have their advantages and limits. LiDAR can be considered more reliable than photogrammetry in terms of the way data is collected: a laser beam has a footprint (i.e. a 20-40 cm diameter circle) on the surface and its size is the function of the divergence angle and the above-target flight height (Bin et al., 2008). Thus, laser beams have multiple echoes and often can penetrate vegetation and roofs covered by tree canopy, so these can also be surveyed (Shan and Toth, 2008). However, due to the footprint, the horizontal accuracy is worse than the vertical (Csanyi and Toth, 2007). A major issue with 3D point clouds is how to handle the dataset, especially in the case of surveys providing a very high point density. Photogrammetry is biased by the vegetation as it can only produce surface models. Furthermore, the technique is sensitive to homogenous area sections, periodic objects and shadows, while LiDAR is independent of them (Paparoditis and Polidori, 2004). According to Baltsavias (1999) the two technologies can be used in a complementary way to exploit the advantages of both. Incoming solar irradiation can be computed with the involvement of slope, aspect, and shadows cast by topographic features (e.g. mounds) or other surface objects (e.g. buildings, trees, chimneys, pylons etc., Boehner and Antonic, 2009; Quazi et al., 2015). If all of these parameters are involved in a model, results can be regarded as reliable (Iqbal, 1983). Calculations can be conducted based on the appropriate equations, or software, such as ArcGIS, SAGA GIS and GRASS GIS, which provide solar radiation models (Wh.m-2.day-1, Hofierka and Šuri, 2002; Hofierka and Kañuk, 2009; Hengl et al., 2009). All models have errors due to the underlying concept or to a lack of appropriate data, but in most cases we do not require exact values, because a good approximation of the possible maximum summed by a given time interval is sufficient. Studies have usually been designed to determine the area of the roof planes and the incoming solar energy, but have not compared the different surveying methods. Our aim was to investigate and compare the surface models of a LiDAR survey and an aerial imaging carried out with a low cost drone system from the perspective of roof detection. We compared the resulting roof shapes and evaluated their suitability for solar panel installation for both models; furthermore, we also compared 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 5 the incoming solar irradiation of the models. We also compared the cost-benefit issues of the drone and LiDAR based techniques. 2. Materials and methods 2.1. Data collection A combined LiDAR and high resolution aerial imaging was carried out over a 7 km2 area in Debrecen (Eastern-Hungary). A Leica ALS70-HP and a Leica RDC 30 RGBN 60 MP were used in the survey (1000 m flight height, 780 m swath, sinusoid scan pattern, 20% overlap). Point density was 12 point/km2, which was in accordance with the suggestion made by Cekada et al., (2010). An accuracy assessment was carried out on the whole study area; however, we used only a smaller part in the analysis to investigate incoming solar irradiation, where the drone survey was possible (Fig. 1). The roofing material of the buildings in the study area was red tile, ensuring the creation of a uniform database independent of LiDAR intensity values. # Fig. 1. approximately here The drone survey was conducted at average altitude of 93 m with a DJI Phantom quadrocopter and a GoPro Hero 3 Black edition camera (focus length: 2.77 mm, lens size: 14 mm) combined with an NDVI stress camera (XNiteCanonELPH110NDVI, focus length: 4.30 mm, lens size: 14 mm; LDP LLC Ltd.). The pilot area was 12 ha, falling within the area of the LiDAR survey, in the university campus (University of Debrecen). 