Mastertheis Ana Gonzalez Quintairos SENDEDVERSION

Technische Universität Berlin

Master Thesis

Analysis of potential distribution and size of photovoltaic systems on rural rooftops

A contribution to an optimized local energy storage system with a remote sensing and GIS-based approach in Swabia, Germany

A thesis submitted in fulfilment of the requirements for the degree of Master of Science Environmental Planning

Ana González Quintairos Matriculation number: 349833

First Supervisor: Prof. Dr. Kleinschmit

Second Supervisor: Dr. Jochen Bühler

Faculty VI Planning-Building-Environment Geoinformation in Enviromental Planning

January 2015

In cooperation with Reiner Lemoine Institut gGmbH

Abstract

This work studies the distribution of photovoltaic systems in rural areas. The aim of the study is to create a method which predicts the size and location of future photovoltaic systems on rooftops. Only very few authors have also attempted to use high-resolution images to quantify the suitable rooftop surface per building and the appropriated location of the panels and none of them have addressed the particular building distribution and typology of rural communities. Addressing rural areas has a tremendous importance in Germany where highest PV potential in Germany is expected. The methodology uses as inputs high-resolution aerial imagery, GIS building footprint from the Land-register map and the Bavarian database of photovoltaic systems. The method has been applied to the village of Freihalden (Bavaria) using two different type of images: official orthophotos from the Bavarian Land-survey Office and Google Earth™ orthophotos and the results on rooftop suitable area and on rooftop potential for PV have been compared with each other. The study area is located in Freihalden, Bavaria.

First, the current status quo was analysed, cross-referencing the files from the Bavarian register of photovoltaic systems and the cadastre data. These buildings were excluded from dataset in order to avoid assigning them as potential building for PV in the future. To calculate the spatial distributed potential of the installed photovoltaic power on rooftops in the future, a pixel-based image analysis is performed on each image to identify suitable rooftop areas, roof obstructions and shadows. The output of the image classification analysis is converted to vector data and the potential suit-able area is assigned to each building. Rooftop orientation and average slope are spatially joined to each rooftop. The results together with the location in coordinates of each building are giving as input to the PV Calculator, developed by the RLI, to obtain the potential annual energy production of each building. Finally, the prognosis step predicts the likely PV-expansion pathway based on each rooftop PV potential and the scenarios of the German Energy Agency. The output of the method is stored in a database including the central coordinates of each building.

Freihalden was found to have 43 buildings with already PV systems in its rooftops which represented an already installed nominal power of 571 kWp. The classification of the Ba- varian land-register orthophoto concluded that 82% of the buildings in the community have adequate areas with more than 10 m2 suitable for PV, where as in the classification of the Google Earth™ orthophoto 78% of the buildings are considered to have adequate

i Abstract

areas for PV. In average 39% of total rooftop area is considered suitable for PV. This number can be used a rule-of-thumb for future studies in the area. The total PV technical potential of Freihalden reaches the 3170 kWp. Individual building potential ranges from 1 potential kWp to 42 kWp and the specific yield varies from 980,1 to 763,1 kWp/kWh. From the 412 buildings composing the village, 98 will be required to installed new PV systems if the municipality has to fulfil its share on the Bavarian renewable energy goals for 2030. The comparison shows that both airborne high-resolution orthophoto and high-resolution Google Earth™ are capable of delivering trustworthy results with 11% and 16% error respectably using open-source software.

The method developed in this study has been further used to estimate the PV potential and prognosis expansion pathway of two more municipalities in Bavaria supporting the research of the project Smart-Power-Flow at the Reiner Lemoine Institute.

ii Table of contents

Table of contents

Abstract ...... i Table of contents ...... iii List of figures ...... iv List of tables ...... vi List of maps ...... vi Abbreviations ...... vii Chapter 1 Introduction ...... 1

1.1 Background ...... 1 1.2 Context and objectives ...... 2 1.3 Structure of the thesis ...... 3

Chapter 2 State of the art ...... 4 2.1 Suitable rooftop area ...... 4 2.1.1 Constant-value methods ...... 5 2.1.2 Remote sensing methods ...... 6 2.1.3 3D/LiDAR methods ...... 7 2.2 Global solar radiation ...... 8 2.3 PV energy production ...... 10 2.4 Data availability ...... 13 2.5 Research approach and research questions ...... 16

Chapter 3 Methods ...... 17

3.1 Description of the study area ...... 17 3.2 Input data ...... 19 3.2.1 General cartography ...... 19 3.2.2 Land register map ...... 20 3.2.3 Official orthophotos from BVW ...... 21 3.2.4 Google Earth orthophotos ...... 22 3.2.5 Bavarian photovoltaic database ...... 23 3.2.6 Statistics per municipality ...... 23 3.3 Research design ...... 23 3.3.1 Data preparation ...... 25 3.3.2 Isolating building rooftops ...... 28

iii

3.3.3 Analysis of status quo ...... 28 3.3.4 Pixel-based classification ...... 29 3.3.5 Computation of total suitable area ...... 34 3.3.6 PV energy production ...... 35 3.3.7 Validation ...... 36 3.3.8 Prognosis of the most probable locations for PV systems ...... 37

Chapter 4 Results ...... 40 4.1 Status quo ...... 40 4.2 Suitable area computation ...... 42 4.2.1 Bavarian land-register orthophoto ...... 43 4.2.2 Google Earth images ...... 44 4.3 PV technical potential ...... 46 4.3.1 Bavarian land-register orthophoto ...... 46 4.3.2 Google Earth images ...... 47 4.4 PV prognosis expansion pathway ...... 48

Chapter 5 Discussion and conclusion ...... 50

5.1 Comparison of results ...... 50 5.2 Review of methodology ...... 52 5.3 Future research and outlook ...... 53 5.4 Conclusion ...... 54

Glossary ...... 55 References ...... 56 Appendix A: Findings in literature review ...... 60 Appendix B: Smarth Power Flow municipalities ...... 64 Appendix C : Accuracy assessments ...... 65 Appendix D : PV potential and expansion pathways ...... 68 Appendix E : Maps ...... 91

iv List of figures

List of figures

Figure 2.1. Components of solar radiation on a slope [31]...... 9 Figure 2.2. Definition of tilt () and azimuth () angles for PV applications...... 10 Figure 2.3. Effect of the azimuth and tilt on the energy yield of a PV panel [45] .... 11 Figure 3.1. Example of settlement category: rural (left), village (centre), suburban (right) [Source: Google Earth™ orthophotos] ...... 17 Figure 3.2 Village of Freihalden in Jettingen-Scheppach municipality ...... 18 Figure 3.3. Residential houses with different roof angles in Jettingen-Scheppach ..... 19 Figure 3.4 Extent of the four orthophotos from BVW...... 22 Figure 3.5 Flowchart for calculating PV potential for this study ...... 24 Figure 3.6. PDF land-register map provided by the regional network operator LVN (above) and land-register map after digitalization (below)...... 27 Figure 3.7 Orthophoto and building footprints before and after correction ...... 29 Figure 3.8 Detail of the obstruction, shadows, suitable and non-suitable areas of one household in Freihalden...... 30 Figure 3.9 Signature plots ...... 31 Figure 3.10 new signature plots after building categorization: red roofs (left), grey roofs (center) and black roofs (right) ...... 31 Figure 3.11 Histograms for band 1, band 2 and band 3 of grey roof image...... 32 Figure 3.12 Original orthophoto clipped by the building footprint (left) and classified output after supervised image classification (right) ...... 33 Figure 3.13 Building before (left) and after applying post-processing (right)...... 34 Figure 3.14 Determination of the rooftop area with roof inclinatio angle = 35º .... 34 Figure 3.15 German Energy Agency classes [5] ...... 38 Figure 4.1 Houses with installed rooftop PV systems (purple) in Freihalden ...... 41 Figure 4.2 Supervised image classification based on the BVW orthophoto ...... 43 Figure 4.3 Photovoltaic suitability map based on BVW orthophoto ...... 44 Figure 4.4 Supervised image classification based on Google Earth™ orthophoto ..... 45 Figure 4.5 Photovoltaic suitability map based on Google Earth™ orthophoto ...... 46 Figure 4.6 PV technical potential map based on BVW orthophoto ...... 47 Figure 4.7 PV technical potential map based on Google Earth™ orthophoto ...... 48 Figure 4.8 Prognosis expansion pathway based on BVW orthophoto ...... 49 Figure 4.9 Prognosis expansion pathway based on Google Earth™ orthophoto ...... 49 Figure 5.1 Comparison of image classification statistics ...... 50 Figure 5.2 Comparision of orthorectification on a small subset of Freihalden ...... 52

v List of tables

List of tables

Table 2.1 Classification of references according to its methodology ...... 5 Table 2.2 Available data sources and used data (green) for the master’s thesis ...... 14 Table 3.1 Correspondence of DXF features with Land register map´s elements ...... 26 Table 3.2 Class types ...... 32 Table 3.3 Input data in EVA PV Calculator ...... 36 Table 3.4 Jettingen-Scheppach in Dena network classes ...... 37 Table 4.1 Installed PV systems in Freihalden ...... 41 Table 4.2 Classification statisitics ...... 43 Table 4.3 Classification statistics ...... 45 Table C.1 Accuracy assessment red roofs in Bavarian land-register orthophoto ...... 65 Table C.2 Accuracy assessment grey roofs in Bavarian land-register orthophoto .... 65 Table C.3 Accuracy assessment black roofs in Bavarian land-register orthophoto ... 66 Table C.4 Accuracy assessment red roofs in Google Earth™ orthophoto ...... 66 Table C.5 Accuracy assessment grey roofs in Google Earth™ orthophoto ...... 66 Table C.6 Accuracy assessment black roofs in Google Earth™ orthophoto ...... 67 Table E.1 Prognosis and expansion pathway based on BVW orthophoto ...... 68 Table E.2 Prognosis and expansion pathway based on Google Earth™ image ...... 80

List of maps

Map 1 Freihalden status quo: installed PV systems ...... 91 Map 2 Building rooftop classification based on BVW orthophoto ...... 92 Map 3 Building rooftop classification based on Google Earth™ orthophoto ...... 93 Map 4 Suitable areas for PV based on BVW orhtophoto ...... 94 Map 5 Suitable areas for PV based on Google Earth™ orthophoto ...... 95 Map 6 Building PV technical potential based on BVW orthophoto ...... 96 Map 7 Building PV technical potential based on Google Earth™ orthophoto ...... 97 Map 8 PV expansion pathway based on BVW orthophoto ...... 98 Map 9 PV expansion pathway based on Google Earth™ orthophoto ...... 99

vi Abbreviations

Abbreviations

AC Alternating Current ALKIS Amtliches Liegenschaftskatasterinformationssystem (Official Real Estate Cadastre Information System) ATKIS Amtliches Topographisch-Kartografisches Informationssystem (Official Topographic-Cartographic Information System) BVW Bayerische Vermessungsverwaltung (Bavarian Land-survey Office) CAD Computer-aided Design DC Direct Current DEM Digital Elevation Model

DSM Digital Surface Model ESRA European Solar Radiation Atlas GIS Geographic Information System GW Gig watt HVAC Heating, Ventilation and Air Conditioning IEA International Energy Agency JRC Joint Research Centre of the European Commission kW Kilowatt

kWh Kilowatt hour kWp Kilowatt peak LVN LEW Verteilnetz GmbH (regional network operator)

LiDAR Light Detection and Ranging NASA National Aeronautics and Space Administration NREL National Renewable Energy Laboratory PV Photovoltaic

PVGIS Photovoltaic Geographical Information System RBT Representative Building Typology

STC Standard Test Conditions for PV modules

vii

Chapter 1

Introduction

The first chapter of this thesis introduces the relevant background information about renewable energy policies in Germany. Secondly, it frames the contexts of the thesis and specifies as well the master thesis objectives. The final research approach and the research questions are not presented here but in the next chapter after the literature research. Finally, the structure of the thesis concludes the chapter.

1.1 Background

The climate policy objectives of the European Union state that by 2020 a share of 20% of primary energy consumption in the European Union should be covered from renewable sources [1]. Germany has even more ambitious goals: the share of renewables in electricity supply must grow as high as 35% by 2020 and 80 % by 2050 [2]. Germany is trying to leave the fossil-nuclear age behind and therefore decentralised solar energy sources are expected to play a major role on the German sustainable power production market, in the near future.

Solar power in Germany has been increasing exponentially for the last 20 years, indeed. Ac- cording to the German Federal Network Agency, the total nominal power of photovoltaic (PV) installed in Germany rose from 17.4 GW in 2010 to 35.7 GW by the end of 2013 [3, 4] and it is expected to reach 42.4 GW of installed capacity by 2020 [5]. Last year, PV- generated energy totalled 29.7 TWh (Figure 1.1) and covered approximately 5.7 percent of Germany’s net electricity consumption [6] . Taken as a whole, renewable energy accounted for 29 percent of net electricity consumption, while the proportion of PV and total RE in Ger- many’s gross electricity consumption stood at 5 percent and 24 percent respectively [3].

Unlike other energy sources, when it comes to photovoltaic generation, small private photovoltaic systems represent, in numerical terms and in terms of performance, the largest share [7]. The Frauhofer Institute (2014) [3] accounts for more than 1.4 million distributed solar PV generation plants just in Germany. With this figure, the installed PV capacity exceeds of all other types of power plants in Germany. These systems are mostly rooftop installations and usually feed into the low-voltage distribution networks, to which this work focus. As a consequence, loads and a reversal power flow direction occurs at certain hours in the year in many low-voltage distribution networks which threatens the network’s stability [7].

Chapter 1. Introduction 2

Figure 1.1. PV-generated energy production in Germany 2000 – 2013 [4]

1.2 Context and objectives

The master thesis will be conducted at the Reiner Lemoine Institute, an independent research institute to support scientifically the turnaround in energy policy towards a climatefriendly energy supply based on renewal energies. The work on this theme is embedded in the joint research project Smart Power Flow, a study aiming to optimize of network expansions and usage of local energy storage systems, due to the rapid increase of renewable energies in Ger- many. In order to accomplish a successful turnaround in German energy policy, it is of great importance to optimize the operation of the low-voltage distribution networks besides the high-voltage networks normally discussed in the media. The project study area comprises Swabia and is funded by the German Federal Ministry of Economy and Energy1.

The objective of the master thesis is to estimate the size, potential and location of future photovoltaic systems on rooftops in rural areas. Rural communities are where the highest photovoltaic growth potential within Germany is expected [5] but also where the availability of data is more scared. The methodology is based on land-register GIS data from the regional network operator and it calculates the status quo and extension of the installed photovoltaic power on rooftops in the future. The ambition is to use only existing or free-of-charge data and open-source software, policy that is trying to support the Institute. Particularly important for the project is the prognosis step: the most likely installation time and the probability that the PV system will be built, based on the size and the scenarios projected by the German Energy Agency. The final model will be later used to estimate the new demand profile in distribution networks to optimize the network expansion in Bayern.

1 Project funded by the Federal Ministry of Economics and Technology (BMWi) under grant agreement no. 0325522A.

Chapter 1. Introduction 3

1.3 Structure of the thesis

The thesis is structured as follows: the first part of Chapter 0 contains a review of the state of the art and the relevant scientific basics, with a focus on existing rooftop estimation methods. The second part of the chapter presents the data available for this study, the limitations and the research approach and questions. Because of the importance of the literature research on the final research approach, which customarily belongs to the first chapter, these are presented together. Chapter 3 describes the study area and input data in detail, and explains how the model development and its subsequent analysis were carried out. After outlining the general structure at the beginning of the chapter, important implementation aspects are described in depth in the corresponding section. In Chapter Chapter 1 methodological results are presented. Calculations are made for one Smart Power Flow exemplary village. Subse- quently, the results are then, in Chapter Chapter 1, compared and discussed. Finally, the closing section of chapter 5 gives a summary and outlook and conclude the thesis.

Chapter 2

State of the art

The potential for photovoltaic electricity generation depends on a number of global, temporal, spatially and technically variable conditions. This chapter discusses the factors that influence PV potential including available rooftop area, incoming solar radiation and the energy production from PV systems. Subsequently, the next section presents the research of the data available for the master thesis, from which the limitations of the study are derived. The availability of free-accessible data sources will likewise determinate the research approach chosen for the thesis as well as the research questions, presented in the last section.

2.1 Suitable rooftop area

The two most crucial components for calculating PV potential include the amount of usable rooftop area for photovoltaic installations and the quantification of the incoming solar radiation [8]. Among those, the main difference in the literature review concerns the method used to determine the available rooftop area, which will be discussed in this section.

The assessment of the rooftop-installed photovoltaic potential on buildings usually starts with the determination of the total rooftop area. Once the total rooftop area for the targeted region is calculated, it is necessary to reduce this area to the one available for solar photovoltaic applications, in order to determine the potential power output [9]. Many approaches for estimating the rooftop potential have been developed. Table 2.1 subdivides the rooftop-installed photovoltaic potential’s literature into three categories according to their methodological approach. In general the methods can be classified depending on the input data used for the analysis as [10]:

2.1.a) constant-value methods 2.1.b) remote sensing methods 2.1.c) 3-D/LiDAR methods

Table A.1 in Appendix A presents a detail description of the main findings of the literature review, which includes a classification of the most significant papers according to the methodology, the input data and the software used in the study, the size of the study area, the small-

Chapter 2. State of the art 5

est sampled unit, the reduction coefficients (if applied) and the results obtain by the authors. The strong variation among results suggests that other literature alone cannot be to later validate estimates of rooftop PV availability for the study area; rather, area- and project- specific validation methods are required [10].

Table 2.1: Classification of references according to its methodology (Table adapted from NREL [10])

Constant-value methods Remote Sensing methods 3D/LiDAR methods

Methodolo- Estimation of a general Direct computation using of Creation of a 3D sur- gy reduction coefficient ac- high resolution images to face elevation model cording to building typolo- identify suitable rooftop from Laser data to gy that applies to the total which a solar radiation roof area model is applied

Advantages • Quick • Building detail-specific • Building detail- specific • Easy to compute • Affordable input data • Excellent solar irradiation prognosis

Disad- • Do not consider build- • Time-intensive • Time-intensive vantages ing specific characteris- • Requires image pro- • Computer-intensive tics cessing knowledge • Expensive or not • Result are difficult to existing input data validate

References [5–7], [9–14] [15–20] [21–25]

2.1.1 Constant-value methods

The National Renewable Energy Laboratory (NREL) 2013 in its latest report classifies the reduction coefficient methodologies as “Constant-Value Methods” . These methods consider typical rooftop configurations and try to estimate a coefficient that can be applied to the entire target area [10]. Rough general assumptions are made about the proportion of sloped versus flat roofs, number of buildings with desirable rooftop orientations, amount of space obstructed by building components (such as HVAC installations, elevators, or penthouses) and shadows to determine the relation between total rooftop area and rooftop available area [11].

Izquierdo et al. [11] presented a hierarchical methodology to systematically compute the phys- ical and technical limits of photovoltaic solar energy in roofs based on representative building typology (RBT). The available rooftop area is calculated by multiplying the RBT building footprint by three coefficients accounting for: a) voids and recesses in buildings; b) shadowing generated by other buildings, objects, or by the roof configuration itself; c) other specific obstructions (e.g. aerials, stacks or HVAC equipment) [10]. These coefficients are created by

Chapter 2. State of the art 6

expert knowledge looking at aerial images. With an error or 32%, the mean available roof area per capita is 14.0±4.5 m2/ca. The economic potential and the social potential are not included in the study. No method validation is included.

The International Energy Agency (IEA) [12] have also estimated the average available rooftop areas for member states; albeit their procedure is not described in detail, their results are often used as a reference given the lack of other data sources [10].

A variation of the constant-value method is to determine a correlation between rooftop area and the region’s population density. Such studies apply a constant value to calculate the suitable rooftop area for PV but using a population-density formula to generate a more specific values [10]. In Germany, Lehman & Peter [13] use a data set of buildings in North Rhine- Westphalia. They studied the correlation curves between rooftop area and population density and applied their results across all European Union countries to estimate potential solar energy generation. 13.4 m!/ca of rooftop area and 7.1 m!/ca of façade area are suitable for PV in Germany [13].

This approach is also used by the German Energy Agency [5] where the prognosis of installed PV potential for the year 2020 is regionalized according to the population density of each municipality. Kerber [7] combined a unified reduction coefficient of 0.5 extracted from literature with population density and economic factors to calculated the PV potential.

2.1.2 Remote sensing methods

Remote sensing methods rely heavily on the existence of high-resolution satellite imaginary, remote-sensed data or aerial orthophotos. Although some reduction coefficients are sometimes applied, remote sensing methods are a more refined—but more time-intensive—method for identifying suitable rooftop space than constant-value methods. The imagery is used mostly to calculate the total rooftop area by pixel- or object-recognition software whne no cadastral or Land-register maps ara accessible. Subsequently it estimates the potential power output in combination with other factors. Only recently, few authors [16–18] have also attempted to use orthophotos to quantify the obstructed space and the appropriated location of the PV systems on building basis.

Vardimon [20] calculated the rooftop area from data compiled by the Israeli Bureaus of Sta- tistics combined with photogrammetric analysis of orthophotos to extract buildings footprint. After extracting the building polygons, hence knowing the total rooftop area, a conservative reduction factor of 30% is used to estimate the mean availability for all rooftops [20]. Two scenarios are developed, “Total Potential” scenario where all rooftops are accounted for and an “Economic Scenario”. The mean available rooftop area per capita in the region was 10.2 m2. No error calculation or validation for results is conducted.

Wiginton et al. [21] demonstrate techniques to merge the capabilities of geographic information systems and object-specific image recognition to determine the total rooftop area for PV in an example large-scale region in south eastern Ontario. The roof extraction procedure, which involved the use of the Feature Analyst (FA) image recognition program an extension to ArcGIS, has not been used previously for PV quantification by any other study [21]. The

Chapter 2. State of the art 7

most conservative estimates of roof area coeficients from the literature are used. Residential and other small buildings with pichted rooftops are considered to have 50% south-facing area on average. For shading a conservative reduction coefficient of 0.30 is selected. These coefficients are then multiplied by the total roof area. The methodology is based on Izquierdo et al. [11] but using FA tool for building selection. Still in rural areas the digitalization was made manually. A relationship across the region was found between available roof area and population of 13.1 m2/capita ± 6.2%

The Chandler study [18] assesses imagery for brightness value to determine the potential areas of shadow and obstructions on commercial and government rooftops in a 4 square mile area of the city. The lengths of shadows are estimated in detail using Google SketchUp. The obstructions identified during the satellite imagery processing are buffered by a distance equal to the shadow lengths estimated with the software SketchUp. These areas are then subtracted from the total rooftop area to estimate the rooftop area suitable for PV. The researchers vis- ited 150 of the 932 buildings included in the study to take ground-truth measurements and validate the accuracy of their model.