2.2. Point cloud processing The LiDAR point cloud was filtered by TerraSolid’s TerraScan module in the MicroStation environment (https://www.terrasolid.com/download/tscan.pdf) over the whole area. TIN interpolation with natural densification (Lin and Zhang, 2014) was carried out for the separation of ground points, then vertical outlying points were removed using filters. Following this, we filtered out the buildings with parameterized algorithms of TerraScan. Afterwards, we extracted the vector features of the buildings as an element of a semi-automated roof identification. We aimed to find the optimal parameters to extract the minimal roof-part size to obtain the most accurate and detailed roof models. Besides, a digital surface model (DSM) was generated from the point cloud with 20 cm cell size (20 cm is the largest reasonable resolution which can be obtained from the 12 points/m2 point cloud) in order to make a comparison (Fig. 2). 2.3. Photogrammetric analysis We applied Agisoft Photoscan Pro 1.1.0. (Agisoft LLC) for the photogrammetric evaluation of 188 images taken by the GoPro 3 camera. We used 16 GCP points (measured with a Stonex S9 RTK 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 6 system) as tie points and the highest precision option was applied to produce the model. Outlier points were filtered out with the aggressive depth filtering mode. The procedure resulted in a classic DSM with a density of 107 point/m2, and also an orthophotograph compiled from the aerial photos. The procedure yielded a true orthophoto (proposed by Amhar et al., 1998); accordingly, both spatial coverages were used in the analysis. Both the DSM and the ortophotograph had a resolution of 20 cm. 2.4. Analysis of digital surface models Image segmentation was carried out on the DSMs, aspect and slope coverages (both the latter were derived from LiDAR and aerial images) using eCognition Developer. Image segmentation is an object-oriented analysis technique that takes into account not only the pixel values but also the texture (Blaschke, 2010); thus, contrary to pixel-based classification, ―salt and pepper‖ type errors can be avoided (Weih and Riggan, 2010). DSM, aspect and slope coverages were segmented using multiresolution segmentation with four different scale parameters (L10, L50, L100, L200) and found that the procedure with two steps from the super-object to the sub-object using L200 and L100 values fulfilled the aims, i.e. separating the input raster coverages into the largest homogenous segments (Kumar et al., 2014; Shao et al., 2014). This procedure was repeated with the use of the orthophoto. An XNiteCanon camera was used to produce a pseudo-color orthophoto with blue-green-infra red bands, which was used to calculate normalized difference vegetation (NDVI, Rouse, 1973) values. NDVI ranges from -1 to 1 and values below zero indicate high reflectance which is characteristic of bare soil/rock or anthropogenic objects (e.g. buildings, roads etc.; Rouse et al., 1973). We applied a roof-mask compiled from NDVI values (<0) and building heights (>3 m). 2.5. Digital building models Finally, four digital representations were produced for the buildings. The representation that provided the roof plane geometry in the most realistic way was the one from the point cloud processed in a CAD environment (LPC), and a segmented digital surface model (SDSM) was produced with the segmentation method from the surface model of the point cloud. PDSM (segmented DSM) and OPDSM (common segmentation of the orthophoto and the DSM) were produced from the surface model of the photogrammetry approach. 2.6. Validation We obtained field measurements with a Stonex S9 RTK GPS pair for 50 buildings to check the contours of the buildings. Besides, 100 measurements were carried out for 100 buildings with a Leica Disto D5 to control the building heights. Root mean square error (RMSE) and quartiles were reported for the calculated differences between the LiDAR data and the measured data. Solar radiation was validated by the comparison of a building’s (student hostel) roof planes based on the blueprint and the LiDAR survey. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 7 2.7. Calculation of the incoming solar irradiation We filtered out those segments of the roof planes that were suited to the following conditions (modifying the approach of Kassner et al., 2008): slope: 20-60o; aspect: 90-270o; area: >2 m2; compactness: >0.3 (Fig. 2). #Fig. 2 approximately here In order to evaluate the results of the modeling procedure, a sample building was analyzed. The building analyzed was the student hostel building of the campus site (N: 47° 33' 18.9000''; E: 21° 37' 24.1932''). The results of the model were compared to a validated method based on a manual approach. The data regarding the building’s roof were acquired from two sources. On the one hand, they were generated automatically from the described procedure and, on the other hand, manually from a digital map. We obtained the following data: area, slope and azimuth. Roof areas have errors due to the surveying method used, i.e. the top view of the roofs results in a smaller area for roof planes. We corrected the roof areas with the cosine of the slope angles (Früh and Zakhor, 2003); the correction was made as in (Eq. 1). )cos( M h t A A   (Eq. 1) where Ah is the roof area measured from above (horizontal roof) [m2]; At is the calculated area of the tilted roof [m2]. The number of PV panels by roof planes was determined manually, and automatically with a Python plugin developed for ArcGIS. In both solutions a 0.5 m buffer was omitted from the calculation, the rest of the roof plane was covered with PV panels. In the first step all available roof areas were covered with solar panels, in the second case the north facing parts of the roof were left empty. The incoming solar irradiation was calculated for the geometrical data acquired. Solar yield calculations were performed with an anisotropic solar irradiation model (Reindl et al., 1990). The direct, diffuse and reflected radiation components were calculated according to (Eq. 2-4) and the global radiation was calculated as a sum of the components (Eq. 5).   bt RDGI  (Eq. 2)                           bi M Mit RAfADD 2 sin1 2 1 cos11 3   (Eq. 3) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 8   2 1 cos1  Mt AGR  (Eq. 4) tttt RDIG  (Eq. 5) where A is the albedo value [-]; Ai is the anisotropy index; D is the diffuse radiation on a horizontal plane [kW/m2]; Dt is the diffuse radiation on a tilted plane [kW/m2]; f is the modulating factor of cloudiness; G is the global radiation on a horizontal plane [kW/m2]; Gt is the global radiation on a tilted plane [kW/m2]; It is the beam radiation on a tilted plane [kW/m2]; Rb is the ratio of beam radiation on a tilted plane to the beam radiation on a horizontal plane; Rt is the reflected radiation on a tilted plane from the surroundings [kW/m2]; αM is the tilt angle of the tilted plane [°]. The meteorological conditions of the building site were described by the insolation time and we applied the Angström-Prescott method (Paulescu et al., 2013) to determine the global irradiation. Insolation data used in the calculations were measured between 1981 and 2000. Calculations were performed for three cases. Firstly, the incoming irradiation was calculated for the entire roof area of the building. In the second case we calculated the incoming solar irradiation for the solar panels which were allocated to the roof planes. In the third case the panels facing in a direction ranging from northeast to northwest were removed since these locations lead to an economically non-viable solution. In the validation process, the traditional (manual) approach based on the blueprints was regarded as providing the most realistic data. Following this we included all the registered roof planes in the analysis and calculated the incoming solar energy for each of them. We summarized the roof planes of the 13 buildings which can be found in the surveyed part of the campus area. 2.8. Statistical analysis We applied non-parametric tests due to the non-normal data distribution of the roof area and solar irradiation. Our null hypothesis (H0) was that solar irradiation derived from the two surface models had the same mean rank, and the alternative hypothesis (H1) was that the mean ranks of the irradiation values were different at the p<0.05 level. Accordingly, Friedman’s ANOVA and the Wilcoxon paired test were applied in the hypothesis testing phase, combined with a Bonferroni correction (Zar, 1999). In the validation process of the building geometry, we calculated Cohen’s Kappa (Kappa Index of Agreement, KIA; Cohen, 1960). KIA is a measure of association, requires nominal values, and indicates whether the agreement occurs by chance (ranges are between 0 and 1; 0 indicates agreement by chance, while 1 is total agreement; Rosenfeld and Fitzpatrick-Lins, 1986). We compared the building contours based on GPS measurements and those that were derived from the 3D point cloud. Heights and area surfaces of the sample building were compared with the Wilcoxon paired test, and RMSE values were calculated. We evaluated statistical tests with p values (p<0.05), and also with effect size as a standardized measure of the difference between groups (Cohen, 1992). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 15 Gatziolis, D., Andersen, H-E., 2008. A Guide to LIDAR Data Acquisition and Processing for the Forests of the Pacific Northwest, USDA Forest Service, Pacific Northwest Research Station, General Technical Report PNW-GTR-768, http://www.fs.fed.us/pnw/pubs/pnw_gtr768.pdf Ghuffar, S., Székely, B., Roncat, A., Pfeifer, N., 2013. Landslide displacement monitoring using 3D Range Flow on airborne and terrestrial LiDAR data. Remote Sens. 5, 2720–2745. Heiskanen, E., Nissila, H., Lovio, R., 2015. Demonstration buildings as protected spaces for clean energy solutions – the case of solar building integration in Finland. J. Clean. Prod. doi:10.1016/j.jclepro.2015.04.090 Hengl, T., Grohmann, C.H., Bivand, R.S., Conrad, O., Lobo, A., 2009. SAGA vs GRASS: A Comparative Analysis of the Two Open Source Desktop GIS for the Automated Analysis of Elevation Data, Proc. Geomorphometry 2009, Zurich, Switzerland, pp. 22–27. Hofierka, J., Kaňuk, J., 2009. Assessment of photovoltaic potential in urban areas using open–source solar radiation tools. Renew. Energ. 34, 2206–2214. Hofierka, J., Šuri, M., 2002. The solar radiation model for open source GIS: implementation and application, Proc. Open-source GIS-GRASS users conference, Trento, Italy, pp. 1–19. Iqbal, M., 1983. An Introduction to Solar radiation, Academic Press Canada, Ontario Izquierdo, S., Rodrigues, M., Fueyo, N., 2008. A method for estimating the geographical distribution of the available roof surface area for large-scale photovoltaic energy-potential evaluations. Sol. Energ. 82, 929–939. Jakubiecz, J.A., Reinhart, C.F., 2013. A method for predicting city-wide electricity gains from photovoltaic panels based on LiDAR and GIS data combined with hourly Daysim simulations. Sol. Energ. 93, 127–143. Jochem, A., Höfle, B., Rutzinger, M., Pfefier, N., 2009. Automatic Roof Plane Detection and Analysis in Airborne LiDAR Point Clouds for Solar Potential Assessment. Sensors 9, 5241–5262. Kampouraki, M., Wood, G.A., Brewer, T.R., 2008. Opportunities and limitations of object–based image analysis for detecting urban impervious and vegetated surfaces using true–colour aerial photography, in Blaschke, T., Lang, S., Hay, G.J. (Eds), Object-Based Image Analysis, Springer, Berlin–Heidelberg, pp. 555–569. Kassner, R., Koppe, W., Schüttenberg, T., Bareth, G., 2008. Analysis of the solar potential of roofs by using official LiDAR data. Int. Arch. Photogramm. Remote Sens. 37 Part B4, 399–404. Korpela, I., Hovi, A., Morsdorf, F., 2012. Understory trees in airborne LiDAR data — Selective mapping due to transmission losses and echo-triggering mechanisms. Remote Sens. Environ. 119, 92–104. Kozma, G., Molnár, E., Czimre, K., Pénzes, J., 2013. Geographical aspect diffusion of passive houses. Int. Rev. Appl. Sci. Eng. 4, 151–156. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 16 Kumar M., Singh, R.K., Raju, P.L.N., Krishnamurthy, Y.V.N., 2014. Road Network Extraction from High Resolution Multispectral Satellite Imagery Based on Object Oriented Techniques. ISPRS Ann. Photogramm. Remote Sens. Spatial Infor. Sci. II–8, 107-110. Lázár, I., 2011. Effects of climate change on renewable energy sources (in Hungarian), in Szabó V, Fazekas I. (Eds.), Környezettudatos energiatermelésés felhasználás, MTA DAB Megújuló Energetikai Munkabizottság, Debrecen, 2011, pp. 92–98. Lewis, N.S., 2007. Towards cost–effective solar energy use. Science 315, 798–801. Lin, X., Zhang, J., 2014. Segmentation–Based Filtering of Airborne LiDAR Point Clouds by Progressive Densification of Terrain Segments. Remote Sens. 6, 1294–1326. Liu, X., 2008. Airborne LiDAR for DEM generation: some critical issues. Prog. Phys. Geog. 32, 31– 49. Lukac, N., Seme, S., Zlaus, D., Stumberger, G., Zalik, B., 2014. Buildings roofs photovoltaic potential assessment based on LiDAR (Light Detection And Ranging) data. Energy 66 (2014) 598–609. Lukac, N., Zlaus, D., Seme, S., Zalik, B., Stumberger, G., 2013. Rating of roofs’ surfaces regarding their solar potential and suitability for PV systems, based on LiDAR data. Applied. Energ. 