Google Earth™ is also used, along with local construction data, in a study for Andalusia, Spain [22]. This analysis uses sampling methods and manual rooftop identification to estimate the availability of residential rooftop space for PV. The analysts sample buildings by type, including pichted/semi-pichted houses, townhouses/row houses, and high-rise buildings. The footprint for each sampled building is manually digitized from Google Earth™, and a 3-D model is created in AutoCAD. Using the 3-D model, areas obstructed by HVAC equipment, antennas, chimneys, and other objects are excluded from the total rooftop area. Rooftops are categorized by flat and pitched roofs, and a 1 meter perimeter is assumed necessary around all installations to account for maintenance work [22]. The sampled buildings suggest that 51% to 55% [22] of flat roof surface area and 16% to 21% [22] of pitched roof surface area could be used for PV.

2.1.3 3D/LiDAR methods

3-D methods use primarily Digital Surface Models (DSM) models to determine incoming solar radiation on tilted surfaces or shadow effects on buildings. Ideal values for rooftop characteristics are input into a computer model, and the GIS software determines areas of high suitability. The DSM models are most often generated from overlapping orthophotography or Light Detection and Ranging (LiDAR) data, and they are combined with slope, orientation, and building structure data to estimate the total solar energy generation potential [10]. As LiDAR data has become more widely available at higher resolutions in recent years, this procedure has become a much more desirable method for estimating rooftop area [10]. A high proportion of rooftop analyses in the past decade have followed this approach [22–26, 28, 29]. These techniques are often only possible with high cost data and feature recognition software.

Nguyen et al. [23, 30] explain how accurate building generation from LiDAR data requires a number of processes including building detection, object segmentation, roof shape reconstruc- tion, and modelling quality analysis. The study processes 3-D building models in r.sun, a GRASS solar radiation tool, to create a solar radiation map of the city. Rooftops with a

Chapter 2. State of the art 8

southeast- through southwest-facing aspect (90 to 270 degrees) and a slope within 15 degrees of the local latitude are considered suitable for PV [20].

A study of Lisbon, Portugal [23] uses LiDAR data to create a DSM that is then used as an input in the ArcGIS Area Solar Radiation tool to create a solar surface map. A solar radiation map is created for each month of the year, and the 12 values for each pixel are averaged to determine a final solar radiation value. Rooftop areas are considered suitable for PV if at least 10 contiguous square meters have more than 1.68 megawatt-hours per square meter (MWh/m2) of solar radiation and have a slope of less than 45 degrees [23].

RADIANCE/DAYSIM, a program designed to simulate lighting scenarios, is used for a study in Fribourg (Switzerland) [27] and an analysis for Cambridge (Massachusetts, USA) [25], conducted by the Massachusetts Institute of Technology (MIT). The Fribourg analysis uses a 3-D building model for 61 buildings in the study area. The methods for developing the 3-D building model are not discussed. The MIT study consider rooftop area to be suitable for PV if it has a slope less than or equal to 60 degrees and a solar resource value of at least 609 kWh/m!. Results of this analysis are presented on a building-specific level and are accessible through an interactive Web interface [25].

2.2 Global solar radiation

The design of PV systems requires the knowledge of incident solar radiation on the installation site. This is one of the essential parameters for photovoltaic electricity generation [31, 32]. The present section will explain the different types of incident solar radiation and the existing methods to calculate the energy received at any point on the terrain.

Solar radiation, or insolation, is the sun´s energy reaching the earth´s surface [33]. There are three sources of insolation on a slope: direct radiation, diffuse radiation and ground-reflected (albedo) radiation [31, 33, 34]. Figure 2.1 displays the way the three components reach a surface on earth. Direct radiation is the direct beam of the sun received by the surface without any interaction with the atmosphere [31, 35]. Diffuse radiation consists of intercepted radiation scattered in the atmosphere by clouds, gases and aerosols [32, 35]. Last, albedo is the amount of radiation reflected by the ground and nearby objects [31, 34]. The total radiation reaching the surface is the sum of the three components of radiation.

In addition, solar radiation is affected by the earth´s rotation and revolution around the sun [32, 33]. It also varies with environmental factors like weather conditions, humidity or water vapour [31]. At the regional and local scale, the relief is the major modifying the radiation received at a specific point. Differences in elevation, surface inclination (slope) and shadows originated by the terrain creates strong local variations [33]. This leads to high spatial and temporal heterogeneity of incoming radiation that can significantly differ from the nearest ground station; data collected from stations 20-30 km from a project can have a mean square error of 25% [32, 34, 36].

Chapter 2. State of the art 9

Figure 2.1. Components of solar radiation on a slope [31].

To overcome the scarcity of accurate insolation maps that would require a dense collection station network, a number of empirical models have been developed to predict the total solar radiation reaching a surface [36]. Šúri & Hofierka [33] as well as Katiyar & Pandey [36] pro- vide an extensive overview of the existing empirical models. One of the first GIS-based irradiation models was SolarFlux [35], a model developed in 1995 for Arc/Info GIS platform. It simulates the influence of terrain patterns on direct insolation at specific intervals of time using as input topographic elevation values, latitude and atmospheric conditions [35]. A more advanced method and worldwide used is Solar Analyst [32] implemented also as an extension of ArcView GIS. Direct and diffuse radiations are calculated using a hemispherical viewshed that represents which sky directions are visible and which are obscured [32, 33]. Although reflected radiation is not included, Solar Analyst extension is regarded as suitable for fine scale studies [33] and it is fully integrated in analysis toolbox available in ArcGIS. The SRAD model [37] was designed to model a complex set of short- and long-wave interactions of solar energy with the earth’s surface and atmosphere. However it is designed for the modelling of mesoscale processes and the calculation over large territories is also limited [38].

Finally, the r.sun irradiance calculation model was developed at the year 2000 on a project of the European Solar Radiation Atlas (ESRA) and is implemented in the GRASS GIS environment. The method is based on the equations published in the ESRA [33, 39] and further expand the previous work of Hofierka [40]. It can be used at various scales and was especially created to overcome the shortcoming of the other models [33]. The r.sun model calculates all components of solar radiation (direct, diffused and reflected) and have been designed to consider diffuse radiation that specifically reflective of European climate conditions [33]. The model output is a solar radiation database in form of maps (raster format) and a web-based tool available for whole Europe and Africa. The database consists of raster maps representing twelve monthly averages and one annual average of daily sums of global irradiation for horizontal surfaces, as well as those inclined at angles of 15, 25, and 40 degrees [41].

This models represent the core of most PV Calculators explain in the following section.

Chapter 2. State of the art 10

2.3 PV energy production

Understanding available solar radiation over the terrain and computing rooftop area are essential components when calculating photovoltaic energy potential but there are also techno- logical considerations to take into account. These include: • azimuth • tilt • panel efficiency

The azimuth angle (or orientation) is the angle between south and the plane of the module in the northern hemisphere. This angle is taken as negative toward east, i.e. goes in the anti- trigonometric direction and it is commonly denoted by [42] (Figure 2.2). Most rooftop PV systems are mounted on racks in a fixed position, and do not have the capability to follow (track) the sun throughout the day. Taking this into consideration, the best orientation in the northern hemisphere is usually due south [43].

The tilt angle (or inclination) is the angle between the horizontal and the plane of the module, denoted by . Generally, a surface with a tilt angle similar to the latitude of the location receives maximum insolation. As a rule of thumb, if the main loads are in winter months when solar availability is reduced, tilt angles should be more vertical (approximately equal to latitude plus 15°) to maximize exposure to the low winter sun. If major loads are cooling and refrigeration the tilt angle should be reduced (approximately latitude minus 10°) to maximize output during summer. For grid connect systems the summer optimum angle should be used to maximize annual output of the modules [41, 44].

Figure 2.2. Definition of tilt () and azimuth () angles for PV applications.

The combination of tilt and azimuth influences the energy output of the PV system. In Ger- many for a south-facing PV system, the optimum tilt angle is 30° to 45° [45]. However, small deviations to the east or west do not reduce the energy yield significantly (Figure 2.3). More

Chapter 2. State of the art 11

substantial output reductions are recorded where the orientation is more than 40 degrees off south [46] for the purpose of this stuy. In rooftop-installed PV systems the tilt and the azimuth is given by the orientation and the inclination angle of the roof.

Tilt (inclination) angle: 0° All azimuth direction: 90%

Tilt (inclination) angle: 30° Azimuth direction: W: 85% SW: 95% S: 100% SO: 95% O: 85%

Tilt (inclination) angle: 45° Azimuth direction: W: 82% SW: 92% S: 96% SO: 92% O: 82%

Figure 2.3. Effect of the azimuth and tilt angle on the energy yield of a PV system [45]

The central component of a PV system is the PV module. A PV module consists of many solar cells made of semiconductors. When the sun shines on the solar cell, it converts the incident solar energy into electrical energy. The incident light is absorbed in the semiconductor, and generates positive and negative charge carriers (particles free to move, carrying an electric charge). In the semiconductor an electric field is created, which separates the positive and negative charge carriers from each other and derive separate contacts. The two outer contacts connected to one another, an electric current (direct current) flows. An inverter converts the direct current generated in the typical household 230 volt AC power. Nowadays the most employed PV modules technologies are essentially three: mono-crystalline silicon, poly-

Chapter 2. State of the art 12

crystalline silicon and thin film (amorphous silicon). Monocrystalline silicon PV cells are made from silicon wafers that are cut into quadratic cells from cylindrical single-crystal silicon ingots [47]. Refined silicon is wasted in the cell production process and due to the careful and slow manufacturing processes required. Although the monocrystalline modules are the oldest and most expensive, are also regarded as the most efficient sunlight conversion technology available. Module efficiency averages about 10% to 18% [9, 12]. Polycrystalline PV cells are made from large blocks of molten silicon, carefully cooled and solidified. They are less expensive to produce than monocrystalline silicon PV cells because the manufacturing costs are also lower, but have a slightly lower conversion efficiency compared to single crystalline. Module efficiency averages about 10% to 13% [10, 46]. Thin film PV cells are constructed by depositing extremely thin layers of photovoltaic semi-conductor materials onto glass, stainless steel or plastic. This leads to reduced processing costs from the raw materials but also tends to reduce energy conversion efficiency. Module efficiency averages 5% to 7% [47].

Solar panel efficiency refers to how much of the incoming solar energy is converted into electrical power. The efficiency depends on several factors, being the module type, the temperature and the irradiation the most important. The performance of PV modules is generally rated according to their maximum DC power output (watts) under Standard Test Conditions (STC). Standard Test Conditions are defined by a module operating temperature of 25°C, and incident solar irradiation level of 1000 W/m! and under Air Mass2 1.5 spectral distribution.

In order to assist coping with the variability associated with PV potential, a number of PV simulation tools have been created. Researches [8, 24] have recognized the utility of these tools to overcome existing barriers to implement PV systems by providing information needed for design, financing and operation of PV systems. The following paragraphs discuss the most well-known PV simulation tools available and the variables considered in the formulas utilized to calculate PV potential from solar radiation data. The PVWATTS [48] Calculator3, created by the US National Renewable Energy Laboratory in the year 2000, estimates the energy production for grid-connected crystalline silicon PV systems. It is among the most popular of its kind and uses meteorological solar datasets from the National Solar Radiation Database. The software translate the station data into a 40-km solar radiation grid along with monthly temperatures and average surface albedo for the US territory [49, 50]. The user selects the PV system parameters including panel rated size, tilt, orientation and DC to AC derate factor [48]. The National Renewable Laboratory suggests a standard system performance ratio of 0.77 to account for the following system losses: inverter and transformer, mismatch, diodes and connections, DC wiring, AC wiring, soiling and system availability [50].

Along with the development of r.sun irradiance calculation model, the Joint Research Centre Institute for Energy and Transportation of the European Commission develop the Photovolta-

2 The Air Mass coefficient quantifies the reduction in the power of light as it passes through the atmosphere and is absorbed by air and dust. 3 Web access: http://pvwatts.nrel.gov/index.php

Chapter 2. State of the art 13

ic Geographical Information System (PVGIS)4 [39], which provides a map-based inventory of solar energy resource and assessment of the electricity generation from photovoltaic systems in Europe, Africa, and South-West Asia [39, 41]. Panel type, mounting position and tracking options, azimuth, slope and estimated system losses are to be selected in the initial page. If required by the user the model also optimized slope for the system location. Furthermore PVGIS is able to estimate losses due to temperature and low irradiance and losses due to angular reflectance effects, which are added to the system losses – by default 14% [41] – to obtain the system performance ratio . The estimation of these losses is made using local ambient temperature, adapting the total system losses to the local conditions instead of applying a general ratio like PVWATTS. The average combined PV system losses for a crystalline silicon panel in Swabia is 26-27% (system performance ratio of 0.73-0.74) [41].

Within Reiner Lemoine Institute, the project “ELUBES – Promoting the commitment of local companies for sustainable awareness in the energy efficiency area” [51] has developed and made available online its own PV energy calculator called Eigen-Verbrauchs-Analyse (EVA)5. ELUBES aims to show the link between renewable energy together with social econ- omies and a sustainable environmentally-friendly development of the region. Within the project active support is given to companies of the region, with tools like EVA, to empower them pushing forward sustainable energy supply independently. The calculation is based on solar radiation data from which hourly production of the photovoltaic system is calculated and standard load profiles to reflect the consumption. The diffuse and direct radiation values are taken from the test reference year data of the DWD6 and using the information of the given azimuth and inclination, the total radiation incident on the PV module is calculated. The hourly energy production of the PV system uses a constant efficiency value of the modules and the inverter. The performance ratio is a constant hourly factor as well.

2.4 Data availability

A relevant part of the thesis was the research of available data sources, as one of the requirements from the Reiner Lemoine Institute. The research was conducted during the first two months of the master thesis parallel with the literature review. The data is originated from a variety of different data sources and most of the data is available at no charge. Some data coould be drawn from public reports or registers, which are as well freely available. Only one dataset used in the study was not free - their use was possible through the thesis’ academic agreement signed with Bavarian Land-survey Office (BVW), which provides with data up to 500€.

Building footprints were considered the basis to start the computation the available rooftops. Access to this dataset, similar to cadastral data, in vector-format was denied by the regional network operator LVN, partner in the Smart Power Flow research project. It was only deliv-

4 http://re.jrc.ec.europa.eu/pvgis/ 5 http://eva.elubes.de/ 6 DWD − Deutscher Wetterdienst: www.dwd.de

Chapter 2. State of the art 14

ered as PDF files. Although the access to shapefile and its metadata was denied, a digitalization and geo-referenziation of the PDF information was possible thanks to the identification of the CAD layers inside the PDF file (section 3.3.1 Data preparation).

After a thourgh study of the possible data available for the work, the methodoly approach was chosen. Table 2.2 gives a summary of all available data for this study considering the needed scale, the ones marked in green were finally used to develop the methodology. Each dataset marked as used (in green) is described in detail when introducing the relevant data in section 3.2

Table 2.2: Available data sources and used data (green) for the master’s thesis

Data type Source Availability Comments Used in this study

Bavarian cartog- BVW Free General map Yes raphy

Administrative BVW Free Federal states, administrative dis- Yes General boards tricts, counties and municipalities. GADM7

cartog- Open street Open- raphy maps StreetMap Free Extension to connect with QGIS Yes

Digital Elevation BVW Free 200 m resolution No Model NAS Free 30m resolution.ASTER WorldEleva- DLR8 Not free tion 12m resolution. TanDEM X Mission

Land-register LVN Not free Building footprint and detail low Yes map distribution networks including household connections. Delivery of shapefile-format was denied. Building

Cadastre ALKIS Not free Expected to be similar to LVN. Data No foot- could be purchased if necessary.

prints

7 GADM - Global Administrative Areas, Boundaries without limits: www.gadm.org/ 8 DLR – Deutsches Zentrum für Luft- und Raumfahrt: http://www.dlr.de/

Chapter 2. State of the art 15

Landsat 7 USGS9/ Free 30 m resolution. Swabia (4 pieces) No images glovis

Orthophotos BVW Not free 20 cm resolution. RGB bands Yes Free access granted for the Remote municipality Jettingen- sensed Scheppach by special agree- imagery ment.

Google Earth Google™ Free High resolution images downloadable Yes images from Google Earth™

LiDAR laser BVW Not free 1m resolution. First-Pulse-points and No points Last- Pulse-points in several classes (ground points, building points). Price: 80€/km2

3-D Buildings BVW Not free LoD1: Existing buildings as a block No Digital model with a flat roof. Price: Surface 0.27€/building LoD2: Realistic 3D representation Model of exist-ing building in LoD2 with standard roof shapes. Under development. Available on-ly for the Neu- Ulm, Ausburg and Ostallgäu-Süd Landkreis. Price: 0.65€/building

Bavarian Energy Bavarian Free Not downloadable but web-based Yes 10 Energy Map Register state accessible: maps Inventory the photovoltaic systems (point layer)

Global radiation and sunshine duration annual mean and monthly values (raster layer, with 200m resolution) Statistics per Bavarian Free Statistics per municipality 2013. Yes Statisti- Municipality Office for Data about: Housing stock, housing, Statistics schools, kindergarten, social help, cal data and Data unemployment, taxes, last election Pro- results... cessing11

9 USGS - United States Geological Survey: http://glovis.usgs.gov/ 10 Bayerische Energie Atlas: http://www.energieatlas.bayern.de/index.html

11 Bayerische Landesamt für Statistik und Datenverarbeitung: https://www.statistik.bayern.de/statistik/

Chapter 2. State of the art 16

2.5 Research approach and research questions

The main objective of this work, as stated in the previous chapter, is to create a model that can estimate the size, potential and location of future photovoltaic systems on rooftops in rural areas. Most methodologies studied in the literature review focus on solar PV potential for urban areas and few consider the typology of small rural communities. More affordable remote-sensed imagery has been used mostly to assess the PV potential on regional scale to calculate total rooftop area and in combination with reduction factors, obtain the mean available surface. Only few authors [16–18], have also attempted to use high-resolution images to quantify the suitable rooftop surface per building and the appropriated location of the panels and none of them have addressed the particularities of rural communities.

With the aim of obtaining a procedure that could build on generally accessible data and that is easily reproducible on a building scale for other rural villages, it was decided to further explore the latest approach based on land-register GIS and high-resolution images for this study. In addition to further contribute on the research of using high-resolution images to directly quantify the suitable rooftop area for rural areas; two different types of images are assessed in this thesis: a) high resolution orthophotos from the Bavarian Land-survey Office and b) Google Earth™ orthophotos. Both data sources are available for the entire study region at acceptable prices in the case of high resolution orthophotos from the Bavarian Land-survey Office, or completely free in the case of Google Earth™ orthophotos.

An pixel-based image analysis is carried out separately for both type of orthophotos to find the suitable area for PV system per building and the results are later compared among each other. The final model would be later used to estimate the new demand profile in distribution networks to optimize the network expansion in Bayern. Derived from the state of research and the objectives of the thesis, the following research questions will be answered and discussed in the progress of the thesis:.

1) How can a model be created to estimate the photovoltaic potential based on a land register map and (a) aerial images or (b) Google Earth images considering the special characteristics of rural environment?

2) What is an appropriate remote sensing approach for identifying accurately the roof-side considering the typical building house in rural Germany?

a. Is pixel-based recognition appropriated to identify roof obstructions? b. Does it accurately identify the adequate roof-side considering the typical building house in rural Germany?

3) How reliable are the results and how much more reliable is one image compared to the other?

4) What are the benefits of the proposed methods in comparison to other methods?

Chapter 3

Methods

This chapter begins with the description of the study area and the input data utilized in this research. The third subsection presents the research design and the steps for identification of buildings rooftops, analysis of the status quo, pixel-based image classification. The final subsection presents the methods for estimating the photovoltaic energy production and prognosis of the most probable location of PV systems. All methodological steps are described thoroughly.

3.1 Description of the study area

The Smart Power Flow project encloses 17 municipalities belonging to the administrative region of Swabia in the federal State of Bavaria. The communities have a population density ranging from 50 inhab/km! to 1167 inhab/km!, being Ebershausen, with 613 inhabitants, the less populated and Mindelheim, with 14205 inhabitants, the most populated (Table B.1 in Appendix B). From the 17 municipalities, eight fall into the category of rural communities, seven into the category of village, and two into suburban areas. Figure 3.1 shows three Smart-Power-Flow examples classified according to its settlement type [6, 14].

Figure 3.1. Example of settlement category rural (left), village (centre), suburban (right). [Source: Google Earth™ images]

Chapter 3. Methods 18

Seventeen municipalities are designated by the regional network operator LVN as particularly critical in terms of the electric grid. The share of photovoltaic energy production is higher than 5% in all cases, reaching in the case of Jegen a PV-share of 279% (Table B.1 in Appen- dix B) with 16.082 MWh produced in 2012 [52]. There are three more municipalities Baisweil, Oberostendorf and Oberrieden that supply more than 100% of the total energy consumption from PV energy production. Worthmentioning is that these three municipalities are classified as rural communities. All municipalities researched by Smart Power Flow, rural municipalities have an average share of photovoltaics in the total municipality energy consumption of 134,5%, while villages and suburban municipalities have a considerable lower share, 26,8% and 16% respectively.

The village of Freihalden (Figure 3.2) in the municipality of Jettingen-Scheppach was chosen as the exemplary community for the purpose of this thesis. Jettingen-Scheppach covers an area of 54.14 km! in the district of Günzburg (latitude 48° 23' 30" N and longitude 10° 27' 44" E) and it is classified as “village” according to its settlement typology. The last statistics of the Bavarian government in 2013 [53] account for 2065 residential buildings in the municipality containing 2872 households. Only a 5% of buildings are blocks with more than 3 households, 17% are two-family homes and the remaining 77% of them single houses [53]. The industrial buildings are located in the outskirts of the village as it is typical for these settlements. This settlement pattern is characteristic of small/medium villages in Central Europe, making Freihalden and Jettingen-Scheppach municipality an excellent example for this master thesis.

Figure 3.2 Village of Freihalden in Jettingen-Scheppach municipality [Source: Google Earth™ images]

The residential buildings have square or rectangular forms covered almost exclusively by pitched roofs [14, 53] with and average living space of 112 m! [53] . Average slope of residential rooftops in Germany varies from 25° to 50° degrees [54, 55]. Figure 3.3 shows the typical shape of buildings in the region. On the other hand, industrial and agricultural buildings present bigger rooftop surfaces and lower inclination angles. Kaltschmitt and Wiese [56] have calculated that more than 50% of German industrial buildings have flat roofs, which are also

Chapter 3. Methods 19

easily identified from space due to its roof material. The rest of non-residential buildings (garages, shelters, covered parking spaces, etc.) have even a higher likelihood of flat rooftops [56]. The Jettingen-Scheppach municipality produced 9615 MWh of energy in 2012 from 9,7 MWp of nominal power installed on its rooftops [52], supplying already 22% of the total electrical consumption of municipality [52]. The study site was selected to ensure the representa- bility of a typical average sized village in Swabia that it is already producing considerable amount of energy from renewable sources, and therefore, shortcomings problems in the network distribution are expected.

Figure 3.3. Residential houses with different roof inclination angles in Jettingen municipality [Source: Wikicommons under free licence]

3.2 Input data

The following subsections describe in detail each dataset used for the creation and validation of the prognosis model, namely: Bavarian general cartography, Land register map with building footprints, Orthophotos from BVW, Google Earth orthophotos and Bavarian municipal statistics.

3.2.1 General cartography

Bavarian Digital Topographic map 1:500 000

The Digitale Topographische Karte is an overview raster-map of Bavaria, which is carto- graphically generalized to the scale 1:500000. The raster dataset has a resolution of 200 pixels/cm and it is delivered under the reference coordinate system EPSG: 31468 DHDN/3- degree Gauss-Kruger zone 4. EPSG 31468 is a projected coordinated system for large and medium scale topographic mapping engineering survey and cadastral survey used in the for- mer West Germany states between 10°30'E and 13°30'E (Bayern, Berlin, Niedersachsen, Schleswig-Holstein). The babarian digital topographic map is freely available to download at the Bavarian Land-survey Office webpage.

Chapter 3. Methods 20

Administrative borders

The Verwaltungsgrenze provided as vector data all administrative boundaries at state level. In a shapefile format, the administrative boundaries are plotted at 1:25000 scale corresponding to the object type group "Administrative Entity" of the ATKIS object type catalogue. The vector dataset is also under the reference coordinate system DHDN/3-degree Gauss-Kruger zone 4 (EPSG: 31468). The Administrative border dataset include:

(a) German federal states (6.8 MB), (b) Bavarian administrative districts (7.7 MB), (c) Bavarian counties (2.8 MB) and (d) Bavarian municipalities (28.5 MB).

It is freely available to download at the Bavarian Land-survey Office webpage. Last updated on 06-08-2013.

Open street maps

OpenStreetMap12 (OSM) is a non-profit collaborative project to create a free editable map of the world. It provides free geospatial data to use and share. OSM project has gained populari- ty because in many countries no free geodata such as digital roadmaps are available. In recent years OSM has also increase its exporting capabilities and it is currently accessible through most GIS software packages. The GIS application QGIS has an OpenStreetMap Plugin that allows the direct viewing of maps. It adds support for OSM raw vector data, bringing it in as a layer either from .osm XML file. OSM data can be exported as entire layers shapefiles with tags concatenated as a single attribute and has detailed and accurated maps for the study area.

3.2.2 Land register map

The Land register map is a floor map that provides geodata about streets, parcels and buildings for each municipality. Based on the ALKIS data German Federal Cadastre, this land register map provided by the regional network operator LVN also includes the medium- voltage and low-voltage distribution lines, as well as each house connection.

Access to original vector data was denied and land register maps were only delivered as PDF with layers. With help from a software, was possible to transform the PDF files into a CAD group layer, which in turn was transformed into separate shapefiles (see section 3.3.1 for further details). The resulting Land register map was a vector dataset containing the following cadastral features:

• Municipal parcel map (polyline layer) • Parcel number (point layer)

12 OpenStreetMaps www.openstreetmap.org

Chapter 3. Methods 21

• Street names (point layer) • Building footprints (polygon layer) • House numbers (point layer) • Household connection (point layer) • Medium-voltage distribution network (polyline layer) • Low-voltage distribution network (polyline layer)

Due to the fact that only pdf files were delivered, all cadastral information existing in the original vector data is missing. Unfortunately, information about number of storey per building, roof typology, actual building usage, number of households, number of persons, legal ownership, and buildings under protection, settlement category, and cadastral reference is not accessible for this study as it was desirable at the beginning of the thesis for the prognosis step.

3.2.3 Official orthophotos from BVW

An orthophoto or orthoimage is an aerial photograph geometrically corrected, such that the scale is uniform. Unlike an normal photographs, orthophotos are accurate representation of the Earth’s surface can be used to measure true distances, having been adjusted for topographic relief, lens distortion, and camera tilt. The German Federal Agency for Cartography and Geodesy (BKG) is responsible for the creation and production of digital orthophotos in Germany. The aerial images are recorded from a plane to achieve very high ground resolution and later orthorectified by the agency using their own digital elevation model [57].

The dataset consists of four georeferrenced images in uncompressed TIFF-format (541MB each image), a reference coordinate system file, worldfile (wld), and a textfile with information for each image. The orthophotos are recordered in three bands Red-Green-Blue (RGB) with 24 bit pixel depth and a ground resolution of 0.2 m. The coordinate system is DHDN/3-degree Gauss-Kruger zone 4 (EPSG: 31468). The orthophotos covers Jettingen- Scheppach municipality with the following boundaries: north 4389342.00, south 4382980.00, 365322.00E and 5360560.00W. The extend of the orthophotos is show in Figure 3.4. They have two different acquisition dates with one month difference, 19th August 2012 (upper left and upper right images) and 27th July 2012 (lower left and lower right images).

As part of a high school agreement signed for the master thesis, digital orthophotos from the Bavarian Land-survey Office (BVW) are available free of charge. The original price of the orthophotos is 240 €

Chapter 3. Methods 22

Figure 3.4 Extent of the four orthophotos from BVW covering Jettingen-Scheppach municipality.

3.2.4 Google Earth orthophotos

Google Earth orthophotos come from satellite imagery. Google Earth™ and Google Maps share the same database of images. Google has contracts with both government and private satellite owners to receive images and maps so that every part of the world can be mapped at some level of detail. Google acquires the best imagery available, most of which is approximately one to three years old. It uses higher resolution images for cities and more heavily populated places. Rural areas tend to have less detail available in the imagery. For the study area, Google Earth™ receives the satellite images from DigitalGlobe, GeoBasis-DE/BKG and GeoContent at a course scale and from DigitalGlobe at detail scale. DigitalGlobe is operates the most sophisticated constellation of high-resolution commercial earth imaging satellites. The company owns the satellites IKONOS, QuickBird, WorldView-1, GeoEye-1 and WorldView-2. Unfortunately, Google does not released detailed information about from which satellite the image was taken, but rather just the date of the image.

The raster data downloaded for the Jettingen-Scheppach municipality is also extracted in three bands Red-Green-Blue (RGB) with a sub-half meter ground resolution (approx. 0.41- 0.46 m) and 16 bit pixel depth. The dataset consists of series of downloaded images composed together and later georeferrenced (see section 3.3.1 for further details). The resulting mosaic

Chapter 3. Methods 23

is a GeoTIFF-image under the coordinate WGS8413 (EPSG: 4326) with a data size of 707 MB. The acquisition date for the images of the Jettingen-Scheppach municipality is the 18th June 2013.

3.2.5 Bavarian photovoltaic database

The Bavarian Energy Atlas [52] offers the possibility to download selected data directly from their website about the themes: biomass, geothermal, solar, wind, hydropower and biogas. Several topics are available in each thematic area. For solar energy, information about yearly solar irradiation, yearly sunshine hours and PV plants installed in Bavaria is accessible. The Bavarian Energy Atlas was used to obtain a georeferenced list of all PV plants in the 17 municipalities involved in the Smart Power Flow project. The output is a CVS file encoded in “UTF-8” that can be opened in Excel or OpenOffice containing the following information for each PV system: Projected coordinate system (EPSG code); Location (X, Y coordinates); Name of the municipality; Municipal unit code; Power (kWp); Energy production in 2012 (kWh); Full load hours per year −calculated− (h); Installation year; Ground-mounted system (yes/no). 4318 rooftop PV systems are reported as installed in the project´s municipalities, as of 31.12.2012.

3.2.6 Statistics per municipality

The Bavarian State Office for Statistics and Data Processing offers statistical data for each municipality in Bavaria. “STATISTIK kommunal” are generated every year from the statistical database; they contain 31 tables and 18 graphs for each community providing up to 2200 statistics, delivering meaningful community profiles. Time series over several years make it possible to identify trends and use them for future decisions developments. It contains data about: population, taxes, municipal finances, past elections, industry, agricultural production, number and type of buildings, new buildings, public facilities and traffic. The PDF version of statistics is available free of charge.

3.3 Research design

Once the available data sources were identified a number of tasks were outlined in order to estimate total rooftop photovoltaic potential and prognosis for the exemplary village. It was first necessary to prepare the input data for the model, i.e. digitalize the land register map PDF and acquire the Google Earth orthophotos. With the digitalized building footprints, both orthophotos were clipped to target only the rooftops.

A pixel-based image analysis is performed on each clipped image to identify suitable rooftop areas, roof obstructions and shadows. The output of the analysis is converted to vector data and the suitable area is assigned to each building. Average slope and inclination are spatially

13 World Geodetic System 1984.

Chapter 3. Methods 24

joined to each rooftop. The results of the image analysis together with the slope, inclination and location of each building are giving as input to RLI´s PV calculator tool to obtain the annual energy production of each building. Finally, prognosis step predicts the PV expansion pathway based on each rooftop PV potential and the scenarios of the German Energy Agency.

Figure 3.5 outlines the research design and the rest of the chapter discusses each step in detail.

Aerial Images (.img) Building footprints (.shp) from a) Google Earth images Land-register map b) Digital aerial photos

Isolating building rooftops

Pixel-based image classification

Validation Computation of suitable rooftop area for PV per building

EVA Photovoltaic calculator Tilt and azimuth

PV energy production

Prognosis of the most probable German Energy Agency location of PV systems (DENA) scenarios

Figure 3.5 Flowchart for calculating PV potential for this study

Chapter 3. Methods 25

3.3.1 Data preparation

Data preparation required a significant amount of time due to the special characteristics of this study: Land register map need to be completely digitalized and Google orthophotos shall be downloaded and georeferenced as a prerequisite for the next methodological steps. Due to these reasons, the digitalization process is the step that requires more time.

3.3.1.1 Land register map digitalization

As mention in section 3.2.2, land register maps where only provided in PDF format. Normal procedure in these cases is to import the PDF map as a raster into the GIS software and digitalize them manually. Only in Freihalden there are more than 400 buildings, which make manual digitalization not viable. There is also no direct procedure to convert PDF files into vector data. A roundabout was created for this special situation.

The software Aide Converter [58]14 was used to transform the PDF into DXF-format. DXF (Drawing Interchange Format) is the native format for CAD packages but can be read also in GIS software. Each PDF is converted into a “DXF Group layer” consisting of a mix of points, polylines, polygons and annotation. The DXF is imported into GIS software where each layer is identified and georeferenced. As regular CAD layers, DXF has its coordinates stored in local coordinate system and no spatial reference. The layers are georeferenced by means of an ASCII pointfile where the local coordinates (origin) and the projected coordinates (target) are stored. The projected reference coordinate system is DHDN/3-degree Gauss-Kruger zone 4, likewise the BVW orthophotos.

At this point the land register map is already a vector-data but no distinction between map elements (e.g. building vs road). In order to have a comprehensive land register map, each urban structure was classified by the value “Colour” stored as hidden metadata from the PDF and then exported into a separate layer. Table 3.1 shows the correspondence of DXF feau- tures and urban structures based on the colour value.

Special treatment was given to the household connections points, a very important feature to later connect the results to the Smart Power Flow database. Household connections points from the PDF were transformed into small round polylines with exactly the same attributes as the building layer, i.e. embebed into the building layer. Therefore it was not possible to extract them with the same procedure as the rest of the features of the DXF Group layer. The procedure to extract the Household connection points was as follow:

1. Transform the Building polyline shapefile into polygon shapefile 2. Run a spatial query to identify small polygons laying within building´s polygons 3. Export the selected features to a new shapefile layer 4. Convert the shapefile to point layer

14 Aide PDF to DWG Converter is a non-free software. For the purpose of this masther thesis. The Free trial version was used, which allows for the convertion of up to 75 PDF files.

Chapter 3. Methods 26

During the digitalization process only the annotation features cannot be successfully transformed. The Aide software package is able to recognize the streets and house numbers as annotations but unfortunately each letter is recorded as an independent feature, distorting the order. Furthermore, German characters (ß, ö, ü, ä) are not recognized making the general map uncomfortable to read. This problem was solved by exporting the street names form OpenStreetMap and replacing the corresponding feature.

Table 3.1 Correspondence of DXF features with Land register map´s elements

CAD layer Colour ID Cadastral element Output Shapefile

Polygon 252 Parcel map Village_Flurkarte

Polyline 251 Parcel map (other features) Village_Flurkarte2

Polyline -2 Buildings Village_Gebaude

Village_Hausanschluss

Polyline 1 Medium-voltage network Village_MS_netzlinie

Polyline 1, 3, 4, 5, 6, 32 Low-voltage network Village_NS_netzlinie

Polygon 7 Transformer stations Village_Trafos

Annotation Parcel number, House number, Village_Hausnummer

Street name, Cadastral codes Village_Strassename

3.3.1.2 Download of Google orthophotos

Google allows downloading their imagery directly from Google Earth™. For public users, the resolution of the downloaded image matches the screen resolution at the moment of downloading. This means that the higher the resolution, the smaller the area visible in the screen and thus the smaller the downloaded area. Each image needs to be georeferenced and later merge these into single big mosaic to obtain a uniform image.

There are several procedures to overcome this manual time-consuming task. With a Google Earth Pro™ license15, users can download high-resolution imagery at desired scale; images can be exported up to 4,800 pixels wide. Google Earth Pro uses the same imagery database as the free Google Earth and Google Maps and due to the expensive licence cost several software packages have been developed to offer a more affordable solution.

15 Available under: http://www.google.com/enterprise/mapsearth/products/earthpro.html One-year licence is 319€/user (as of 23.07.214)

Chapter 3. Methods 27

Figure 3.6. PDF land-register map provided by the regional network operator LVN (Above) and land register map after digitalization (Below).

Chapter 3. Methods 28

The shareware Universal Map Downloader16 or the freewares El-Shayal Smart GIS [59], an Arabian GIS software, and SAS.Planet [60], a Russian application, are softwares that help to get small tile images from Google Maps, Yahoo maps, Bing maps, OpenStreetMap, etc. and automatically combine them into one big georeferenced image. The licence allows for non- commercial or personal use only.

For this master thesis, SAS.Planet was used to obtain Google Earth georeferenced orthophotos. In order to obtain the raster image it is necessary to first define the area of interest via rectangular selection, polygonal selection or by coordinates. The software calculates the number of tiles necessary for the area of interest at the given resolution 210 tiles for Frei- halden and store then into the program cache. Finally, the image can be created and saved together with the georeferencing file. It is possible to choose between seven output formats, five map projection, six different georeferencing files and tens of overlaying layers, making SAS.Planet extremely dynamic. The image is afterwards imported into the GIS software and saved in GEOTIFF-format.

3.3.2 Isolating building rooftops

The extraction of the image of each building passes through the overlapping of the cadastral metadata to the orthophoto and using the building footprints as a mask to clip the orthophotos. The result is a clipped image composed by small subsets, same number of buildings as in the land register map. Directly integrating the building footprint shapefile limits the subsequent pixel-based image analysis within the rooftop surface and thus excludes data from out- side the building rooftop. The classification focus therefore only in spectral characteristics of the building’s rooftops, leading to a reduction of pixel´s misclassification and method error.

In the study area, building footprint shapefile do not match completely building rooftops on the images (Figure 3.7). Although this is a known problem when working with different GIS data sources [18], it still needs to be corrected. This is achieved by slightly moving the image to centre each building rooftop in the building footprint and giving each building footprint a one meter buffer to cover the entire building rooftop. The procedure to adjust the building footprints to the building rooftops goes as follow:

1. Re-georeferenced the image to centre the rooftops within the building footprints 2. Buffer and dissolve building footprint to cover the entire rooftop area. 3. Clip the raster image to extract a subset image of the building rooftops, using as mask, the shapefile obtained in the previous step.

3.3.3 Analysis of status quo

As already mentioned, Jettingen-Scheppach municipality has already 9,7 MWp installed PV nominal power, which represent 22% of the total municipal electrical consumption [52]. The aim of this step is to find where and how much PV installed capacity has Freihalden.

16 Available under: http://www.allmapsoft.com Shareware licence is 45€/user (as of 23.07.214)

Chapter 3. Methods 29

Figure 3.7 Original orthophoto and building footprints (left). Orthophoto and building footprints after correction (right).

Therefore for the status quo, the CSV file from the Bavarian Energy Atlas and LVN database were imported in the QGIS project with the tool Add Delimited Text Layer. The georeferenced point shape file is overlapped with the building footprint shapefile and both layers in- tersected. The output is the location of all the roof which already have installed PV systems, with the installed capacity (kWp) and the energy production for each rooftop (Table 4.1 in chapter 4).

3.3.4 Pixel-based classification

The objective of this step is to perform a image classification using the pixel spectral properties to identify the suitable area for PV on rooftops (Figure 3.8). Particularly, the purpose is to classify: a) shadow areas b) obstructions c) suitable areas d) non-suitable areas

Suitable areas are consider to be parts of the roof south-oriented with no obstructions (chimneys, windows, antenns, etc) and therefore adequate for PV modules, while non-suitable areas do not have obstructions but are north-oriented. South-oriented and north-oriented roof sides differ from each other on the brightness of its pixels, meaning that they present different spectral properties.

To identify the suitable areas a supervised classification is conducted on the subset image of the building rooftops. Supervised image classification is an image processing technique that allows for the identification of materials in an image, according to their spectral signatures. Supervised classifications require the user to select training sites for each target and assign them a ID code (i.e. identifier). Training samples are created to represent classes in a supervised classification. After training sample collection, spectral characteristics of roof targets are

Chapter 3. Methods 30

Figure 3.8 Detail of the obstruction, shadows, suitable and non-suitable areas of one household in Freihalden. calculated by the algorithms considering the values of each pixel belonging to the training samples that have the same class ID. Therefore, the classification algorithm classifies the whole image by calculating the spectral characteristics of each pixel, and comparing them to the defined classes. The result of the classification process is a raster, where pixel values correspond to class IDs.

The following steps were performed for image classification process 1. Select number of classes

2. Creation of Areas of interest - Collect training sites - Evaluating training sites (histograms) - Editing classes

3. Compute Supervised classification - Create signature file - Run supervised classification

4. Post-classification processing - Majority Filter - Boundary smoothing - Raster vectorization

3.3.4.1 Select number of classes

Pixel values were first anylized to get an idea of the number of needed classes. Redish and greysh rooftops present spectral differences between suitable and non-suitable sides of the roof. Unsupervised classification and spectral signatures (Figure 3.9) confirmed that not all the roofs could be processed together in the same classification as it would be desirable. Non-

Chapter 3. Methods 31

suitable red roof-areas and suitable grey roof-areas have similar signatures leading to misclassification of the pixels (see brown and dark blue lines in Figure 3.9).

Seeking to solve this problem, the buildings were classified into three diferrent categories: red, grey and black roofs. Building footprints were visually classified according to roof category and the orthophotos were re-clipped to obtain three different subset images each one containing only one category of roof type. The resulting signatures are shown in Figure 3.10.

no-suitable red obsturction shadow suitable grey suitable red no-suitable red

Figure 3.9 Signature plots

Figure 3.10 New signature plots after building categorization: red roofs (left), grey roofs (center) and black roofs (right)

3.3.4.2 Creation of Areas of interest

The number of target classes targeted in this study are four, in accordance with the roof features: suitable areas, non-suitables areas, roof obstructions and shadows. Supervised classification requires to select several training sample for each class. Training sites are polygons

Chapter 3. Methods 32

defined over the image that overlay pixels belonging to the same class identified with an ID. For each class 20 to 25 training samples where created assignating them the corresponding class ID (Table 3.2).

Table 3.2 Class types

Classes Class ID Colour

Suitable areas 1 Green

Non-suitable areas 2 Red

Obstructions 3 Pink

Shadows 4 Black

The appropriateness of the training sites was check studying their separability and distribution.The histograms and the scatter plots allows the user to compare the distribution of mul- tiple training samples. When training sites represent different classes, neither histograms nor scatter plots should not overlap each other. Figure 3.11 shows the histograms for the training sites of the grey roof image: black training samples representing shadows, red representing non suitable areas, green representing suitable areas and pink representing obstructions. After this evaluation some training samples were remove and new ones created until a satisfied training sample set was created.

Figure 3.11 Histograms for band 1, band 2 and band 3 of grey roof image.

3.3.4.3 Compute supervised classification

The final training sample set is stored in a signature file. The signature file records the spectrum signatures of different classes across a series of bands. For each class, the signature contains means and covariances calculated from its training sample; i.e. the spectral characteris-

Chapter 3. Methods 33

tics of classes. Each material has a unique signature, as shown in Figure 3.10, and it is used for the image classification.

Minimum distance method is applied as classification algorithm. The classification algorithm places each unknown pixel in the class closest to the mean vector in the band space. The distance is defined as an index of similarity so the minimum distance is identical to the maximum similarity. Figure 3.12 shows the result of the supervised classification using the minimum distance algorithm on one household from the orthophoto of the Bavarian Land-register Office (BVW).

Figure 3.12 Original orthophoto clipped by the building footprint (left) and classified output after supervised image classification [black=shadow, red=non suitable, green=suitable, pink=obstruction] (right)

3.3.4.4 Post-processing classification

Post-classification processing refers to the process of removing the noise and improving the quality of the classified output. Image post-processing helps to remove misclassified isolated pixels or isolated regions of pixels that may exist. Small misclassficated pixels give the output a "noisy" or speckled effect (left image of Figure 3.13). Post-processing is conducted in three steps: majority filtering, boundary smoothing and raster vectorization. The result is a more consistent image classification output.

1. Filtering the classified image. The tool “majority filter” removes the isolated pixels or noise from the classified output.

2. Smoothing class boundaries of the classified output. The tool “boundary smoothing” smooths the class boundaries and clumps the classes.

3. Vectorazing the classified image. This step converts each the raster classified image into a polygon shape layer to apply the following steps of the methodology. The “raster to vector” tool from

Chapter 3. Methods 34

GRASS, accessible from the QGIS processing toolbox extension, was found to give the best results as it simultaniusly generalize the vector output layer.

Figure 3.13 Building before (left) and after applying post-processing steps (right).

3.3.5 Computation of total suitable area

To computate the real rooftop area, the projection of image has to be considered. From the building footprint the real rooftop area can be easily decudted. Average roof inclination (tilt) of residential buildings in Germany varies from 25° to 50° degrees [54]. For the simplicity of this study an average tilt angle of = 35º was chosen (Figure 3.14), following the example of two other german publications [54, 55]. The roof area of one side of the roof is calculated then as follows:

Figure 3.14 Determination of the rooftop area considering the roof tilt angle = 35º

Although the supervised classification identify the suitable areas on rooftop, not the 100% of this area can be completely covered by PV modules. The standard size of the modules and the existing obstructions inside the suitable areas (obstructions are not computated as suita-

Chapter 3. Methods 35

ble areas but the PV modules need to be place around them) will demand a specific disposition of the PV modules in each rooftop leaving some small parts of the roof uncovered. A reduction coefficient of =0,9 [9, 13] is applied to account for disposition of PV modules on suitable areas. Therefore the suitable area is calculated as:

Flat roofs receive a similar treatment. In this case, the total roof area is equal to the one measured from the image. But differing from pitch roofs where the PV modules are laid on top of the roof, in flat roofs the PV modules have to be mounted in rows and are elevated for the optimum utilization of solar radiation and the distance between rows is defined to avoid shading between the modules rows. In Germany, a rooftop ratio of about 1/3 has been proven for PV module inclination angle of 30° [61]. Therefore the suitable area for flat roofs is calculated as:

To extract the suitable areas of each rooftop, only suitable areas (green) with more than 10 m2 are selected using the SQL-expression selection tool and saved as a separated shapefile. Areas with less than 10 m2 are considered not to be costeffective for PV systems. For each suitable area bigger than 10 m2 four extra columns are added to the layer attribute table:

• the tilt angle ( = 35º for pitch roofs and = 30º for flat roofs), • the azimuth (orientation) calculated runing the Minimum Bounding Geometry tool • X coordinates of the surface centroid. • Y coordinates of the surface centroid.