102, 803–812. May, C.N., Toth, C.K., 2007. Point positioning accuracy of airborne LiDAR systems: a rigorous analysis. Int. Arch. Photogramm. Remote Sens. 36(3/W49B), 107–111. Mongus, D., Lukac, N., Zalik, B., 2014. Ground and building extraction from LiDAR data based on differential morphological profiles and locally fitted surfaces, Int. Arch. Photogramm. Remote Sens. 93, 145–156. Mücke, W., Deák, B., Schroiff, A., Hollaus, M., Pfeifer, N., 2013. Detection of fallen trees in forested areas using small footprint airborne laser scanning data. Can. J. Remote Sens. 39, 32-40. Nagyváradi, L., Gyenizse, P., Szebényi, A., 2011. Monitoring the changes of a suburban settlement by remote sensing. Acta Geographica Debrecina Landscape Env. 5, 76–83. Nex, F., Remondino, F., Agugiaro, G., de Filippi, R., Poletti, M., Furlanello, C., Menegon, S., Dallago, G., Fontanari, S., 2013. 3D solarweb: A solar cadaster in the Italian alpine landscape. Int. Arch. Photogramm. Remote Sens. XL-7/W2, 173–178. Nguyen, H.T., Pearce, J.M., Harrap, R., Barber, G., 2012. The Application of LiDAR to Assessment of Rooftop Solar Photovoltaic Deployment Potential in a Municipal District Unit. Sensors, 12, 4534–4558. Paparoditis, N., Polidori, L., 2004. Overview of digital surface models, In Egels, Y and Kasser, M. (Eds.), Digital Photogrammetry, Taylor and Francis, London, pp. 159–163. Paulescu, M., Paulescu, E., Gravila, P., Badescu, V., 2013. Weather modeling and forecasting of PV systems operation, Springer, London Priestnall, G., Jaafar, J., Duncan, A., 2000. Extracting urban features from LiDAR digital surface models. Comput. Env. Urban Syst. 24, 31, 65–78. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 17 Protic, D., Kilibarda, M., Vucetic, I., Nestorov, I., 2002. 3D roof modelling for accurate assessment of solar potential, Proc. EuroSun 2002 Int. Conference, Bologna, Italy, 5 p. Quazi, A., Fayaz, H., Wadi, A., Raj, R.G., Rahim, N.A., Khan, W.A., 2015. The artificial neural network for solar radiation prediction and designing solar systems: a systematic literature review. J. Clean. Prod. 104, 1–12. Reindl, D.T., Beckman, W.A., Duffie, J.A., 1990. Evaluation of hourly pitched surface radiation models. Solar Energy 45, 9–17. Rosenfeld, G.H., Fitzpatrick-Lins, K., 1986. A Coefficient of Agreement as a Measure of Thematic Classification Accuracy. Photogramm. Eng. Remote Sens. 52, 223–227. Rouse, J.W., Haas, R.H., Schell, J.A., Deering, D.W., 1973. Monitoring vegetation systems in the Great Plains with ERTS, Third ERTS Symposium, NASA SP-351, Vol. 1. NASA, Washington, DC, pp. 309–317. Shan, J., Toth, C.K., 2008. Topographic Laser Scanning and Ranging: Principles and Processing, CRC Press, Boca Raton Shao, P., Yang, G., Niu, X., Zhang, X., Zhan, F., Tang, T., 2014. Information Extraction of HighResolution Remotely Sensed Image Based on Multiresolution Segmentation. Sustainability 6, 5300–5310. Shortis, M.R., Johnston, G.H.G., Pottler, K., Lüpfert, E., 2008. Photogrammetric analysis of solar collectors. Int. Arch. Photogramm. Remote Sens. 37 Part B5, 81–88. Smyth, M., 2012. Solar photovoltaic installations in American and European winemaking facilities. J. Clean. Prod. 31, 22–29. Stevanovic, S., 2013. Optimization of passive solar design strategies: A review. Renew. Sust. Energ. Rev. 25, 177–196. Weih, R.C., Riggan, N.D., 2010. Object-Based Classification vs. Pixel–Based Classification: Comparative Importance of Multi-Resolution Imagery. Int. Arch. Photogramm. Remote Sens. (GEOBIA 2010), XXXVIII–4/C7 Yu, B., Liu, H., Wu, J., Hu, Y., Zhang, L., 2010. Automated derivation of urban building density information using airborne LIDAR data and object-based method, Landscape Urban Plan. 98, 210–219. Zar, J.H., 1999. Biostatistical Analysis. Prentice Hall, New Jersey, Zhou, Q., Neumann, U., 2013. Complete residential urban area reconstruction from dense aerial LiDAR point clouds. Graph. Models 75, 118–125. Zlinszky, A., Schroiff, A., Kania, A., Deák, B., Mücke, W., Vári, Á., Székely, B., Pfeifer, N., 2014. Categorizing Grassland Vegetation