These four columns of information plus suitable area in m2 are the input data for EVA in the next step.

3.3.6 PV energy production

The potential annual energy production is calculated with EVA [51], the RLI’s PV calculator, accounting for geographical location, the inclination, the azimuth, the kilowatt peak (kWp) and the solar irradiation on the tilted surface for each rooftop.

The geographical location in X and Y coordinates, the tilt and the azimuth are taken from the attribute table of the previous step. The kWp is derived from the amount of suitable area for PV production: 1 kWp requires about 10 m! roof area [51]. The solar irradiation on the tilted surface is determined by the calculator choosing the corresponding weather region. Table 3.3 shows the input data need to calculate the PV yearly energy production with EVA. Each single suitable area is processed by the calculator and an extra column with the energy production is added to the output table (last row Table 3.3).

Chapter 3. Methods 36

Table 3.3 Input and output data in EVA PV Calculator

Attribute Unit Freihalden example

Feature ID - 146

X coordinate in WGS 84 10.496059

Y coordinate in WGS 84 48.384980

Polygon area m2 66.066

Suitable area m2 72.673

Kilowatt peak kWp 7.26

Tilt degrees 35

Azimuth south degrees 14.01

Energy output kWh 7081.26489041

The output from the PV Calculator is imported to the GIS software and spatially joined with the initial building footprints representing each household, since rooftops with complex shape may have two or three suitable areas with different orientations on them. The result is saved as a separate shapefile and CSV table.

3.3.7 Validation

The methodology bases its results on how well the supervised image classification is implemented. Therefore the performance of the methodology is analysed with accuracy assessments. Accuracy assessments are performed on classified images to determine how well the classification process accomplished the task by means of an error matrix. The error matrix is a table of values that compares the value assigned during the classification process to the true value that pixel has. 210 random points were taking with the following distribution: 65 points for reddish rooftop image, 65 points for the greyish rooftop image and 30 points for the black roofs. The reference points were manual assigned on screen from the high resolution orthophotos. On this basis the error matrix the overall accuracy, the user and producer accuracy are calculated.

Tables C.1-C.6 in Appendix C gather the accurancy assessments of the six supervised classification conducted. It was found that the classification decreases its accuracy depending on the material of the roofs. In this way, reed roofs have the highest accuracy. On the other hand black and very dark roofs delivered worse result. Overall the Bavarian land-register orthopho- to presented a higher accuracy in all roof types (red roofs: 91%; grey roofs: 87%; black roofs: 76%) compared to the Google Earth™ image (red roofs: 88%; grey roofs: 82%; black roofs: 71%). As a consequence, the Bavarian land-register orthophoto has a weighted average accuracy of 89% while the Google Earth™ image has a weighted average accuracy of 85%.

Chapter 3. Methods 37

3.3.8 Prognosis of the most probable locations for PV systems

The German Energy Agency [5] generated two scenarios for the expansion of photovoltaic and wind energy sources in Germany at municipal level with a horizon year of 2030. The scenarios are made to fulfill german national goals (scenario 1) and german “bundesländer” goals (scenario 2) concerning the implantation of renewable energy sources in the electricity market. Each municipality is assigned to a category depending on the population density and the already existing solar and wind power plants in relation to the german average in 2010: 54,6 kW/km 2 for photovoltaics and 94,8 KW/km2 for wind energy. There are 11 categories, six describing rural municipalities (A), two describing suburban municipalities (B) and three describing intensities of urban settlements (C).

The municipalty of Jettingen-Scheppach falls in German Energy Agency class A2 (Figure 3.15). According to its class assignation and its municipal area, Jettingen-Scheppach should have installed 587 kWp of photovoltaic energy in 2010, if we consider the PV-factor for 2010. In fact, 2.906 kWp of photovoltaic energy have been already installed in the municipality by 2010 and 9.730 kWp by 2013 [52]. This would mean that the municipality had already exceed its expected PV development fro 2030 in the past year (Table 3.4). The reason behind this bias is that, Jettingen-Scheppach has a population density of 125 inhab/ km2, very close to the border between classes A and classes B: 150 inhab/ km2. Looking at the installed PV nominal power by 2010 it should fall in the class B (suburban), which is also consistent with its population density. For the sake of this study, Jettingen-Scheppach will be assigned the the class B1 by agreement with Smart-Power-Flow researchers. Consequently a PV factor of 4,9 is used to create the PV prognosis expansion pathway.

Table 3.4 Jettingen-Scheppach in Dena network classes

Installed Nominal Nominal Popula- PV PV power (kWp) PV power (kWp) Dena Popula- Area tion nominal factor in 2010 factor in 2030 Netz tion km2 density power 2010 Bundeslän- 2030 Bundeslän- class hab/km! kWp [52] derszenario [52] derszenario 2010 [52] [52]

6769 54,14 125,0 2906 A2 0,2 587,96 2,7 7937,47

B1 1 2939,80 4,9 14405,03

For the prognosis, first the PV technical potential is calculated and subsequently the PV expansion pathway is inferred. PV technical potential represents which building have the higher probability based on how profitable the installation of a PV system in the roof will be. This concept is illustrated by the specific yield.

Chapter 3. Methods 38

Figure 3.15 German Energy Agency classes [5]

Chapter 3. Methods 39

Based on the specific yield, a ranking is made starting with the most profitable building for photovoltaic systems. The output of this ranking can be seen in section 4.3 PV technical potential.

The PV factor 4,9 assigned to Jettingen-Scheppach municipality implies that, in order to fulfil the Bavarian renewable energy goals, the village of Freihalden should boost its installed nominal power from 2.906 kWp in 2010 to 14.405 kWp in 2030. Considering that the municipality had already reached 9.730 kWp by 2012, 4.675 kWp are left to be installed by 2030. Freihal- den enclose 412 buildings from the 2.065 that comprise the whole municipality. This represents 19,9%. Thus it is entitle to intall 930 kWp by 2030. This value determinates how many of the buildings with the highest PV potential are forecasted to have PV systems installed in the next 17 years.

Due to the lack of futher population data on household bases, a linear growth is chosen for the expasion pathway. As a result approximately 54,7 kWp per year will be installed in Frei- halden until 2030. A installation year is assigned to each building with higher potential until the 930 kWp are achived, regarding the 54,7 kWp per year.

Chapter 4

Results

The following chapter shows the results of modelling efforts described in the methodology as well as a validation of the outputs. As of the current status quo, Freihalden was found to have 43 buildings with already PV systems in its rooftops. The suitable area computation compares the results of the Bavarian land-register orthophoto and Google Earth™ orthophoto. It was found that 82% and 78%, respectably, of the buildings in the community have adecuate areas with more than 10 m2 suitable for PV. The total PV technical potential of Freihalden is between 2740 kWp and 3170 kWp. In every result subsection the maps created for the Smart Power Flow project are shown, as well as being display in DIN A3 format in Appendix E. Lastly Appendix D presents the final database with the master thesis results used for the creation of the maps.

4.1 Status quo

Among the 412 buildings composing Freihalden, 43 buildings where found to have already PV systems installed on its rooftops (Figure 4.1). A total of 571 kWp of photovoltaic sources are installed, which produced 461,6 MW of electricity in 2012. This represents the 10.4% of all building in the village and accounts already for the 95% of the electricity comsumption private households (Table B.1 in Appendix B).

The smallest PV systems installed has 2 kWp of nominal power and it also represents the oldest one in the village, from 2002. The largest PV systems reaches the nominal power of 64 kWp and has been installed shortly last year 2013 (Table 4.1). It is located on a industrial building (most upper-left marked building in Figure 4.1) but no energy production data is available yet. The average full load hours per year in the village is 987,8 hours, machting the average full load hours of the municipality according to the Bavarian Energy Atlas Register [52].

Chapter 4. Results 41

Figure 4.1 Houses with installed rooftop PV systems (purple) in Freihalden community.

Table 4.1 Installed PV systems in Freihalden

Nominal Full load Location PV system Electricity pro- Installation Power hours per X coordinate Y coordinate duction (kWh) year (kWp) year (h) 4388602 5362303 64 - - 2013 4388381 5361793 32 31697 1002 2010 4388595 5361782 30 16102 543 2012

4388465 5361903 26 25614 981 2010 4388506 5361990 26 27360 1072 2008 4388490 5361782 25 24088 1042 2011 4388368 5361763 25 22613 910 2010 4388553 5361882 25 28599 1135 2009 4388591 5362101 20 16770 1002 2010 4388376 5361783 20 20620 1053 2009 4388352 5361676 18 - - 2012 4388451 5361560 17 17700 1035 2011 4388447 5361734 17 17880 1062 2007

Chapter 4. Results 42

4388333 5361592 15 11255 863 2011 4388727 5361635 15 17680 1142 2010 4388697 5361782 15 16022 1065 2008 4388553 5361882 12 11318 962 2010 4388694 5361685 12 12415 1043 2008 4388873 5361969 12 13835 1130 2008 4388855 5361823 12 12807 1046 2007 4389021 5361809 11 10824 1001 2010 4388523 5362098 11 10929 979 2004 4388583 5361925 9 8438 983 2007 4388698 5361644 8 8987 1080 2007 4388855 5361823 8 8893 1068 2007 4388523 5362098 8 8196 979 2004 4388613 5362176 7 - - 2012 4388490 5361782 6 5420 1042 2011 4388376 5361783 6 6403 1053 2009 4388719 5361674 6 4862 844 2009 4388546 5362027 6 5389 936 2003 4388421 5361600 6 5715 992 2003 4388663 5362128 5 71 33 2012 4388890 5361771 5 5020 919 2011 4388988 5361835 5 5172 1071 2009 4388795 5361964 5 4560 974 2009 4388352 5361621 5 5075 940 2003 4388610 5362206 4 2584 591 2011 4388795 5361964 4 3507 974 2009 4388719 5361674 3 2431 844 2009 4388546 5362027 3 3494 1329 2005 4388954 5361824 2 1276 851 2002

TOTALS 571 kWp 461,629 kWh

4.2 Suitable area computation

The results of the supervised image classification is shown in Figure 4.2 and Figure 4.4. Red, grey and black roofs where processed separately for each data set and the classification output of the three images is merged in one shapefile, as displayed in the figures. Figure 4.3 and Figure 4.5 illustrate the PV suitable areas with more than 10 m2.

Chapter 4. Results 43

4.2.1 Bavarian land-register orthophoto

The results from the supervised classification on the official orthophotos from the Bavarian land-register office are presented in Figure 4.2. Based on the classification, the buildings in Freihalden were found to have 39,5% of their rooftops suitable for PV systems, 40,7% non suitable for PV systems, 1,3% covered with obstructions and 13% shadowed (Table 4.2).

Figure 4.3 shows PV suitable areas with more than 10 m2 deduced from the supervised classification output. From the 412 buildings in the community, 338 are considered suitable for PV and 74 not suitable for PV, meaning that 18% of the buildings have no photovoltaic potential. The computation of the error matrix in the accuracy assessment displays a 91% accuracy in red roofs, 87% accuracy in grey roofs and a lower 76% accuracy in black roofs (Tables 1-3 in Appendix C).

Figure 4.2 Supervised image classification based on the Bavarian land-register orthophoto

Table 4.2 Classification statisitics

Class PixelSum Percentage % Area [square meter] 1 816269 0.39563942 32649.8453 2 841491 0.40786433 33658.7524 3 27737 0.01344391 1109.41163 4 377667 0.18305234 17105.5978

Chapter 4. Results 44

Figure 4.3 Freihalden photovoltaic suitability map based on Bavarian land-register orthophoto

4.2.2 Google Earth images

The results from the supervised classification on the Google Earth™ image are presented in Figure 4.4. Based on this classification, the buildings in Freihalden were found to have 40,1% of their rooftops suitable for PV systems, 46,6% non suitable for PV systems, 2,4% covered with obstructions and 10,8% shadowed (Table 4.3). The numbers differ slightly from the ones obtained in the classification of the Bavarian land-register orthophoto; they are discuss in the following chapter.

Figure 4.5 shows PV suitable areas with more than 10 m2 deduced from the supervised classification output on the Google Earth™ orthophoto. The output is very similar to the results obtained from the Bavarian land-register orthophoto. 314 buildings are considered suitable for PV and 98 not suitable for PV, representing 23% of the buildings have no photovoltaic potential. Small side-building like garages or shed, where the suitable does not reach the 10 m2 in this classification are the reason of this decrecement. The accuracy assessment displays a 88% accuracy in red roofs, 82% accuracy in grey roofs and low 71% accuracy in black roofs (Tables 4-6 in Appendix C).

Chapter 4. Results 45

Figure 4.4 Supervised image classification based on Google Earth™ orthophoto

Table 4.3 Classification statistics

Class PixelSum Percentage % Area [square meter] 1 905661 0.401199886 29368.0387 2 1052178 0.466105633 32824.8181 3 55016 0.024371606 1247.90368 4 244526 0.108322875 7566.21591

Chapter 4. Results 46

Figure 4.5 Freihalden photovoltaic suitability map based on Google Earth™ orthophoto

4.3 PV technical potential

Based on the specific yield, a ranking is made starting with the most profitable building for photovoltaic systems. In order to better compared the results from the classification outputs, the PV technical potential is determined for both images.

4.3.1 Bavarian land-register orthophoto

The total PV technical potential of Freihalden reaches the 3170 kWp. The building range from 1 potential kWp to 49,5 kWp and the specific yield varies from 980,1 to 761,1 kWh/kWp. The PV potentiality map is display in Figure 4.6. All buildings with the highest potential are south oriented (see Figure 4.6). In addition, 78 of the buildings where found to have no PV potential. The final database is displayed in Table D.1 in Appendix D.

Chapter 4. Results 47

Figure 4.6 PV technical potential map based on Bavarian land-register orthophoto

4.3.2 Google Earth images

The total PV technical potential of Freihalden reaches the 2740 kWp. The building range from 1 potential kWp to 42 kWp and the specific yield varies from 980,1 to 763,1 kWh/kWp. The PV potentiality map based on Google Earth™ classification is display in Figure 4.7. All buildings with the highest potential are as well south oriented (see Figure 4.7). 98 buildings where found to have no PV potential. The final database displayed in Table D.2 in Appendix D.

Chapter 4. Results 48

Figure 4.7 PV technical potential map based on Google Earth™ orthophoto

4.4 PV prognosis expansion pathway

Figure 4.8 displayed the buildings of the expansion pathway and the installation year according to the technical potential map of the Bavarian land-register orthophoto and Figure 4.9 displayed the buildings of the expansion pathway and the installation year according to the technical potential map of the Google Earth™ orthophoto. If the German Energy Agency forecast are to be accomplish, 99 buildings are required to installed PV in the first expansion pathway and 96 buildings in the second expansion pathway. Between three and eight buildings per year are required to install PV modules in both prognosis.

Chapter 4. Results 49

Figure 4.8 Prognosis expansion pathway based on Bavarian land-register orthophoto

Figure 4.9 Prognosis expansion pathway based on Google Earth™ orthophoto

Chapter 5

Discussion and conclusion

This chapter discusses the results exhibited in the previous chapter. It compares the findings obtained from both images: Bavarian Land-register orthophoto and Google EarthTM orthopho- to. It continues with the review the methodology and finally, it gives an insight on the future research. The chapter closes with the final conclusion.

5.1 Comparison of results

The most significant different among results takes place at the image classification step, as it was expected. For the same training areas, the classification outputs presents similarities but also differences. Looking at the classification statistics both images present almost the same suitable area — 39,5% and 40,1% —, which supports the classification procedure (Figure 5.1). A lot of pixels from darker roofs were classificed as shadows instead of non-suitable on the Bavarian land-register orthophoto, raising the percentage of shadows in the image to a 18%. This misclassification is not occurring on the Google Earth™ image. More obstructions have been also identified in the Google Earth™ image — 1,3% and 2,4% —.

Bavarian land-register orthophoto Google Earth™ orthophoto 50,0% 50,0% 45,0% 45,0% 40,0% 40,0% 35,0% 35,0% 30,0% 30,0% 25,0% 25,0% 20,0% 20,0% 15,0% 15,0% 10,0% 10,0% 5,0% 5,0% 0,0% 0,0% 1 2 3 4 1 2 3 4 Percentage % 39,6% 40,8% 1,3% 18,3% Percentage % 40,1% 46,6% 2,4% 10,8% Figure 5.1 Comparison of image classification statistics

Chapter 5. Discussion and conclusion 51

On the other hand, Google Earth™ image presented a very “noisy” classification output, that required a much more carefull image post-processing in order to obtain the same consistancy of suitable areas. In this process, due to further filtering needed to remove the salt and peper effect, part of the pixels representing small obstruction and shadows on the rooftop where erased together with the “noisy” pixels. This conflicting point was not encounter with the Bavarian land-register orthophoto, the classification presented almost no “noise effect” delivering cleaner outputs.

The Bavarian land-register orthophoto has a average accuracy of 89% while the Google Earth™ image has an average accuracy of 85%. Overall the orthophoto presented a higher accuracy in all rooftypes (red roofs: 91% ; grey roofs: 87% ; black roofs: 76%) compared to the Google Earth™ image (red roofs: 88% ; grey roofs: 82% ; black roofs: 71%) (Tables 1-6 in Appendix C). This can be explained because the Google Earth™ image has lower ground resolution: 0,4 m versus 0,2 m of the Bavarian land-register orthophoto.Furthermore, Google Earth™ image has a lower pixel depth. Original images from DigitalGlobe are processed by Google™ in order to be displayed in Google Earth and Google Maps. The files that Google™ offer to download are the processed images that can be seen in the screen, not the original images that they have bought from DigitalGlobe. On the process the pixel depth per image is lowered in comparision to the origal image, lowering consequently the radiometric resolution (ability to discriminate small differences in the magnitude of radiation within the ground area). This is also the reason behind the “noisy” classification output of the Google Earth™.

It was found that the classification decrease its accuracy depending on the material of the roofs. In this way, reed roofs have the highest accuracy presenting the most separated spec- taral signatures among classes (specially in the red band). On the other hand black and very dark roofs exhibit very similar spectral signatures between suitable, non-suitable and shadow areas, ergo the classification delivered worse results. Figure 3.10 Signature plots of red, grey and black roofs in Chapter 3 clearly illustrate this concept. For that reason, accuracy of the classification on black roofs is a 15% lower than red roofs.

The PV technical potential of Freihalden reaches the 3170 kWp basing the calculation on the Bavarian land-register orthophoto classification.There is -430 kWp difference between the PV technical potential based on the Google Earth™ image classification in comparision to the Bavarian land-register orthophoto classification, which represents a 16% lower technical potential. The lower technical potential is a consequence of the higher number of buildings identified as ‘non suitable for PV’ in the Google Earth™ image classification in comparision to Bavarian land-register orthophoto. Furthermore both have the same maximum and minium buildings with the highest potential and specif yield ranges between 981 and 760 kWh/kWp in both images.

Special mention deserves the orthorectification of the Bavarian land-register orthophoto. Rooftops are elevated from the ground and is a common known problem that even after the orthorectication process, elevated surfaces do not reflect ground-true measures. The further the sensor is to Earth, the better it reflect ground-true measures from elevated surfaces. Fig- ure 5.2 compares rooftops from the Bavarian land-register orthophoto −airborne sensor, clos- er to Earth− to the Google Earth™ image −satellite sensor, futher from Earth−. The Bavar- ian land-register orthophoto (left) illustrate how rooftops are not correctly orthorectified; roof

Chapter 5. Discussion and conclusion 52

do not fall completely inside the buildings (part of the walls can be seen) and the size of each roof-side is distorted. On the contrary, Google Earth™ image (right) display a lower spatial resolution but all roofs fall buildings completely inside the buildings.

Figure 5.2 Comparision of orthorectification on a small subset of Freihalden. [Source: Bavari- an land-register orthophoto (left), Google Earth™ (right)]

The prognosis expansion pathways was found to identify with the highest implementation potential only small to medium size buildings, all facing south, in both cases. Since the PV technical potential maps are not completely the same, the prognosis expansion pathways show also differences on the buildings identified as future buildings installing PV.

With this work, it is proved that it is possible to use open-source software and freely available images downloaded from Google Earth™, and still be able to obtain significantly thruthfull results. Google Earth™ images are already processed images and have lower spatial and radiometric resolution. Nevertheless, accuracy assessments give reasonable good results. Where no airborne high-resolution orthophotos are accessible, Google Earth™ images presents the options to users or administrations in remote rural areas the possibility to assess their community PV potential in a accurate and simple manner.

5.2 Review of methodology

Compare with constant-value methods, this methodology delivers trustworthy building scale results if classification is well made. It was found that in average 39,6 - 40,1% of Freihalden’s rooftop area is suitable for PV. The results are consonant with the studies conducted by Löld [15], Kerber [7] and the International Energy Agency [12] — 44,8% , 40% and 40% respectively—, with the advantage that the PV potential per building is also calculated. If applied to another village, the quantity of suitable area will be adapted to amount of obstructions existing in that village and to the its roof configuration, making the methodoly much more precise.

Chapter 5. Discussion and conclusion 53

On the other hand, supervised classification can hardly be automatizied. Even if signatures for each class are stored and saved as separated files, it is likely that they cannot be applied to other images that were taken on a different date or under other weather conditions. This implies longer working times and may require the need of a person with some remote sensing knowledge if the methodology is to be correctly applied. Additionally some steps previous to the supervised classification cannot be automatized, like isolating the building rooftops. Building footprints do not macht completely building rooftops on the images. Although this is a known problem when working with different GIS data sources, it still needs to be corrected on invididual bases, depending on the input image.

The use of high-resolution aerial orthophotos can also produce inaccuracies. Aerial orthophotos could be not correctly orthorectified for surfaces above the ground connoting inaccuracies when calculating suitable areas. Nevertheless, this does not always have to be case and will depend on the orthorectification applied to the image.

Google earth images are processed images and do not allow the access to the original data. They have lower spatial and radiometric resolution yielding less accurated results. Classifica- tion output presented a lot of noise, which demands longer and carefully post-processing techniques. Still accuracy assessments gives reasonable good results.

Lastly, the prognosis expansion pathway is based exclusively on how profitable the installation of the PV system will be for each rooftop. This variable is represented by the specific yield. However not only profitability plays a role when making the decision of adquiring and installing a PV module. Studies [62–64] have shown that other conditions like environmental convictions, economic status or even political orientation influence this decision. The number of household per building, if known, will also have a non-negligible impact. Therefore, the prognosis expansion pathways should be used only as an orientation of possible PV development in the village.

5.3 Future research and outlook

If future research is to be made following the approach of this thesis, further steps can be implemented in order to improve the classification output. Moving from a pixel-based to a object-based classification may offer more precise results and better recognition of obstructions and shadows. Segmentation of rooftops can delivered meaningful improvements.

Automatize the sequential steps, by writing a R or python code, would speed up the methodology. As this method was implemented using open-source software, programming knowledge will be required for the automatization.

To improve the prognosis expansion pathway, the adquisition of extra data including new variables which reflect the number of households per building, social variables or the energy market rise as an important requirment. Without additional data that can inlight the situation when a person decides on installing a PV system, the expansion pathways may be ob- served only as an orientation.

Chapter 5. Discussion and conclusion 54

The use of LiDAR point data offers another opportunity for future research in both automat- ing rooftop inventories and calculating incoming solar radiation and PV potential for home- owners. This data type make available the creation of 3D models, the extraction of the exact tilt angle and azimuth of each house and permit the direct integration of solar radiation models in the GIS environment. As a drawback, isolation of vegetation near the houses present several problem and flat obstructions (like roof windows) are hardly identified. When as- sessing villages with a reduce number of houses, LiDAR data may become affordable and allow more accurate calculations on the potential energy production per rooftop on a building bases.

5.4 Conclusion

In the present work, an approach to compute the available roof surface for PV on rural rooftops has been proposed. Only very few authors [16–18] have also attempted to use high- resolution images to quantify the suitable rooftop surface per building and the appropriated location of the panels and none of them have addressed the particular building distribution and typology of rural communities. Addressing rural areas has a tremendous importance in Germany where highest PV potential in Germany is expected [5]. The metholodology accounts for the roof surface, roof typology, shadowing, already occupied surface and azimuth angle of the rooftops. The method has been applied to the village of Freihalden (Bavaria) using QGIS open-source software and two different type of orthophotos. The results on rooftop suitable area and on rooftop potential for PV have been compared with each other. The comparision shows that both airborne high-resolution orthophoto and high-resolution Google Earth™ are capable of delivering thrustworthy results with a 11% and 16% error respectabe- ly. According to the present methodology, the total PV technical potential for Freihalden is ranging between 2740 kWp and 3170 kWp and between 98 and 99 buildings will be required to installed new PV systems if the municipality has to fullfil its share on the Bavarian renewable energy goals for 2030.

As regards of additional methodological perspectives, despite the presented method is able to identify shadows and presents a step forward in the evaluation of the roof surface available for PV, the computation of real (in-time) shadowing still remains a open issue. The in-time analysis would required indeed a complete 3D village model and, at today, the data at this detail is not disposable at reasonable prices.

As a concrete application, the method developed in this study has been already successfully used to estimate the PV potential and prognosis expansion pathway of two more municipalities in Bavaria, the village of Tussenhausen (604 buildings) and Hiemenhofen (66 buildings) employing Google Earth™ as part of the research conducted in the Smart-Power-Flow project at the Reiner Lemoine Institute.

Glossary 55

Glossary

Albedo: the ratio of diffusely reflected radiation on a surface.

Azimuth/orientation: the surface azimuth angle, the deviation of the projection of the normal to the plane in question on the equator plane from the local meridian.

Latitude: the angular location north or south of the equator’s plane.

Longitude: the angular location east or west of the prime meridian, Greenwich, England.

Orthophoto: An orthophotograph has the spectral qualities of a photograph but the spatial attributes of a map. Unlike an uncorrected aerial photograph, an orthophotograph can be used to measure true distances, because it is an accurate representation of the Earth's surface, having been adjusted for topographic relief, lens distortion, and camera tilt.

Panel efficiency: is the ratio of the electrical output of a solar panel to the incident energy in the form of sunlight. By convention, solar cell efficiencies are measured under standard test conditions (STC).

Radiometric resolution: is the ability of the sensor to discriminate small differences in the magnitude of radiation within the ground area that corresponds to a single raster cell. It is the range of available brightness values, which in the image correspond to the maximum range of DNs. The greater the bit depth (number of data bits per pixel) of the images that a sensor records, the higher its radiometric resolution.

Solar irradiation: is a measure of the irradiance (power per unit area on the Earth's surface) produced by the Sun in the form of electromagnetic radiation incident on a surface.

Spatial resolution: the resolving power of an instrument needed for the discrimination of features and is based on detector size, focal length, and sensor altitude [62]. It refers to the coarseness or fineness of a raster grid; usually measured in meters.

Spectral resolution: is the ability of a sensor to detect small differences in wavelength [62]; thus the number and location in the electromagnetic spectrum (defined by two wavelengths) of the spectral bands in multispectral sensors [62], for each band corresponds an image.

Tilt/inclination: the angle between the plane of the surface in question and the equator plane.

References 56

References

[1] Directive 2009/29/EC of the European Parliament and of the Council of 23 April 2009 amending Directive 2003/87/EC so as to improve and extend the greenhouse gas emission allowance trading scheme of the Community. OJ L 140/16. 2009.

[2] Bundesministerium für Umwelt Naturschutz und Reaktorsicherheit, “Energiekonzept für eine umweltschonende, zuverlässige und bezahlbare Energieversorgung,” Berlin, Oct. 2010.

[3] H. Wirth, “Recent facts about photovoltaics in Germany,” Berlin, 2014.

[4] Bundesverband Solarwirtschaft e.V., “Statistische Zahlen der deutschen Solarstrombranche (photovoltaik),” Berlin, 2010.

[5] Deutsche Energie-Agentur GmbH (dena), “dena-Netzstudie II . Integration erneuerbarer Energien in die deutsche Stromversorgung im Zeitraum 2015 – 2020 mit Ausblick 2025.,” Berlin, 2010.

[6] Fraunhofer Institute for Solar Energy Systems ISE, “Electricity production from solar and wind in Germany,” Berlin, 2014.

[7] G. Kerber, “Aufnahmefähigkeit von Niederspannungsverteilnetzen für die Einspeisung aus Photovoltaikkleinanlagen,” Technischen Universität München, 2011.

[8] J. A. Jakubiec and C. F. Reinhart, “Towards validated urban photovoltaic potential and solar radiation maps based on LIDAR measurements, GIS data and hourly DAYSIM simulations,” in SimBuild 5th National Conference of IBPSA-USA Madison, Wisconsin August 1-3, 2012, no. Table 1.

[9] J. Schallenberg-Rodríguez, “Photovoltaic techno-economical potential on roofs in regions and islands: The case of the Canary Islands. Methodological review and methodology proposal,” Renew. Sustain. Energy Rev., vol. 20, pp. 219–239, Apr. 2013.

[10] J. Melius, R. Margolis, and S. Ong, “Estimating Rooftop Suitability for PV: A Review of Methods , Patents , and Validation Techniques,” Golden, CO., 2013.

[11] S. Izquierdo, M. Rodrigues, and N. Fueyo, “A method for estimating the geographical distribution of the available roof surface area for large-scale photovoltaic energy-potential evaluations,” Sol. Energy, vol. 82, no. 10, pp. 929–939, Oct. 2008.

[12] IEA -International Energy Agency, “Potential for Building Integrated Photovoltaics. IEA-PVPS T7-4,” Paris, 2002.

[13] H. Lehmann and S. Peter, “Assessment of roof & façade potentials for solar use in Europe,” Insitute Sustain. Solut. Innov., pp. 2–4, 2003.

[14] L. Bergamasco and P. Asinari, “Scalable methodology for the photovoltaic solar energy potential assessment based on available roof surface area: Application to Piedmont Region (Italy),” Sol. Energy, vol. 85, no. 5, pp. 1041–1055, May 2011.

References 57

[15] M. Lödl, G. Kerber, R. Witzmann, C. Hoffmann, and M. Metzger, “Abschätzung des Photovoltaik-Potentials auf Dachflächen in Deutschland,” in 11th Symposium Energieinnovation, 10-12 February 2010, Graz, Austria, 2010, p. 14.

[16] E. Kjellsson, “Potential for building integrated photovoltaics - A study for Sweden,” IEA-PVPS Task, pp. 2–4, 2000.

[17] L. Bergamasco and P. Asinari, “Scalable methodology for the photovoltaic solar energy potential assessment based on available roof surface area: Further improvements by ortho-image analysis and application to Turin (Italy),” Sol. Energy, vol. 85, no. 11, pp. 2741–2756, Nov. 2011.

[18] J. H. Jo and T. P. Otanicar, “A hierarchical methodology for the mesoscale assessment of building integrated roof solar energy systems,” Renew. Energy, vol. 36, no. 11, pp. 2992– 3000, Nov. 2011.

[19] M. H. Kabir, W. Endlicher, and J. Jägermeyr, “Calculation of bright roof-tops for solar PV applications in Dhaka Megacity, Bangladesh,” Renew. Energy, vol. 35, no. 8, pp. 1760–1764, Aug. 2010.

[20] R. Vardimon, “Assessment of the potential for distributed photovoltaic electricity production in Israel,” Renew. Energy, vol. 36, no. 2, pp. 591–594, Feb. 2011.

[21] L. K. K. Wiginton, H. T. Nguyen, and J. M. M. Pearce, “Quantifying rooftop solar photovoltaic potential for regional renewable energy policy,” Comput. Environ. Urban Syst., vol. 34, no. 4, pp. 345–357, Jul. 2010.

[22] J. Ordóñez, E. Jadraque, J. Alegre, and G. Martínez, “Analysis of the photovoltaic solar energy capacity of residential rooftops in Andalusia (Spain),” Renew. Sustain. Energy Rev., vol. 14, no. 7, pp. 2122–2130, Sep. 2010.

[23] T. Santos, N. Gomes, M. Brito, S. Freire, A. Fonseca, and J. A. Tenedorio, “Solar potential analysis in Lisbon using LiDAR data,” in 31st Symposium European Association of Remote Sensing Laboratories (EARSeL), 30th May - 2nd June 2011, Prague, Czech Republic, 2011, pp. 13–19.

[24] H. T. Nguyen and J. M. Pearce, “Incorporating shading losses in solar photovoltaic potential assessment at the municipal scale,” Sol. Energy, vol. 86, no. 5, pp. 1245–1260, May 2012.

[25] J. A. Jakubiec and C. F. Reinhart, “A method for predicting city-wide electricity gains from photovoltaic panels based on LiDAR and GIS data combined with hourly Daysim simulations,” Sol. Energy, vol. 93, pp. 127–143, 2013.

[26] J. Hofierka and J. Kaňuk, “Assessment of photovoltaic potential in urban areas using open-source solar radiation tools,” Renew. Energy, vol. 34, no. 10, pp. 2206–2214, Oct. 2009.

[27] R. Compagnon, “Solar and daylight availability in the urban fabric,” Energy Build., vol. 36, no. 4, pp. 321–328, Apr. 2004.

[28] F. Borfecchia, “Remote Sensing and GIS in planning photovoltaic potential of urban areas,” Eur. J. Remote Sens., no. 195, pp. 195–216, Apr. 2014.

[29] N. Lukač, S. Seme, D. Žlaus, G. Štumberger, and B. Žalik, “Buildings roofs photovoltaic potential assessment based on LiDAR (Light Detection And Ranging) data,” Energy, vol. 66, pp. 598–609, Mar. 2014.

References 58

[30] H. T. Nguyen and J. M. Pearce, “Automated quantification of solar photovoltaic potential in cities,” Int. Rev. Spat. Plan. Sustain. Dev., vol. 1, no. 1, pp. 57–70, 2013.

[31] A. Labouret and M. Villoz, Solar Photovoltaic Energy, 4th ed. London, UK: The Institution of Engineering and Technology, 2010.

[32] P. Fu and P. Rich, “Design and implementation of the Solar Analyst: an ArcView extension for modeling solar radiation at landscape scales,” in Proceedings of the 19th annual ESRI user …, 1999.

[33] M. Šúri and J. Hofierka, “A New GIS-based Solar Radiation Model and Its Application to Photovoltaic Assessments,” Trans. GIS, vol. 8, no. 2, pp. 175–190, 2004.

[34] R. Perez, R. Seals, P. Ineichen, R. Stewart, and D. Menicucci, “A new simplified version of the perez diffuse irradiance model for tilted surfaces,” Sol. Energy, vol. 39, no. 3, pp. 221–231, Jan. 1987.

[35] R. Dubayah and P. M. Rich, “Topographic solar radiation models for GIS,” Int. J. Geogr. Inf. Syst., vol. 9, no. 4, pp. 405–419, Jul. 1995.

[36] C. K. katiyar, A.K.; Pandey, “A Review of Solar Radiation Models—Part I,” Renew. Energy, vol. 2013, p. 11, 2013.

[37] D. W. Mckenney, “Calibration and sensitivity analysis of a spatially-distributed solar radiation model,” Int. J. Geogr. Inf. Sci., vol. 13, no. 1, pp. 49–65, 1999.

[38] J. P. Wilson and J. C. Gallant, Terrain analysis: principles and applications. John Wiley & Sons, 2000.

[39] M. Šúri, T. A. Huld, and E. D. Dunlop, “PV-GIS: a web-based solar radiation database for the calculation of PV potential in Europe,” Int. J. Sustain. Energy, vol. 24, no. 2, pp. 55–67, Jun. 2005.

[40] J. Hofierka, “Direct solar radiation modelling within an open GIS environment,” in Proceedings of JEC-GI’97 conference in Vienna, Austria, IOS Press Amsterdam, 1997, pp. 575–584.

[41] Joint Research Centre Institut for Energy and Transportation, “Photovoltaic Geographical Information System (PVGIS) - About PVGIS,” European Commission, 2012. [Online]. Available: http://re.jrc.ec.europa.eu/pvgis/about_pvgis/about_pvgis.htm.

[42] S. Luque, A.; Hegedus, Handbook of Photovoltaic Science and Engineering, 2nd Edition, 2nd ed. Chichester, West Sussex, England: John Wiley & Sons, Ltd., 2011.

[43] A. Luque and S. Hegedus, Handbook of Photovoltaic Science and Engineering, 1st ed. Chichester, West Sussex, England: John Wiley & Sons, Ltd., 2003, p. 1179.

[44] G. N. Tiwari and S. Dubey, Fundamentals of photovoltaic modules and their applications, no. 2. Royal Society of Chemistry, 2010.

[45] Oberste Baubehörde im Bayerischen Staatsministerium des Innern für Bau und Verkehr, “Solaranlagen gut gestalten.” Oberste Baubehörde im Bayerischen Staatsministerium des Innern, für Bau und Verkehr Kontaktadresse:, München, p. 16, 2013.

[46] SolarPV.co.uk, “Solar PV - Solar PV Orientation - Photovoltaics,” 2014. [Online]. Available: http://www.solarpv.co.uk/solar-pv-orientation.html. [Accessed: 24-Jun-2014].

[47] Energypedia UG, “Solar cells and modules,” Energypedia.info. 2014.

References 59

[48] B. Marion and M. Anderberg, “PVWATTS - an online performance calculator for grid- connected PV systems,” in Solar Conference, 2000, pp. 119–124.

[49] B. Marion, M. Anderberg, and P. Gray-Hann, “Recent Revisions to PVWATTS,” in DOE Solar Energy Technologies, 2005.

[50] National Renewable Energy Laboratory (NREL), “PVWatts Calculator,” 2014. [Online]. Available: http://pvwatts.nrel.gov/index.php.

[51] Elubes Projekt, “Eigen-Verbrauchs-Analyse,” Reiner Lemoine Institut gGmbH, 2014. [Online]. Available: http://eva.elubes.de/. [Accessed: 18-Sep-2014].

[52] Bayerische Staatsregierung, “Bavarian Energy Atlas.” 2014.

[53] Bayerisches Landesamt für Statistik und Datenverarbeitung, “Statistik kommunal 2013 Jettingen-Scheppach. 09 774 144.,” Munich, Germany, 2014.

[54] Statistische Ämter des Bundes und der Länder, “Flächenerhebung nach Art der tatsächlichen Nutzung,” Regionaldatenbank Deutschland, 2014. [Online]. Available: https://www.regionalstatistik.de. [Accessed: 14-Jul-2014].

[55] S. Dahm, “Bau- und Wohnflächenreserven in kleinen Kommunen Baden-Wüttembergs,” Schriftenr. des Instituts für Städtebau und Landesplannung an der Univ. Karlsruhe, no. 36, 2007.

[56] M. Kaltschmitt and A. Wiese, Erneuerbare Energieträger in Deutschland, Potentiale und Kosten. Berlin: Springer Verlag, 1993.

[57] Bundesamt für Kartographie und Geodäsie, “Digitale Orthophotos (DOP),” Geodaten Produkten, 2014. .

[58] Aide CAD Systems, “PDF to DWG Converter.” [Online]. Available: http://www.aidecad.com/pdf-to-dwg.html. [Accessed: 21-May-2014].

[59] M. Elshayal, “El-Shayal Smart GIS.” [Online]. Available: http://freesmartgis.blogspot.com/. [Accessed: 10-Jun-2014].

[60] SASGIS, “SAS.Planet software.” [Online]. Available: http://sasgis.ru/sasplaneta/. [Accessed: 10-Jul-2014].

[61] R. Haselhuhn and C. Hemmerle, Leitfaden Photovoltaische Anlagen, 3rd ed. Deutsche Gesellschaft für Sonnenenergie, 2005.

[62] NASA, “Glosary,” Science Data User Book: Lansat 7, 2011. [Online]. Available: http://landsathandbook.gsfc.nasa.gov/glossary. [Accessed: 28-Aug-2014].

Reports:

Berlin, January 2015

Signature

BSc. Ana Gonzalez Quintairos

Appendix A: Findings in literature review 60

Appendix A: Findings in literature review

Table A.1 Literature review

Smallest Utilization Literature Method- Study area sampled Input data Software factor Results Comments reference ology unit [5] Dena (2012) Germany Municipali- Population Constant- Not speci- Not specified 3587 MW in Bayern by ty density value fied 2030 Total PV from 63% increasing poten- Netzentwick- tial in rural areas lungsplan prognosis

[7] Kerber Bayern, Ger- Building Digital low- Constant- GIS C=0.4 -- (2011) many voltage net- value Estimation works from literature

[9] Schallen- Canary Is- Municipali- Cadaster data Constant- -- Cvoid=0.9 43 km2 of available Carried out a through method- berg- lands ty value Cshadow=0.66 rooftops in Canary ology review Rodríguez Corienta- Islands (16.8 km2 in- (2013) tion=0.5 dustrial, 15.6 km2 Coefficients residential) from literature

[11] Izquierdo et Spain Municipali- Cadaster data Constant- GIS For rural areas mean available roof al. (2008) ty Land-use da- value Cvoid=0.47 area per capita: taset Cshadow=0.46 14.0±4.5 m2/ca (32% Google Earth Cfuntion- Error) DEM al=0.92

Appendix A: Findings in literature review 61

[12] IEA (2002) 23 countries Country Building sta- Constant- GIS C=0.4 mean available roof This study is referenced very tistics value area per capita: often as it was the first of its 18.0m2/ca kind 1295 km2 of available rooftops in Germany. [13] Lehmann & Northrine- State Cadaster data Constant- Not speci- C=0.9 mean available roof Functions to correlate popula- Peter (2003) Westfalia, Statistical value fied area per capita: tion density with rooftop area Germany data 13.4m2/ca 985 km2 of available rooftops in Germany.

[14] Bergamasco North Italy Municipali- Cadaster data Constant- ArcGIS Cresiden- mean available roof Followed the Izquierdo et al. et al. ty Statistical value Matlab tial=0.06 area per capita: (2008) methodology. (2011a) data Cindustri- 9.65m2/ca Google Earth al=0.34 Coefficients from literature [15] Lödl et al. Germany Land Cadaster data Constant- -- C=0.448 PV potential in Bayern: No shadow effect consider (2010) value 2.8 GWp in Rural 12.0 GWp in villages 8.3 GWp in small cities 2.2 GWpin big cities Total 25,3 GWp

[16] Kjellson Sweden Country Building foot- Constant- GIS C=0.65 mean available roof (2000) print value Estimation area per capita: Statistical from literature 51m2/ca data [17] Bergamasco North Italy Building Building foot- Remote ArcGIS -- 17% shaded Considering a10 m2 cut-off et al. print Sensing MATLAB 1% obstructed (2011b) Orthophotos 38% suitable but not bright 43% suitable and bright

Appendix A: Findings in literature review 62

[18] Jo & Chandler, Building Building foot- Remote ArcGIS -- 21% of total rooftop The % of invalid rooftop are is Otanicar Arizona print Sensing Definiens area is unshaded or calculated through object (2011) Quickbird Developer obstructed. brightness classification image Google Sketchup

[19] Kabir et al. Dhaka, Bang- Building Quickbird Remote Object- -- 10.55 km2 of adequate No shadow analysis (2010) ladesh images from Sensing specific rooftop-areas(all flat 2006 image roofs) in Dhaka Meg- recognition acity

[20] Vardimon Israel Municipali- Building foot- Remote GIS C=0.3 mean available roof (2011) ty print Sensing Estimation area per capita: Building sta- from literature 10.2m2/ca tistics review Orthophotos

[21] Wiginton et Ontario, Census Building foot- Remote ArcGIS, Cflat=0.3 mean available roof al. (2010) Canada subdivision print Sensing Object- Cpitched=0.15 area per capita: 13.1± Statistical specific 6.2m2/ca data image Orthophotos recognition

[22] Ordonez et Andalusia, Building Building foot- Remote GIS -- 16-21% suitable for al. (2010) Spain print Sensing AutoCAD pitched roofs Google Earth Sketchup 50-52% suitable for flat roofs

[24] Nguyen & Kingston, Building Land-use da- 3D Model ArcGIS -- 33% of total rooftop manual building-extraction Pearce Ontario tabase PVGIS area is unshaded or processes (2012), LiDAR data GRASS obstructed (2013) Orthophotos

Appendix A: Findings in literature review 63

[25] Jakubiec & Cambridge, Building LiDAR data 3D Model Daysim -- Annual PV yield per The Daysim results are com- Reinhart Massachu- Orthophotos building (kW h) pared with r.sun, PVWatts and (2013) setts With a 3.6–5.3% ac- Solar Analyst curancy

[26] Hofierka & Bardejov, Building LiDAR data 3D Model GIS -- 34.9% suitable for resi- Kanuk Slovakia Orthophotos PVGIS dential roofs (2009) Topographic 16.7% suitable for in- map dustrial roofs

[27] Compagnon Firbourg, Building LiDAR data 3D Model GIS -- 6.5-21% rooftops suita- (2004) Switzerland Orthophotos Daysim ble for PV 30% facade area is adequate for passive solar techniques [23] Santos et al. Lisbon, Por- Building Land-use da- 3D Model ArcGIS -- 1.68 MWh/m2 Considering a10 m2 cut-off (2011) tugal tabase Solar Ra- LiDAR data diation extension