with Full-Waveform Airborne Laser Scanning: A Feasibility Study for Detecting Natura 2000 Habitat Types. Remote Sens. 6, 8056–8087. Web references 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 18 https://www.maxmax.com/RemoteSensingcamerasi.htm, LDP-LLC Ltd. https://www.terrasolid.com/download/tscan.pdf, Terrasolid, Ltd., Terra Scan User’s Guide, 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 19 Table 1. Difference of roof heights (relative heights of LiDAR surface subtracted from measurements) Roof type Lower quartile Median Upper quartile RMSE Flat -0.01 0.17 0.45 0.30 Shed -0.18 0.03 0.20 0.31 Gable -0.02 0.11 0.29 0.34 Combination -0.25 0.01 0.23 0.43 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 20 Table 2. Number and area of roof planes in the campus area Method Number of detected planes Sum of the area [m2] LPC 68 5432 SDSM 79 4893 PDSM 78 5197 OPDSM 52 5239 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 21 Table 3 Statistical characteristics of roof planes of the hostel building, considering the four calculation methods (N=4 according to the four methods; unit: m2) Roof plane IDs 16 20 23 24 31 Minimum 150.76 78.17 244.30 76.44 30.60 Maximum 182.54 100.47 284.89 96.97 70.32 Mean 160.19 90.53 257.15 88.12 52.08 Standard error 7.49 5.35 9.45 5.21 9.48 Standard deviation 14.97 10.70 18.90 10.43 18.97 Median 153.72 91.74 249.69 89.53 53.70 25 percentile 151.45 79.90 244.75 77.88 33.42 75 percentile 175.39 99.95 277.00 96.94 69.13 Coefficient of variation 9.35 11.82 7.35 11.83 36.42 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 22 Table 4. Comparison of the incoming solar energy on the roof planes of the hostel building (p values, Wilcoxon test with Bonferroni correction; (LPC: LiDAR point cloud processing; SDSM: DSM from LiDAR + segmentation; PDSM: DSM from aerial photographs + segmentation; OPDSM: DSM from aerial photographs combined with orthophoto + segmentation) LPC SDSM PDSM SDSM 0.5455 PDSM 0.9091 1 OPDSM 1 1 1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 23 Table 5. Incoming solar energy on the hostel building’s roof planes (Incoming solar irradiation [kWh]; Roof plane IDs correspond to Fig. 4) Roof plane ID Roof plane surface Calculated roof plane surface Surface for PV panels Calculated surface for PV panels Surface for PV panels except northern directions Calculated surface for PV panels except northern directions 0 462203 438925 245800 256056 245800 256056 1 6622 0 0 2 7839 0 0 3 163404 109964 79207 33421 0 0 4 27739 25715 11044 7305 0 0 5 197275 54191 32885 19126 0 0 6 21756 26707 7650 7586 7650 7586 7 22204 24842 0 5565 0 0 8 74579 76779 26940 32458 26940 32458 9 117252 82262 37154 29033 37154 29033 10 30465 36849 2055 8254 2055 8254 11 428893 448378 228987 290448 228987 290448 12 62024 76814 21623 32472 21623 32472 13 158239 146923 72752 63890 72752 63890 14 127699 122964 56923 61851 0 0 15 275257 284305 142410 176715 142410 176715 16 37727 76814 7437 32472 7437 32472 17 276441 275325 139278 173186 139278 173186 18 158103 152502 73845 73886 73845 73886 19 68755 76779 25157 32458 25157 32458 Total 2710015 2551500 1211148 1336183 1031089 1208917 Ratio1 [%] 100 94.2 100 110.3 100 117.2 Ratio2 [%] 100 94.2 44.7 49.3 38.0 44.6 1solar energy calculated for area provided by manual method/solar energy calculated for area provided by automated method 2solar energy calculated for separate cases /solar energy calculated for the total roof area provided by the automated method 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 24 Table 6. Comparison of different surveying techniques and possible returns in terms of solar panels Calculations based on blueprints Drone survey LiDAR survey area/object usually 1 house ~1-5 km2/day ~800-1000 km2/day absolute cost low low High relative cost (price/building) high low Low cost/benefit can be financed by a single household can be financed by a local authority or firm should be financed by a project fund return soon soon there is no direct return IT infrastructure requirement low medium high HR expertise requirement medium high high Figure 4 Click here to download high resolution image Figure 5 online Click here to download high resolution image Figure 5 printed Click here to download high resolution image