Appendix B: Smarth Power Flow municipalities 64

Appendix B: Smarth Power Flow municipalities

Table B.1 Energy data of Smart-Power-Flow researched municipalities

Munici- Population Popula- pal area density Name tion (km²) (hab/km²) 4.956 31,31 158,3 26.500 7.726 361 7,29 6.655 913 26 % 91% B2 1 4.9 1.721 33,53 51,32 7.779 2.683 261 4,22 5.022 921 66% 191% A4 2.7 6.3 1.296 26,31 49,3 2.724 2.020 240 3,91 3.803 973 140% 189% A4 2.7 6.3 9.361 25,94 360,9 62.500 14.600 343 11,24 10.378 923 17% 73% C1 1.5 5.8 3.055 10,24 298,3 22.600 4.762 110 1,39 1.344 968 6% 30% B1 1.2 4.9 605 9,09 66,6 1.278 943 71 1,16 1.100 946 86% 117% A4 2.7 6.3 1.330 11,64 114,3 9.325 2.073 89 0,88 716 817 8% 36% A2 0.2 2.7 1.655 23,34 70,9 7.052 2.580 103 1,87 1.850 989 26% 72% A4 2.7 6.3 2.393 33,75 70, 5.767 3.730 397 15,34 16.082 1.048 279% 431% A4 2.7 6.3 6.769 54,14 125,0 45.800 10.600 347 9,73 9.615 988 22% 95% A2 0.2 2.7 2.755 2,36 1167,4 5.560 4.295 117 0,83 805 974 15% 19% D1 1.2 6.3 3.718 51,43 72,3 16.600 5.796 441 7,10 6.925 976 43% 122% A4 2.7 6.3 14.223 56,44 252,0 214.000 22.200 497 10,30 10.592 1.028 5% 48% B2 1 4.9 1.312 21,04 62,4 3.964 2.045 206 8,44 8.588 1.018 218% 423% A4 2.7 6.3 1.237 20,81 59,4 2.322 1.928 229 5,58 5.779 1.036 251% 303% A4 2.7 6.3 2.434 21,15 115,1 24.100 3.794 219 3,97 3.667 923 15% 98% A4 2.7 6.3 2.933 41,76 70,2 14.400 4.572 286 5,64 4.722 838 33% 105% A4 2.7 6.3 1.865 17,54 106,3 17.300 2.907 213 13,41 14.511 1.082 84% 501% A4 2.7 6.3

Appendix C : Accuracy assessments 65

Appendix C : Accuracy assessments

Table C.1 Accuracy assessment RED ROOFS in Bavarian land-register orthophoto

→Reference 0 1 2 3 4 data ↓Class Unclassified Suitable Non- Obstruction Shadow Total data suitable 0 Unclassified 0 0 0 0 1 1 1 Suitable 0 34 4 0 0 38 2 Non- 0 4 25 0 0 29 suitable 3 Obstruction 0 0 0 3 0 3 4 Shadow 0 0 1 0 6 7 Total 0 38 26 3 7 74

Overall accuracy [%] = 90.9722222222 Class 0 producer accuracy [%] = nan user accuracy [%] = 0.0 Class 1 producer accuracy [%] = 90.2777777778 user accuracy [%] = 94.2028985507 Class 2 producer accuracy [%] = 91.0714285714 user accuracy [%] = 87.9310344828 Class 3 producer accuracy [%] = 100.0 user accuracy [%] = 100.0 Class 4 producer accuracy [%] = 90.9090909091 user accuracy [%] = 90.9090909091

Table C.2 Accuracy assessment GREY ROOFS in Bavarian land-register orthophoto

→Reference 0 1 2 3 4 data ↓Class Unclassified Suitable Non- Obstruction Shadow Total data suitable 0 Unclassified 0 0 0 0 0 0 1 Suitable 0 20 3 0 0 23 2 Non- 0 0 24 0 1 25 suitable 3 Obstruction 0 0 0 4 0 4 4 Shadow 0 1 3 0 10 14 Total 0 21 30 4 11 66

Overall accuracy [%] = 87.8787878788 Class 0 producer accuracy [%] = nan user accuracy [%] = nan Class 1 producer accuracy [%] = 95.2380952381 user accuracy [%] = 94.2028985507 Class 2 producer accuracy [%] = 80.0 user accuracy [%] = 96.0 Class 3 producer accuracy [%] = 100.0 user accuracy [%] = 100.0 Class 4 producer accuracy [%] = 90.9090909091 user accuracy [%] = 71.4285714286

Appendix C : Accuracy assessments 66

Table C.3 Accuracy assessment BLACK ROOFS in Bavarian land-register orthophoto

→Reference 0 1 2 3 4 data ↓Class Unclassified Suitable Non- Obstruction Shadow Total data suitable 0 Unclassified 0 0 0 0 0 0 1 Suitable 0 9 0 0 2 11 2 Non- 0 4 12 0 1 16 suitable 3 Obstruction 0 0 0 2 0 2 4 Shadow 0 1 0 1 2 1 Total 0 13 12 3 2 30

Overall accuracy [%] = 76.6666666667 Class 0 producer accuracy [%] = nan user accuracy [%] = nan Class 1 producer accuracy [%] = 69.2307692308 user accuracy [%] = 81.8181818182 Class 2 producer accuracy [%] = 100.0 user accuracy [%] = 75.0 Class 3 producer accuracy [%] = 66.6666666667 user accuracy [%] = 100.0 Class 4 producer accuracy [%] = 0.0 user accuracy [%] = 0.0

Table C.4 Accuracy assessment RED ROOFS in Google Earth™ orthophoto

→Reference 0 1 2 3 4 data ↓Class Unclassified Suitable Non- Obstruction Shadow Total data suitable 0 Unclassified 0 0 0 0 0 0 1 Suitable 0 33 1 0 0 34 2 Non- 0 9 28 0 0 37 suitable 3 Obstruction 0 3 0 5 0 8 4 Shadow 0 2 0 0 6 8 Total 0 47 29 5 6 87

Overall accuracy [%] = 87.9813084112 Class 0 producer accuracy [%] = nan user accuracy [%] = nan Class 1 producer accuracy [%] = 75.4385964912 user accuracy [%] = 97.7272727273 Class 2 producer accuracy [%] = 97.4358974359 user accuracy [%] = 80.8510638298 Class 3 producer accuracy [%] = 100.0 user accuracy [%] = 62.0 Class 4 producer accuracy [%] = 100.0 user accuracy [%] = 75.0

Table C.5 Accuracy assessment GREY ROOFS in Google Earth™ orthophoto

Appendix C : Accuracy assessments 67

→Reference 0 1 2 3 4 data ↓Class Unclassified Suitable Non- Obstruction Shadow Total data suitable 0 Unclassified 0 0 0 0 0 0 1 Suitable 0 25 3 2 0 30 2 Non- 0 4 22 0 4 30 suitable 3 Obstruction 0 0 0 6 0 6 4 Shadow 0 0 0 0 3 3 Total 0 29 25 8 7 69

Overall accuracy [%] = 82.3783783784 Class 0 producer accuracy [%] = nan user accuracy [%] = nan Class 1 producer accuracy [%] = 91.2352941176 user accuracy [%] = 81.9473684211 Class 2 producer accuracy [%] = 85.5714285714 user accuracy [%] = 78.3333333333 Class 3 producer accuracy [%] = 76.6666666667 user accuracy [%] = 100.0 Class 4 producer accuracy [%] = 63.3333333333 user accuracy [%] = 100.0

Table C.6 Accuracy assessment BLACK ROOFS in Google Earth™ orthophoto

→Reference 0 1 2 3 4 data ↓Class Unclassified Suitable Non- Obstruction Shadow Total data suitable 0 Unclassified 0 0 0 0 0 0 1 Suitable 0 10 4 1 0 15 2 Non- 0 3 11 0 4 18 suitable 3 Obstruction 0 2 0 4 0 6 4 Shadow 0 0 0 0 0 0 Total 0 15 15 5 4 39

Overall accuracy [%] = 71.2148760331 Class 0 producer accuracy [%] = nan user accuracy [%] = nan Class 1 producer accuracy [%] = 80.1587301587 user accuracy [%] = 76.2264150943 Class 2 producer accuracy [%] = 63.1951219512 user accuracy [%] = 58.8387096774 Class 3 producer accuracy [%] = 96.875 user accuracy [%] = 93.9393939394 Class 4 producer accuracy [%] = 0.0 user accuracy [%] = 0.0

Appendix D : PV potential and expansion pathways 68

Appendix D : PV potential and expansion pathways

Table E.1 Prognosis and expansion pathway based on Bavarian land-register orthophoto classification

TRAFO AREA AREA ENERGY SPEC_YIELD IMPLEMENT X COORD Y COORD ID KWP RANKING NAME TOTAL SUITABLE PROD (kWh) (kWh/kWp) YEAR 4388854.099 5361979.739 11 759C 230.09532 130.31568 13.031568 12772.42271 980.1140364 1 1 4388866.27 5362041.017 194 759 294.0984 163.65085 16.365085 16039.57176 980.1092853 2 1 4388871.181 5362004.438 345 759 279.79224 47.86287 4.786287 4691.055294 980.1032186 3 1 4388676.282 5361982.714 343 759X 259.76952 150.63554 15.063554 14763.81624 980.1017901 4 1 4388745.356 5361758.687 85 231.1476 58.3902 5.83902 5722.801052 980.0961552 5 1 4388620.698 5361851.825 114 759X 62.05548 36.06372 3.606372 3534.589375 980.0956126 6 1 4388853.717 5362013.514 14 759 233.5242 110.77341 11.077341 10856.83601 980.0940505 7 2 4388709.46 5361704.933 62 759F 223.71792 144.0527 14.40527 14118.48029 980.0913336 8 2 4388681.69 5361719.321 279 759F 243.89532 108.55625 10.855625 10639.47551 980.0887111 9 2 4388681.441 5361756.685 289 759F 293.81484 159.81372 15.981372 15663.08601 980.0839383 10 2 4388521.85 5362033.23 195 759X 129.66096 94.42378 9.442378 9254.302528 980.0817684 11 2 4388526.098 5361751.567 79 759X 84.44292 47.3902 4.73902 4644.625197 980.0813664 12 2 4388499.626 5361791.926 93 759X 71.49372 33.96679 3.396679 3329.019737 980.0807603 13 2 4388613.549 5361627.769 253 759F 268.17192 118.74841 11.874841 11638.26717 980.0777263 14 3 4388756.742 5362051.371 384 759C 269.4024 107.45625 10.745625 10531.42431 980.0662416 15 3 4388494.061 5361759.431 285 759X 741.11724 292.32071 29.232071 28649.23769 980.0618536 16 3 4388928.371 5361973.848 175 143.81256 49.53872 4.953872 4855.065758 980.0547446 17 4 4388647.093 5361643.817 262 759F 287.025 117.96213 11.796213 11560.92805 980.0541962 18 4 4388475.436 5361653.501 54 759D 109.725 67.11287 6.711287 6577.408344 980.0517164 19 4 4388871.449 5361983.725 342 759 45.75708 14.19693 1.419693 1391.367579 980.0482072 20 4 4388660.615 5362044.467 348 759X 363.3258 89.66287 8.966287 8787.337201 980.0419283 21 4 4388901 5361827.905 759 70.70388 21.78088 2.178088 2134.612582 980.0396413 22 4

Appendix D : PV potential and expansion pathways 69

4388465.188 5361672.823 56 759D 173.37888 58.70392 5.870392 5753.186332 980.0344392 23 4 4388586.205 5361908.878 149 759X 580.34532 138.44105 13.844105 13567.63366 980.0296704 24 4 4388654.408 5361781.245 90 759F 57.28596 28.90074 2.890074 2832.354581 980.028394 25 5 4388589.272 5361857.183 10 759X 247.58676 82.74497 8.274497 8109.227749 980.0266709 26 5 4388526.273 5361770.416 290 759X 223.77192 84.67426 8.467426 8298.295771 980.0257801 27 5 4388393.692 5361411.058 224 416.41404 70.01324 7.001324 6861.474334 980.0252543 28 5 4388643.83 5361740.845 380 759F 157.0992 153.05895 15.305895 15000.16288 980.0252048 29 5 4388765.793 5361665.557 227 759F 49.93356 15.24963 1.524963 1494.498609 980.0228657 30 5 4388705.331 5361686.917 18 759F 84.84144 19.84301 1.984301 1944.657282 980.0213181 31 5 4388710.145 5361646.345 263 759F 83.24064 23.86483 2.386483 2338.798591 980.0189615 32 5 4388565.543 5361744.929 77 759D 325.7016 27.26372 2.726372 2671.89357 980.0179762 33 5 4388638 5361680.333 271 759F 266.69292 85.38321 8.538321 8367.70596 980.0177295 34 6 4388649.801 5361583.622 242 759F 45.75468 11.30074 1.130074 1107.486227 980.012129 35 6 4388479.376 5361743.589 280 759X 736.53516 319.11605 31.911605 31273.62112 980.0077783 36 6 4388414.101 5361611.823 16 759D 65.68356 40.95784 4.095784 4013.880226 980.0029069 37 6 4388367.822 5361593.99 42 759D 167.98824 97.74105 9.774105 9578.636548 980.0013964 38 6 4388384.839 5361551.592 34 759D 282.06564 100.31054 10.031054 9830.428448 979.9995541 39 7 4388795.5 5361882.924 140 759C 199.5258 59.30122 5.930122 5811.512189 979.998757 40 7 4388501.692 5361716.144 67 759X 305.58516 157.60932 15.760932 15445.65335 979.9961925 41 7 4388606.658 5361939.29 336 759X 550.66644 201.52341 20.152341 19748.70024 979.9705273 42 7 4388897.834 5361872.617 759 289.73208 80.60503 8.060503 7899.02968 979.9673396 43 8 4388993.49 5361813.009 6 759 241.67112 101.19571 10.119571 9916.724869 979.9550662 44 8 4388833.717 5361759.925 759 334.73436 155.2375 15.52375 15212.53718 979.9524716 45 8 4388557.392 5361859.996 123 759X 157.56096 40.63983 4.063983 3982.503915 979.9509287 46 8 4388735.092 5361909.553 153 759C 264.06792 160.12304 16.012304 15691.0327 979.9359732 47 8 4388677.803 5361796.09 295 759F 244.6758 114.73946 11.473946 11243.45633 979.911909 48 9 4388914.431 5361973.522 173 759 52.78596 17.05858 1.705858 1671.574229 979.9023298 49 9 4388768.745 5361974.497 177 759C 265.83276 91.09375 9.109375 8926.222803 979.8940984 50 9 4388576.348 5361890.968 392 759X 611.08824 195.43909 19.543909 19150.65269 979.8783187 51 9

Appendix D : PV potential and expansion pathways 70

4388492.447 5361705.795 63 759X 47.75628 30.31875 3.031875 2970.820756 979.8625459 52 9 4388487.988 5361645.922 53 759D 261.69372 132.53713 13.253713 12986.76245 979.8584327 53 9 4388477.156 5361631.265 256 759D 250.19292 115.05956 11.505956 11274.19154 979.8570012 54 10 4388803.558 5362012.926 189 759C 77.04144 18.59253 1.859253 1821.793879 979.8525964 55 10 4388866.356 5361750.264 759 463.32888 172.30048 17.230048 16882.39364 979.8227865 56 10 4388648.992 5362021.1 347 759X 426.0726 149.81912 14.981912 14679.43507 979.8105256 57 10 4388834.068 5361990.19 344 759C 134.2266 37.67071 3.767071 3690.968899 979.7980709 58 10 4388765.876 5361807.495 297 759C 44.61324 34.9723 3.49723 3426.528346 979.7835276 59 10 4388691.948 5362334.032 33 120.73128 57.35906 5.735906 5619.668483 979.7351078 60 11 4388455.302 5361795.534 94 759X 40.66176 11.70895 1.170895 1147.159165 979.7284683 61 11 4388897.196 5361901.619 759 50.09292 13.83162 1.383162 1355.10465 979.7150659 62 11 4388924.301 5361954.968 166 759C 202.1274 133.03554 13.303554 13033.55448 979.7047078 63 11 4388622.75 5361890.569 142 759X 152.51724 60.68909 6.068909 5945.735167 979.7041227 64 11 4388786.749 5361735.894 75 378.76644 58.08517 5.808517 5690.600612 979.6993987 65 11 4389067.481 5361850.91 9 759 168.00936 49.07034 4.907034 4807.409447 979.6976029 66 11 4388414.051 5361667.839 17 759D 630.0984 358.85355 35.885355 35156.11181 979.6785293 67 11 4388441.646 5361702.18 64 759D 637.00308 311.42463 31.142463 30507.72104 979.6181196 68 12 4388637.903 5362100.46 209 395.62968 106.55821 10.655821 10437.96597 979.5553034 69 12 4388539.681 5361814.035 301 759X 184.28676 34.68872 3.468872 3397.887979 979.5368578 70 12 4389109.584 5361828.775 308 759 171.62808 43.95699 4.395699 4305.679939 979.5211043 71 12 4388448.801 5361722.203 69 759X 328.38276 105.73761 10.573761 10356.74419 979.4759112 72 12 4388825.392 5362033.003 198 759C 267.94692 89.98946 8.998946 8814.145986 979.4642602 73 13 4388643.965 5361999.037 401 759X 517.57968 220.825 22.0825 21628.58376 979.4445264 74 13 4388656.038 5361866.41 318 759C 174.51096 52.37892 5.237892 5130.222798 979.4441729 75 13 4388487.391 5362002.266 186 759X 249.88128 100.11298 10.011298 9805.501941 979.4436187 76 13 4388989.234 5361778.892 293 759 176.05548 32.02892 3.202892 3136.965819 979.4166705 77 13 4388858.217 5361858.681 316 759C 149.92032 36.08517 3.608517 3534.241265 979.4165485 78 13 4388833.429 5361793.267 759 338.08824 128.64841 12.864841 12599.97105 979.4113313 79 14 4388592.632 5361868.627 131 759X 74.53824 32.51446 3.251446 3184.487937 979.4066817 80 14

Appendix D : PV potential and expansion pathways 71

4388732.335 5361841.272 112 759C 240.21096 81.03051 8.103051 7935.751273 979.3534896 81 14 4389026.898 5361843.15 310 759 196.5516 43.56176 4.356176 4265.763275 979.244933 82 14 4388366.6 5361611.436 49 759D 334.03356 162.35747 16.235747 15898.69136 979.2399055 83 14 4388855.462 5361767.723 759 48.47112 16.44841 1.644841 1610.578729 979.1698586 84 14 4388633.235 5361617.158 250 759F 233.49372 99.65318 9.965318 9757.227821 979.1185611 85 15 4388691.199 5361859.769 317 759F 290.31096 93.9598 9.39598 9199.636012 979.1034051 86 15 4388698.607 5361619.333 252 759F 279.45468 106.99216 10.699216 10475.26241 979.0682239 87 15 4388284.837 5361666.363 394 759D 174.03756 141.33713 14.133713 13837.06484 979.0113069 88 15 4388708.622 5362071.456 204 759C 527.24292 278.35159 27.835159 27250.04948 978.9794797 89 16 4388859.509 5361781.898 759 267.71256 88.7304 8.87304 8686.404806 978.9660371 90 16 4388659.12 5361768.702 381 759F 275.21724 109.48872 10.948872 10718.48329 978.9577674 91 16 4388483.71 5361951.872 340 759X 578.33208 131.61753 13.161753 12884.0199 978.898472 92 16 4388885.239 5361906.787 759 107.96256 66.18051 6.618051 6478.180625 978.8653223 93 16 4388581.838 5361698.433 274 46.96872 15.79534 1.579534 1546.071868 978.8151875 94 17 4388859.293 5361842.946 110 759C 109.875 44.83358 4.483358 4388.237689 978.7836904 95 17 4388362.07 5361575.06 38 759D 208.34064 124.47193 12.447193 12180.79073 978.5974018 96 17 4388446.177 5361887.266 322 759X 324.74532 113.15392 11.315392 11072.54602 978.5384388 97 17 4389009.024 5361831.239 306 759 71.55 47.61801 4.761801 4659.433825 978.5024249 98 17 4388480.398 5361838.229 309 759X 171.98436 19.42622 1.942622 1900.846572 978.4953387 99 no 4388576.765 5361646.936 260 759F 290.86644 113.21838 11.321838 11077.82821 978.4478646 100 no 4389016 5361848.684 313 759 106.49064 11.4598 1.14598 1121.245969 978.4166991 101 no 4388277.707 5361604.196 225 759D 635.56872 112.23003 11.223003 10980.30466 978.3749201 102 no 4388619.716 5361963.544 172 759X 472.01256 153.43284 15.343284 15010.66022 978.3212131 103 no 4388523.268 5362046.415 199 759X 122.71644 85.21997 8.521997 8336.995373 978.2912823 104 no 4388665.927 5362001.042 405 759X 181.73676 138.01128 13.801128 13501.36883 978.2800961 105 no 4388376.791 5361632.189 378 759D 353.6484 118.16409 11.816409 11559.48924 978.2573742 106 no 4388951.062 5361896.697 324 759 236.06484 73.66568 7.366568 7206.246659 978.2366305 107 no 4388950.482 5361798.764 95 759 363.2016 82.14338 8.214338 8034.347305 978.0882288 108 no 4388793.878 5361777.94 759 759F 305.4 85.70122 8.570122 8382.332342 978.0878664 109 no

Appendix D : PV potential and expansion pathways 72

4388811.854 5361949.273 759 190.98048 15.29693 1.529693 1496.073431 978.0220156 110 no 4388530.419 5362013.636 346 759X 229.31952 71.61176 7.161176 7003.682938 978.0073744 111 no 4388766.076 5361746.248 78 343.57968 120.76372 12.076372 11810.78034 978.0073304 112 no 4388597.08 5361718.59 278 48.38436 16.77071 1.677071 1640.16689 977.9949029 113 no 4388412.255 5361858.696 124 759X 564.77808 248.92659 24.892659 24339.81433 977.7908552 114 no 4388255.697 5361642.299 264 759D 327.79452 101.12267 10.112267 9886.930375 977.7165076 115 no 4388694.46 5361943.36 165 759C 269.40468 101.1098 10.11098 9884.284014 977.5792271 116 no 4388566.349 5362091.464 206 759X 496.88436 237.27341 23.727341 23193.43964 977.4984749 117 no 4388959.886 5361867.212 128 759 90.14292 54.0375 5.40375 5281.723723 977.4182231 118 no 4388695.662 5361584.079 40 759F 49.78824 24.05392 2.405392 2351.008945 977.3911882 119 no 4388741.4 5361872.909 136 759C 256.90308 91.99259667 9.19925967 8988.390523 977.077596 120 no 4388872.308 5361810.81 298 759C 156.78276 100.63284 10.063284 9832.502936 977.0670226 121 no 4388501.696 5362012.078 190 759X 214.5516 34.74449 3.474449 3394.730764 977.0558624 122 no 4388536.267 5362059.884 349 759X 968.22192 386.03125 38.603125 37716.82215 977.040645 123 no 4388739.977 5362074.053 203 759C 31.65708 21.6777 2.16777 2117.957281 977.0212158 124 no 4389004.485 5361788.453 4 759 44.08596 16.08321 1.608321 1571.052795 976.8278811 125 no 4388462.285 5362255.354 218 59.01096 24.53517 2.453517 2396.243376 976.6565204 126 no 4388700.863 5362147.354 360 112.85628 23.55122 2.355122 2299.935536 976.5674711 127 no 4388504.779 5361975.844 178 759X 186.41484 81.02622 8.102622 7912.04934 976.4801246 128 no 4388388.086 5361642.792 383 759D 418.68048 194.03395 19.403395 18946.69063 976.4626568 129 no 4388075.122 5361705.014 65 167.47032 86.2125 8.62125 8418.085446 976.4344435 130 no 4388836.484 5361972.939 341 759C 228.40548 38.60318 3.860318 3769.176746 976.3902211 131 no 4388951.05 5361883.055 139 759 111.42888 39.5098 3.95098 3857.427421 976.3216775 132 no 4388395.38 5361845.789 115 759X 824.43516 379.22071 37.922071 37019.6885 976.2042928 133 no 4388402.893 5361827.19 22 759X 70.62888 52.69693 5.269693 5144.294663 976.2038629 134 no 4388902.38 5362019.014 192 759 230.18208 82.77929 8.277929 8080.778753 976.1836267 135 no 4388232.601 5361676.399 57 759D 40.61484 22.95392 2.295392 2240.416034 976.0494216 136 no 4388279.309 5361692.707 61 759D 818.49144 495.58443 49.558443 48367.6913 975.9727783 137 no 4388356.041 5361845.618 315 759X 181.1976 103.10784 10.310784 10061.19598 975.7934971 138 no

Appendix D : PV potential and expansion pathways 73

4388219.399 5361604.377 47 759D 366.0726 84.48517 8.448517 8243.809066 975.7699565 139 no 4388783.52 5361855.074 122 759C 319.67808 94.44534 9.444534 9215.667327 975.7672879 140 no 4388425.603 5361686.382 60 759D 737.86644 351.48443 35.148443 34292.54809 975.6491373 141 no 4388918.759 5361940.993 164 759C 279.16404 96.63247 9.663247 9426.388628 975.4887387 142 no 4388235.713 5361620.826 51 759D 340.66644 46.9777 4.69777 4581.928493 975.341171 143 no 4388659.018 5362161.205 361 216.27888 129.32733 12.932733 12613.47009 975.3135779 144 no 4388952.849 5361932.43 160 759 261.48984 101.63824 10.163824 9910.909847 975.1162404 145 no 4388404.664 5361652.383 266 759D 135.75708 59.2152 5.92152 5773.847345 975.0616978 146 no 4388271.573 5361651.846 267 759D 257.93904 79.39338 7.939338 7740.665603 974.9762011 147 no 4388255.042 5361949.827 233 271.80936 195.31446 19.531446 19041.50414 974.9152284 148 no 4388764.14 5362132.865 211 230.0742 67.64142 6.764142 6592.160561 974.5745374 149 no 4388925.088 5361780.718 91 759 219.525 75.47892 7.547892 7355.820246 974.5529277 150 no 4388938.516 5361867.036 133 759 338.99532 63.77426 6.377426 6214.332617 974.4264562 151 no 4388715.178 5361889.019 143 759C 209.3274 72.67304 7.267304 7081.26489 974.4005329 152 no 4388801.877 5361916.889 759 277.6242 94.12304 9.412304 9171.297748 974.3945529 153 no 4388432.683 5361867.157 132 759X 199.4508 76.83676 7.683676 7485.666601 974.229861 154 no 4388423.564 5361576.868 241 759D 256.7766 69.18395 6.918395 6739.320468 974.1161741 155 no 4388758.787 5361863.314 400 759C 275.79372 81.92053667 8.19205367 7979.595965 974.0653915 156 no 4388412.316 5361903.874 150 759X 41.01564 22.61446 2.261446 2202.763917 974.0510791 157 no 4388364.429 5361879.037 135 759X 252.40776 149.12304 14.912304 14521.54343 973.7960969 158 no 4388408.747 5361561.365 36 759D 329.54064 124.81997 12.481997 12152.80777 973.626878 159 no 4388827.228 5361870.506 134 759C 169.1016 61.40233 6.140233 5977.988431 973.5768057 160 no 4388371.5 5361401.508 223 483.71016 96.24142 9.624142 9365.319292 973.1069317 161 no 4388426.219 5361887.19 138 759X 184.78356 57.75429 5.775429 5615.396115 972.2907363 162 no 4388591.355 5362213.028 366 85.725 52.02659 5.202659 5058.19014 972.2317261 163 no 4388639.583 5362138.942 359 145.40388 58.91017 5.891017 5727.432581 972.2315487 164 no 4388595.354 5362180.843 213 25.81872 12.98088 1.298088 1262.040596 972.2303853 165 no 4388248.543 5361624.799 52 759D 98.72112 69.10662 6.910662 6718.363316 972.1736234 166 no 4388585.548 5362136.838 355 523.15548 195.66679 19.566679 19019.56907 972.0386925 167 no

Appendix D : PV potential and expansion pathways 74

4388882.462 5362030.544 385 759 246.7524 75.68946 7.568946 7350.082297 971.083992 168 no 4388444.999 5361601.194 44 759D 655.55856 297.0473 29.70473 28845.44622 971.0724931 169 no 4388963.816 5361854.839 120 759 176.1 52.86017 5.286017 5132.624768 970.9815099 170 no 4388907.521 5361725.874 73 759 303.96564 146.17537 14.617537 14192.37429 970.9142035 171 no 4388644.113 5361850.162 119 759C 267.78756 85.98051 8.598051 8334.863706 969.3898891 172 no 4388456.293 5362279.45 370 360.1008 135.60943 13.560943 13145.17722 969.3409392 173 no 4388754.763 5361829.387 104 759C 337.0266 99.62744 9.962744 9652.412908 968.8508415 174 no 4388343.772 5361753.62 284 759D 656.9508 234.97474 23.497474 22759.00325 968.5723344 175 no 4388895.227 5361841.398 759 191.94144 60.81801 6.081801 5882.704306 967.2635304 176 no 4388538.91 5362032.046 197 759X 199.16724 52.70122 5.270122 5095.607527 966.8860659 177 no 4388506.765 5361806.487 296 759X 237.78048 68.42341 6.842341 6604.220922 965.1990337 178 no 4388466.772 5361939.704 27 759X 170.9766 50.32071 5.032071 4852.217603 964.2585733 179 no 4388527.585 5361922.311 333 759X 570.39612 170.8223 17.08223 16460.42251 963.5991616 180 no 4388599.147 5362325.215 371 880.88436 167.9596333 16.79596333 16179.52879 963.2986492 181 no 4388555.061 5362298.308 374 1528.36872 362.9570333 36.29570333 34963.54953 963.2972039 182 no 4388682.005 5361835.827 373 759F 260.1774 60.89453333 6.08945333 5865.79379 963.2709986 183 no 4388656.416 5361650.778 372 759F 49.82112 7.79296667 0.77929667 750.6658622 963.2607081 184 no 4388878.238 5361849.146 116 759C 172.38048 55.43395 5.543395 5338.114227 962.9684025 185 no 4388497.079 5361892.164 323 759X 349.59372 95.78591 9.578591 9218.679385 962.4254115 186 no 4388543.825 5361965.236 171 759X 519.7758 161.18443 16.118443 15499.01418 961.5701825 187 no 4388407.49 5361939.804 163 759X 66.53436 45.925 4.5925 4408.094134 959.8463004 188 no 4388743.255 5362058.652 386 759C 161.72808 69.06372 6.906372 6628.305947 959.7377533 189 no 4389158.379 5361824.609 230 759E 177.3516 50.36372 5.036372 4833.073988 959.6340357 190 no 4388977.435 5361759.13 84 759 329.27808 104.18639 10.418639 9998.038472 959.6299931 191 no 4388889.963 5361995.564 182 759 176.92032 50.19608 5.019608 4816.829712 959.6027641 192 no 4388418.164 5361927.432 334 759X 103.9266 40.43787 4.043787 3877.857569 958.9668223 193 no 4388685.585 5361924.372 332 759C 353.23596 113.09375 11.309375 10840.23329 958.5174506 194 no 4388740.801 5361939.107 759 244.37808 60.30233 6.030233 5779.739817 958.460447 195 no 4388729.871 5362009.662 187 759C 246.99372 90.05392 9.005392 8620.125549 957.2182476 196 no

Appendix D : PV potential and expansion pathways 75

4388856.593 5361904.663 759 73.1016 16.79216 1.679216 1606.61732 956.7663241 197 no 4388325.54 5361753.251 288 759D 147.48516 47.46753 4.746753 4538.898511 956.2112271 198 no 4388645.862 5362398.098 237 64.6476 35.1098 3.51098 3354.67592 955.4813529 199 no 4388239.448 5361556.733 35 759D 294.9258 188.7402 18.87402 18017.36861 954.6121391 200 no 4388851.511 5362057.466 201 759 293.75388 121.0946 12.10946 11559.53017 954.5867584 201 no 4388631.404 5361830.646 8 759D 258.11484 80.81568 8.081568 7703.343243 953.1990875 202 no 4388808.397 5361722.251 72 340.3476 107.40048 10.740048 10215.66993 951.1754449 203 no 4388759.042 5361960.593 759 46.29372 13.3375 1.33375 1267.287802 950.1689239 204 no 4388658.957 5362075.703 351 268.0734 77.1848 7.71848 7325.163877 949.0422825 205 no 4388385.965 5361817.694 299 759D 582.1476 165.9669 16.59669 15743.00959 948.5632127 206 no 4388836.569 5362073.421 202 759 451.60308 188.36642 18.836642 17862.07334 948.2620808 207 no 4388703.94 5361992.831 13 759C 57.06324 21.32108 2.132108 2017.444008 946.2203639 208 no 4388718.574 5361982.263 12 759C 255.2484 49.58162 4.958162 4686.030191 945.1143772 209 no 4388497.097 5361958.419 168 759X 286.65468 108.20381 10.820381 10214.59172 944.0140523 210 no 4388502.928 5361860.684 126 759X 166.56792 40.66568 4.066568 3836.592396 943.4472498 211 no 4388714.17 5361849.506 395 759C 313.22112 90.8446 9.08446 8550.256227 941.1958693 212 no 4388920.374 5361752.197 82 759 374.53356 109.32966 10.932966 10244.8067 937.056486 213 no 4388803.498 5361663.211 3 759F 300.84612 13.23872 1.323872 1238.866279 935.7900759 214 no 4388474.6 5361917.693 330 190.35 60.28517 6.028517 5635.2614 934.7674395 215 no 4388725.726 5361734.169 76 759F 288.92808 180.25821 18.025821 16812.22078 932.6743439 216 no 4388690.028 5362124.133 352 233.1774 81.54608 8.154608 7594.768073 931.3468009 217 no 4388643.654 5361977.384 234 759X 49.76016 23.74449 2.374449 2209.185887 930.3993839 218 no 4388510.445 5362333.343 221 335.18436 166.51679 16.651679 15485.5939 929.9719208 219 no 4388911.622 5361900.371 759 358.38516 169.50747 16.950747 15670.26995 924.4589604 220 no 4388666.359 5361704.463 276 759F 290.65308 108.62071 10.862071 10027.51347 923.1677343 221 no 4388480.549 5361700.741 275 759X 50.03676 20.47892 2.047892 1889.497563 922.6548875 222 no 4388227.372 5361748.464 20 759D 400.18596 153.11923 15.311923 14118.94991 922.0886177 223 no 4388828.805 5361830.41 105 759C 336.17112 113.78983 11.378983 10474.54327 920.5166463 224 no 4388588.465 5361607.721 249 759F 283.81404 99.52426 9.952426 9143.15986 918.6865454 225 no

Appendix D : PV potential and expansion pathways 76

4388265.197 5361568.019 37 759D 539.55 208.32108 20.832108 19124.51973 918.0309418 226 no 4388804.728 5361806.648 1 759C 311.3484 160.25196 16.025196 14684.27582 916.324257 227 no 4388939.96 5362001.882 184 759 319.91484 160.23051 16.023051 14621.87268 912.5523396 228 no 4388679.244 5362095.188 207 236.6274 72.48824 7.248824 6593.871059 909.6470075 229 no 4388726.439 5361693.905 409 759F 304.02888 157.9875 15.79875 14370.94247 909.6252847 230 no 4388928.082 5361835.153 107 759 176.21016 56.95943 5.695943 5180.158638 909.447064 231 no 4388961.266 5361964.041 169 759 215.05308 62.10699 6.210699 5646.835535 909.2109495 232 no 4388465.871 5361620.586 393 759D 262.79064 116.07794 11.607794 10540.80177 908.0796724 233 no 4388893.422 5361804.484 759 310.91256 90.2473 9.02473 8137.80202 901.7224914 234 no 4388472.806 5362257.489 219 179.1258 76.02034 7.602034 6843.555338 900.2268785 235 no 4388826.433 5361931.1 759 93.4476 61.07145 6.107145 5473.974037 896.3229195 236 no 4388781.948 5361817.157 303 759C 329.63904 136.11642 13.611642 12167.02013 893.8686555 237 no 4388616.427 5362315.322 371 586.91952 349.79571 34.979571 31263.90547 893.7761264 238 no 4388499.929 5362273.439 220 222.17808 74.60233 7.460233 6613.266054 886.4691028 239 no 4388520.289 5362240.602 369 309.0726 122.17304 12.217304 10789.86299 883.1623564 240 no 4388687.828 5362209.883 215 74.5758 35.40625 3.540625 3121.09007 881.5082282 241 no 4388990.187 5361952.826 167 759 517.39452 147.96716 14.796716 13004.78321 878.8965882 242 no 4388884.772 5361890.234 759 247.45548 118.78713 11.878713 10434.66879 878.4342872 243 no 4388267.656 5362109.385 236 126.61176 25.3473 2.53473 2224.103398 877.4517987 244 no 4388550.301 5362226.423 368 306.8016 103.6277 10.36277 9063.508903 874.62222 245 no 4388464.797 5361846.967 312 759X 130.03128 20.42733 2.042733 1782.930309 872.8161289 246 no 4388534.404 5361633.111 257 759D 239.34372 96.53358 9.653358 8423.013182 872.5474785 247 no 4388668.177 5361597.866 247 759F 481.50936 193.31642 19.331642 16842.11649 871.2201731 248 no 4388342.057 5361644.569 265 759D 159.14292 75.03628 7.503628 6524.170273 869.4687788 249 no 4388616.286 5362208.907 367 141.6234 50.04142 5.004142 4349.684025 869.2167459 250 no 4388597.821 5362191.579 404 127.5726 96.3402 9.63402 8369.087897 868.7015283 251 no 4388622.684 5362157.531 362 182.36724 68.31605 6.831605 5934.608582 868.6990219 252 no 4388500.88 5362098.895 208 759X 202.42968 169.82108 16.982108 14748.03877 868.4457061 253 no 4388637.512 5362183.084 363 160.53048 62.03395 6.203395 5379.289398 867.1524864 254 no

Appendix D : PV potential and expansion pathways 77

4388957.419 5361991.487 181 759 273.47808 140.94608 14.094608 12209.68064 866.2660674 255 no 4389037.14 5361862.474 121 759 201.81564 89.15159 8.915159 7706.85438 864.4662848 256 no 4388631.458 5362198.327 365 44.17032 20.28554 2.028554 1752.789005 864.0583417 257 no 4389236.012 5361839.269 23 759E 206.13984 50.69878 5.069878 4380.438695 864.0126439 258 no 4388840.681 5361695.212 59 759F 243.05628 30.49926 3.049926 2634.518978 863.7976719 259 no 4388540.455 5362240.965 216 320.10468 191.91568 19.191568 16547.65736 862.2358194 260 no 4388257.403 5361685.064 58 759D 105.09612 22.79926 2.279926 1962.91512 860.9556274 261 no 4388552.518 5362242.807 217 89.83824 59.74375 5.974375 5142.139399 860.6991357 262 no 4388370.641 5361898.642 25 759X 91.28904 66.17193 6.617193 5691.984007 860.1810476 263 no 4388318.716 5361621.469 50 759D 232.16952 121.26642 12.126642 10415.36325 858.8827186 264 no 4388248.751 5361939.781 232 92.03904 22.71324 2.271324 1938.541011 853.4850206 265 no 4388748.368 5362149.758 212 85.75776 70.87696 7.087696 6043.646771 852.6955404 266 no 4388253.406 5361668.058 55 759D 203.73516 99.95392 9.995392 8518.038668 852.1965589 267 no 4388657.29 5362121.75 353 130.28676 43.54449 4.354449 3708.583595 851.6768929 268 no 4388780.374 5361642.023 261 759F 332.31564 150.29179 15.029179 12799.68071 851.6553505 269 no 4388386.986 5361704.871 19 759D 258.77112 153.31679 15.331679 13049.82717 851.1675186 270 no 4388722.411 5362135.167 358 182.21952 92.57193 9.257193 7869.877197 850.1364503 271 no 4388385.424 5361603.482 45 759D 89.44452 16.90392 1.690392 1435.768036 849.3698718 272 no 4388535.306 5361845.524 314 759X 227.26644 69.21838 6.921838 5875.640898 848.8555928 273 no 4388710.235 5362122.209 354 157.93128 25.16679 2.516679 2133.71056 847.8278556 274 no 4388240.037 5361581.484 41 759D 204.69612 23.65 2.365 1999.080891 845.2773322 275 no 4388929.571 5361804.202 98 759 289.83756 95.26176 9.526176 8043.882303 844.3978259 276 no 4388588.95 5361713.154 277 166.2234 130.66372 13.066372 11024.45638 843.7274231 277 no 4388484.749 5361559.846 240 759D 39.7734 14.89301 1.489301 1255.806399 843.2186635 278 no 4388624.529 5361658.853 268 759F 291.36564 108.16949 10.816949 9113.203509 842.4929718 279 no 4388434.589 5361815.127 101 759X 234.56016 105.81912 10.581912 8906.796592 841.7001192 280 no 4388556.12 5361828.774 307 759X 123.94452 34.75747 3.475747 2925.515256 841.6939598 281 no 4389122.055 5361819.267 21 759E 50.9484 15.78247 1.578247 1323.133991 838.3567281 282 no 4388488.894 5361580.254 244 759D 300.62808 141.2598 14.12598 11839.41818 838.1307477 283 no

Appendix D : PV potential and expansion pathways 78

4388921.5 5362046.651 200 759 63.02112 12.16875 1.216875 1019.265117 837.6087252 284 no 4388425.503 5361800.286 97 759X 201.71256 142.30392 14.230392 11905.27606 836.6091434 285 no 4388309.404 5361739.191 283 759D 202.8258 24.65122 2.465122 2058.574325 835.0800994 286 no 4389091.272 5361835.86 108 759 131.95548 37.27537 3.727537 3112.636294 835.0383361 287 no 4388919.357 5362011.3 191 759 286.88208 191.19804 19.119804 15921.86491 832.7420568 288 no 4389083.364 5361852.299 118 759 97.73208 75.59926 7.559926 6273.942617 829.8947128 289 no 4388249.371 5361595.416 43 759D 133.0734 40.22304 4.022304 3337.217875 829.6781831 290 no 4388339.349 5361730.002 74 759D 274.59612 91.70392 9.170392 7608.456732 829.6762813 291 no 4388403.932 5361753.304 81 759D 285.83904 25.66091 2.566091 2129.016274 829.6729439 292 no 4388552.087 5361658.142 382 759D 214.63824 230.85392 23.085392 19105.32937 827.59389 293 no 4388591.27 5361894.012 144 759X 42.2508 27.28517 2.728517 2254.57698 826.3012398 294 no 4388640.803 5361778.521 88 759F 302.83824 218.41017 21.841017 18035.81294 825.7771579 295 no 4388834.182 5361899.986 759 159.19224 53.84412 5.384412 4443.486204 825.2500374 296 no 4388785.063 5361974.718 759 52.7976 24.91753 2.491753 2055.039735 824.7365348 297 no 4388574.101 5361686.513 272 152.56404 74.1598 7.41598 6114.35179 824.4833172 298 no 4388544.048 5361867.081 130 759X 57.56016 37.25392 3.725392 3070.578936 824.2297552 299 no 4388845.714 5361893.509 759 69.69372 50.15747 5.015747 4134.043158 824.2128557 300 no 4388810.519 5362031.913 193 759C 102.06792 21.35122 2.135122 1754.814287 821.880102 301 no 4388736.152 5361765.687 86 82.10388 44.02145 4.402145 3612.298525 820.5769063 302 no 4388760.284 5361621.533 254 759F 302.65548 204.37659 20.437659 16770.58231 820.5725671 303 no 4388439.275 5361531.229 239 759D 430.69452 172.61409 17.261409 14160.70627 820.367924 304 no 4388620.82 5361696.477 273 759F 272.7 113.87574 11.387574 9335.538863 819.8005004 305 no 4388747.444 5361721.323 71 759F 359.29692 198.79497 19.879497 16291.90459 819.533039 306 no 4388966.879 5361920.618 157 759 194.15388 100.80466 10.080466 8242.768925 817.6972102 307 no 4388280.92 5361577.534 243 759D 182.5524 28.14878 2.814878 2298.872026 816.6862031 308 no 4388771.796 5361723.237 70 35.84064 12.14301 1.214301 991.6713485 816.660242 309 no 4388744.082 5361597.013 248 759F 309.81324 104.17341 10.417341 8507.419347 816.65939 310 no 4388891.832 5361819.479 759 86.94612 59.71801 5.971801 4876.901276 816.6550218 311 no 4388797.449 5362036.06 196 759C 159.9258 54.79804 5.479804 4462.391962 814.3342283 312 no

Appendix D : PV potential and expansion pathways 79

4388667.56 5361886.978 141 759C 284.48676 106.70858 10.670858 8686.937841 814.0805398 313 no 4389017.453 5361868.969 127 759 316.43904 124.56642 12.456642 10134.11135 813.5508227 314 no 4388732.002 5362059.731 386 759C 281.73048 53.62929 5.362929 4361.690399 813.3037747 315 no 4388628.499 5362083.496 205 202.56324 25.18395 2.518395 2045.632854 812.2764116 316 no 4388758.032 5361900.048 26 759C 264.02808 63.55074 6.355074 5150.526829 810.4589858 317 no 4388996.033 5361867.626 129 759 150.72888 64.03628 6.403628 5186.480691 809.9284798 318 no 4388782.92 5361936.642 759 50.01096 26.27108 2.627108 2122.390384 807.8809031 319 no 4388855.037 5361928.846 759 209.48436 50.51409 5.051409 4077.016166 807.1047437 320 no 4388759.624 5361934.639 759 249.06096 82.01446 8.201446 6609.866896 805.9392083 321 no 4388842.061 5361932.648 759 235.5492 46.94338 4.694338 3782.791883 805.8200928 322 no 4388945.301 5361831.83 103 759 124.3266 26.95 2.695 2167.515853 804.2730439 323 no 4388871.024 5361920.085 759 197.75856 17.94375 1.794375 1439.50036 802.2293892 324 no 4389021.744 5361924.361 158 759 278.88516 96.16838 9.616838 7709.810004 801.699062 325 no 4389015.242 5361906.712 327 759 135.5766 43.19216 4.319216 3460.550993 801.1988734 326 no 4388444.531 5361838.703 109 759X 147.18048 41.94179 4.194179 3350.88701 798.9375299 327 no 4388321.955 5361708.184 397 759D 548.76792 132.27071 13.227071 10564.30794 798.6883827 328 no 4389028.339 5361815.39 300 759 100.2726 25.94449 2.594449 2068.101843 797.125649 329 no 4388982.703 5361922.127 156 759 123.76404 35.87463 3.587463 2854.220413 795.6097143 330 no 4388991.567 5361881.764 320 759 177.98436 63.95037 6.395037 5083.096375 794.8501901 331 no 4388781.42 5361920.495 759 144.18756 55.90662 5.590662 4424.11854 791.3407285 332 no 4388461.56 5361927.258 335 170.46096 69.45466 6.945466 5449.384229 784.5959118 333 no 4388475.971 5361869.744 125 759X 379.88208 96.31875 9.631875 7502.020695 778.8743827 334 no 4388482.819 5361977.084 179 759X 225.00936 77.4598 7.74598 6031.228807 778.6269532 335 no 4388798.325 5362003.297 185 759C 164.76096 56.16875 5.616875 4308.363369 767.039211 336 no 4388618.815 5362109.219 210 294.77808 238.38199 23.838199 18232.61597 764.8487189 337 no 4388881.685 5361774.704 759 115.61952 53.5348 5.35348 4078.852619 761.9067632 338 no

Appendix D : PV potential and expansion pathways 80

Table E.2 Prognosis and expansion pathway based on Google Earth™ image classification

TRAFO AREA AREA ENERGY SPEC_YIELD IMPLEMENT X COORD Y COORD ID KWP RANKING NAME TOTAL SUITABLE PROD (kWh) (kWh/kWp) YEAR 4388660.924 5362045.426 348 759X 363.3258 163.841992 16.3841992 16058.28975 980.1083078 1 1 4388866.562 5361752.142 759 463.32888 108.483203 10.8483203 10632.45573 980.1015671 2 1 4388681.694 5361718.284 279 759F 243.89532 88.341602 8.8341602 8658.27659 980.0905116 3 1 4388705.335 5361685.88 18 759F 84.84144 21.159961 2.1159961 2073.86598 980.0896987 4 1 4388675.082 5361777.683 291 759F 52.61484 20.526172 2.0526172 2011.747102 980.0887872 5 1 4388854.026 5362014.473 14 759 233.5242 66.773438 6.6773438 6544.38743 980.0884343 6 1 4388710.148 5361645.307 263 759F 83.24064 34.243945 3.4243945 3356.203609 980.0867305 7 1 4388643.834 5361739.808 380 759F 157.0992 124.085156 12.4085156 12161.41674 980.0863481 8 1 4388499.626 5361791.926 93 759X 71.49372 46.765039 4.6765039 4583.374886 980.0857614 9 2 4388666.362 5361703.425 276 759F 290.65308 78.716602 7.8716602 7714.901498 980.0856874 10 2 4388919.666 5362012.259 191 759 286.88208 100.407227 10.0407227 9840.768138 980.0856404 11 2 4388606.967 5361940.249 336 759X 550.66644 265.067773 26.5067773 25978.88096 980.0844766 12 2 4388747.729 5361722.283 71 759F 359.29692 102.869336 10.2869336 10082.04759 980.0828876 13 2 4388902.689 5362019.973 192 759 230.18208 78.763867 7.8763867 7719.479406 980.0787721 14 3 4388676.591 5361983.673 343 759X 259.76952 36.06582 3.606582 3534.728805 980.0772047 15 3 4388576.348 5361890.968 392 759X 611.08824 279.791016 27.9791016 27421.66492 980.0766769 16 3 4388833.924 5361761.802 759 334.73436 83.634375 8.3634375 8196.790323 980.0743203 17 3 4388677.803 5361796.09 295 759F 244.6758 97.332813 9.7332813 9539.337013 980.0741105 18 3 4388478.746 5361742.323 280 759X 736.53516 240.541211 24.0541211 23574.79032 980.0728209 19 4 4388685.894 5361925.331 332 759C 353.23596 95.2875 9.52875 9338.692951 980.0543566 20 4 4388423.53 5361576.362 241 759D 256.7766 59.268945 5.9268945 5808.645656 980.0487686 21 4 4388787.034 5361736.854 75 378.76644 42.743164 4.2743164 4189.035727 980.0481142 22 4 4388698.61 5361618.296 252 759F 279.45468 82.3625 8.23625 8071.851526 980.039645 23 4 4388882.771 5362031.503 385 759 246.7524 77.739063 7.7739063 7618.724113 980.0380683 24 5 4388557.392 5361859.996 123 759X 157.56096 59.533203 5.9533203 5834.405586 980.0254803 25 5

Appendix D : PV potential and expansion pathways 81

4388766.361 5361747.207 78 343.57968 142.067578 14.2067578 13922.82395 980.0141695 26 5 4388371.5 5361401.508 223 483.71016 186.654102 18.6654102 18292.33001 980.0122161 27 5 4388695.662 5361584.079 40 759F 49.78824 21.941992 2.1941992 2150.33633 980.0096224 28 5 4388597.083 5361717.553 278 48.38436 20.867773 2.0867773 2045.061143 980.0092914 29 5 4388866.579 5362041.975 194 759 294.0984 81.127148 8.1127148 7950.467028 980.0008042 30 6 4388631.404 5361830.646 8 759D 258.11484 65.38125 6.538125 6407.353205 979.9985783 31 6 4388681.441 5361756.685 289 759F 293.81484 147.064844 14.7064844 14412.29575 979.995991 32 6 4388928.681 5361974.807 175 143.81256 52.48418 5.248418 5143.426343 979.9955612 33 6 4388898.143 5361873.576 759 289.73208 111.62207 11.162207 10938.87063 979.9917373 34 6 4388477.122 5361630.759 256 759D 250.19292 96.619531 9.6619531 9468.537666 979.9817458 35 7 4388901.285 5361828.865 759 70.70388 26.198047 2.6198047 2567.348888 979.9772052 36 7 4388393.692 5361411.058 224 416.41404 151.737695 15.1737695 14869.89044 979.9733967 37 7 4388465.837 5361620.08 393 759D 262.79064 88.614453 8.8614453 8683.858069 979.9595636 38 7 4388620.025 5361964.502 172 759X 472.01256 246.735156 24.6735156 24178.92 979.9543928 39 7 4388989.234 5361778.892 293 759 176.05548 58.957422 5.8957422 5777.468596 979.9391493 40 8 4388451.007 5361809.876 99 759X 49.8024 16.626758 1.6626758 1629.249106 979.8958435 41 8 4388735.402 5361910.511 153 759C 264.06792 72.90293 7.290293 7143.699676 979.8919846 42 8 4388745.641 5361759.647 85 231.1476 84.629102 8.4629102 8292.633247 979.8796219 43 8 4388769.054 5361975.456 177 759C 265.83276 118.997656 11.8997656 11660.14063 979.8630509 44 8 4388487.954 5361645.416 53 759D 261.69372 99.904492 9.9904492 9789.224506 979.8582937 45 8 4388632.461 5361615.943 250 759F 233.49372 91.156055 9.1156055 8931.486534 979.8017843 46 8 4388448.767 5361721.697 69 759X 328.38276 113.61582 11.361582 11131.20943 979.7235482 47 9 4388757.051 5362052.33 384 759C 269.4024 71.631055 7.1631055 7017.849224 979.721606 48 9 4388940.269 5362002.841 184 759 319.91484 84.2875 8.42875 8257.712969 979.7079009 49 9 4388909.37 5361985.419 180 759 151.11564 56.615625 5.6615625 5546.630226 979.6995486 50 9 4388656.038 5361866.41 318 759C 174.51096 56.834766 5.6834766 5567.994786 979.6811315 51 9 4388425.569 5361685.876 60 759D 737.86644 312.08418 31.208418 30572.93626 979.6374895 52 10 4388539.681 5361814.035 301 759X 184.28676 66.032227 6.6032227 6468.718188 979.6304747 53 10 4388859.499 5361844.823 110 759C 109.875 41.42832 4.142832 4058.404815 979.6209005 54 10

Appendix D : PV potential and expansion pathways 82

4388644.274 5361999.996 401 759X 517.57968 230.179297 23.0179297 22548.54874 979.6080286 55 10 4388804.934 5361808.525 1 759C 311.3484 59.266797 5.9266797 5805.776813 979.6002326 56 11 4388691.405 5361861.647 317 759F 290.31096 117.614063 11.7614063 11521.08426 979.5668954 57 11 4389067.79 5361851.868 9 759 168.00936 45.695117 4.5695117 4475.982419 979.5318872 58 11 4388741.11 5361940.066 759 244.37808 85.477734 8.5477734 8372.528659 979.4982 59 11 4388655.153 5361805.48 5 759F 231.15708 17.967383 1.7967383 1759.892408 979.4928999 60 11 4388576.769 5361645.899 260 759F 290.86644 106.345508 10.6345508 10416.31557 979.47866 61 11 4388404.668 5361651.346 266 759D 135.75708 77.792773 7.7792773 7619.562499 979.4691981 62 11 4388919.733 5361754.043 82 759 374.53356 144.632813 14.4632813 14165.51439 979.4122162 63 11 4388639.583 5362138.942 359 145.40388 68.025977 6.8025977 6661.854065 979.3103104 64 12 4388808.682 5361723.211 72 340.3476 120.245898 12.0245898 11775.55717 979.2897192 65 12 4388993.799 5361813.968 6 759 241.67112 59.288281 5.9288281 5804.375499 979.0089038 66 12 4388285.403 5361665.58 394 759D 174.03756 124.033594 12.4033594 12141.71439 978.9053105 67 12 4388694.769 5361944.319 165 759C 269.40468 116.509766 11.6509766 11405.11462 978.8977365 68 12 4388938.825 5361867.995 133 759 338.99532 96.995508 9.6995508 9494.543706 978.8642693 69 12 4388497.097 5361958.419 168 759X 286.65468 90.668359 9.0668359 8874.765621 978.8161735 70 13 4388483.71 5361951.872 340 759X 578.33208 200.65332 20.065332 19640.25344 978.8152742 71 13 4388709.46 5361704.933 62 759F 223.71792 111.248242 11.1248242 10888.63323 978.7690152 72 13 4388446.177 5361887.266 322 759X 324.74532 139.766602 13.9766602 13678.97921 978.7015648 73 13 4388461.56 5361927.258 335 170.46096 71.613867 7.1613867 7007.717466 978.5419723 74 13 4388858.423 5361860.558 316 759C 149.92032 53.317773 5.3317773 5217.137997 978.4988577 75 14 4388729.871 5362009.662 187 759C 246.99372 97.730273 9.7730273 9562.526745 978.4610696 76 14 4388414.017 5361667.333 17 759D 630.0984 274.411328 27.4411328 26849.46124 978.4385153 77 14 4388612.775 5361626.554 253 759F 268.17192 117.345508 11.7345508 11480.63674 978.3618426 78 14 4388526.098 5361751.567 79 759X 84.44292 41.048047 4.1048047 4015.809359 978.3192264 79 15 4388977.643 5361759.898 84 759 329.27808 163.474609 16.3474609 15992.16117 978.2657546 80 15 4388376.169 5361631.309 378 759D 353.6484 133.486719 13.3486719 13058.43524 978.2572629 81 15 4388264.458 5361584.034 245 759D 102.08436 60.448438 6.0448438 5913.353146 978.2474687 82 15 4389027.207 5361844.109 310 759 196.5516 71.422656 7.1422656 6986.820091 978.2358264 83 15

Appendix D : PV potential and expansion pathways 83

4388859.715 5361783.775 759 267.71256 73.966406 7.3966406 7235.58707 978.2261247 84 15 4388536.113 5362061.913 349 759X 968.22192 300.858594 30.0858594 29430.02003 978.2010757 85 16 4388586.205 5361908.878 149 759X 580.34532 232.085742 23.2085742 22700.89311 978.1252788 86 16 4388441.612 5361701.674 64 759D 637.00308 187.799219 18.7799219 18368.97442 978.1177214 87 17 4388659.12 5361768.702 381 759F 275.21724 85.176953 8.5176953 8330.958958 978.0766586 88 17 4388366.566 5361610.93 49 759D 334.03356 116.975977 11.6975977 11440.50346 978.0216205 89 17 4388754.97 5361831.265 104 759C 337.0266 44.055859 4.4055859 4308.714947 978.0117888 90 17 4388951.371 5361897.656 324 759 236.06484 81.438672 8.1438672 7962.614671 977.7436782 91 no 4388487.391 5362002.266 186 759X 249.88128 62.184375 6.2184375 6079.446653 977.6485898 92 no 4388649.301 5362022.059 347 759X 426.0726 180.842578 18.0842578 17678.86568 977.5831488 93 no 4388795.706 5361884.802 140 759C 199.5258 82.921094 8.2921094 8106.188799 977.5786121 94 no 4388219.425 5361605.728 47 759D 366.0726 90.633984 9.0633984 8859.478455 977.5007193 95 no 4388566.349 5362091.464 206 759X 496.88436 190.499805 19.0499805 18614.90529 977.1613829 96 no 4388480.398 5361838.229 309 759X 171.98436 19.069531 1.9069531 1863.36341 977.1417083 97 no 4388395.38 5361845.789 115 759X 824.43516 401.497851 40.1497851 39221.3414 976.8755002 98 no 4388996.342 5361868.585 129 759 150.72888 59.041211 5.9041211 5766.233754 976.6455763 99 no 4388402.893 5361827.19 22 759X 70.62888 59.700781 5.9700781 5829.570222 976.4646499 100 no 4388502.928 5361860.684 126 759X 166.56792 67.733789 6.7733789 6613.14924 976.3442054 101 no 4388493.469 5361757.633 285 759X 741.11724 192.976367 19.2976367 18840.41629 976.3069218 102 no 4388344.196 5361755.493 284 759D 656.9508 202.778125 20.2778125 19791.02288 975.9939775 103 no 4388412.255 5361858.696 124 759X 564.77808 287.905664 28.7905664 28097.76259 975.9364301 104 no 4388432.683 5361867.157 132 759X 199.4508 73.953516 7.3953516 7216.898957 975.8696202 105 no 4388953.158 5361933.389 160 759 261.48984 91.407422 9.1407422 8919.971046 975.847568 106 no 4388715.384 5361890.897 143 759C 209.3274 93.072461 9.3072461 9082.369964 975.8385957 107 no 4388925.296 5361781.486 91 759 219.525 81.539648 8.1539648 7956.936599 975.8365157 108 no 4388726.439 5361693.905 409 759F 304.02888 86.994531 8.6994531 8488.537278 975.755278 109 no 4388384.805 5361551.086 34 759D 282.06564 105.522396 10.5522396 10295.59238 975.6784127 110 no 4388506.765 5361806.487 296 759X 237.78048 114.803906 11.4803906 11200.43714 975.6146394 111 no 4388426.219 5361887.19 138 759X 184.78356 38.132617 3.8132617 3719.905016 975.5178923 112 no

Appendix D : PV potential and expansion pathways 84

4388236.551 5361621.629 51 759D 340.66644 71.607422 7.1607422 6984.707137 975.4166456 113 no 4388255.659 5361641.576 264 759D 327.79452 47.716797 4.7716797 4654.040881 975.3464552 114 no 4388239.448 5361556.733 35 759D 294.9258 134.24082 13.424082 13090.07582 975.1188813 115 no 4388918.759 5361940.993 164 759C 279.16404 128.21875 12.821875 12501.84649 975.0404285 116 no 4388289.813 5361686.724 61 759D 818.49144 144.14082 14.414082 14048.42981 974.632294 117 no 4388530.419 5362013.636 346 759X 229.31952 104.139063 10.4139063 10149.69862 974.6293396 118 no 4388272.401 5361650.913 267 759D 257.93904 61.170313 6.1170313 5959.647916 974.2712803 119 no 4388339.349 5361730.002 74 759D 274.59612 105.011328 10.5011328 10230.19954 974.1996159 120 no 4388278.181 5361603.704 225 759D 635.56872 87.651953 8.7651953 8537.692064 974.0447043 121 no 4388500.88 5362098.895 208 759X 202.42968 47.61582 4.761582 4637.715132 973.9861946 122 no 4388897.506 5361902.577 759 50.09292 18.235938 1.8235938 1775.713566 973.7440248 123 no 4388456.293 5362279.45 370 360.1008 128.990039 12.8990039 12546.76114 972.6922509 124 no 4388659.018 5362161.205 361 216.27888 58.09375 5.809375 5649.154841 972.4204138 125 no 4388691.948 5362334.032 33 120.73128 43.185742 4.3185742 4198.717306 972.2461885 126 no 4388585.857 5362137.797 355 523.15548 173.447656 17.3447656 16859.78401 972.0387349 127 no 4388732.541 5361843.15 112 759C 240.21096 90.154883 9.0154883 8759.924978 971.6528585 128 no 4388644.113 5361850.162 119 759C 267.78756 120.671289 12.0671289 11725.04843 971.6518754 129 no 4388741.607 5361874.787 136 759C 256.90308 121.030209 12.1030209 11757.91123 971.4856589 130 no 4388834.377 5361991.149 344 759C 134.2266 57.891797 5.7891797 5621.432883 971.0240785 131 no 4388455.302 5361795.534 94 759X 40.66176 13.374023 1.3374023 1298.375147 970.8186884 132 no 4389016.309 5361849.643 313 759 106.49064 12.291211 1.2291211 1193.037506 970.6427675 133 no 4388638.212 5362101.419 209 395.62968 106.826758 10.6826758 10362.03366 969.9848475 134 no 4389158.379 5361824.609 230 759E 177.3516 79.354688 7.9354688 7697.137763 969.9663571 135 no 4388895.433 5361843.275 759 191.94144 70.455859 7.0455859 6831.269882 969.5815194 136 no 4388782.154 5361819.035 303 759C 329.63904 85.133984 8.5133984 8253.573653 969.48049 137 no 4388616.286 5362208.907 367 141.6234 68.013086 6.8013086 6592.490764 969.2974031 138 no 4388957.728 5361992.446 181 759 273.47808 48.44082 4.844082 4694.381047 969.0961151 139 no 4388501.657 5361715.638 67 759X 305.58516 87.811002 8.7811002 8504.444941 968.4942373 140 no 4388871.49 5362005.397 345 759 279.79224 78.542578 7.8542578 7606.093195 968.4038122 141 no

Appendix D : PV potential and expansion pathways 85

4388890.272 5361996.522 182 759 176.92032 66.807813 6.6807813 6469.024507 968.3035886 142 no 4388364.429 5361879.037 135 759X 252.40776 85.75293 8.575293 8295.70045 967.3955689 143 no 4388538.91 5362032.046 197 759X 199.16724 26.574023 2.6574023 2568.509037 966.5488122 144 no 4388640.803 5361778.521 88 759F 302.83824 105.02207 10.502207 10150.72834 966.5328767 145 no 4388645.862 5362398.098 237 64.6476 41.677539 4.1677539 4027.343217 966.3102269 146 no 4388758.032 5361900.048 26 759C 264.02808 84.345313 8.4345313 8144.451368 965.6080556 147 no 4388783.726 5361856.951 122 759C 319.67808 109.755078 10.9755078 10578.07825 963.7894155 148 no 4388964.125 5361855.798 120 759 176.1 63.971875 6.3971875 6163.988164 963.5465842 149 no 4388599.147 5362325.215 371 880.88436 246.434245 24.6434245 23738.97848 963.2986877 150 no 4388555.061 5362298.308 374 1528.36872 425.39388 42.539388 40978.07585 963.2972589 151 no 4388523.268 5362046.415 199 759X 122.71644 34.397135 3.4397135 3313.418155 963.283179 152 no 4388521.85 5362033.23 195 759X 129.66096 36.119141 3.6119141 3479.293432 963.2824412 153 no 4388412.316 5361903.874 150 759X 41.01564 11.199219 1.1199219 1078.793283 963.2754592 154 no 4388370.641 5361898.642 25 759X 91.28904 25.349609 2.5349609 2441.865073 963.2752413 155 no 4388622.75 5361890.569 142 759X 152.51724 12.714844 1.2714844 1224.788437 963.2744505 156 no 4388616.326 5361879.036 137 759X 71.5008 8.811198 0.8811198 848.7594764 963.2736393 157 no 4388682.005 5361835.827 373 759F 260.1774 72.208333 7.2208333 6955.620251 963.2711298 158 no 4388526.273 5361770.416 290 759X 223.77192 22.936198 2.2936198 2209.369442 963.2675138 159 no 4388736.437 5361766.646 86 82.10388 22.619792 2.2619792 2178.889988 963.2670308 160 no 4388492.413 5361705.289 63 759X 47.75628 13.127604 1.3127604 1264.534879 963.2640343 161 no 4388253.406 5361668.058 55 759D 203.73516 37.1875 3.71875 3582.131668 963.2622972 162 no 4388465.154 5361672.317 56 759D 173.37888 26.224609 2.6224609 2526.117112 963.262069 163 no 4388656.416 5361650.778 372 759F 49.82112 14.048828 1.4048828 1353.268431 963.2607293 164 no 4388414.105 5361610.786 16 759D 65.68356 18.234375 1.8234375 1756.442565 963.2589903 165 no 4388367.788 5361593.484 42 759D 167.98824 17.447917 1.7447917 1680.684444 963.2579314 166 no 4388362.035 5361574.554 38 759D 208.34064 43.124349 4.3124349 4153.98287 963.2569455 167 no 4388647.097 5361642.779 262 759F 287.025 99.55 9.955 9588.118908 963.146048 168 no 4388836.793 5361973.897 341 759C 228.40548 70.864063 7.0864063 6822.865046 962.8103099 169 no 4388827.435 5361872.384 134 759C 169.1016 71.605273 7.1605273 6891.469939 962.4249235 170 no

Appendix D : PV potential and expansion pathways 86

4388527.585 5361922.311 333 759X 570.39612 184.406835 18.4406835 17743.67095 962.2024559 171 no 4388794.084 5361779.818 759 759F 305.4 118.460547 11.8460547 11383.92166 960.9884427 172 no 4388510.445 5362333.343 221 335.18436 131.877604 13.1877604 12670.78873 960.7991305 173 no 4388833.635 5361795.144 759 338.08824 136.558984 13.6558984 13110.05402 960.0286732 174 no 4388840.681 5361695.212 59 759F 243.05628 11.850781 1.1850781 1137.231975 959.6261842 175 no 4388444.965 5361600.688 44 759D 655.55856 233.754297 23.3754297 22402.05405 958.3590266 176 no 4388851.82 5362058.425 201 759 293.75388 134.803711 13.4803711 12903.69122 957.2207713 177 no 4388878.444 5361851.024 116 759C 172.38048 67.381445 6.7381445 6448.806075 957.059629 178 no 4388758.994 5361865.192 400 759C 275.79372 100.832617 10.0832617 9642.773116 956.3148714 179 no 4388950.691 5361799.532 95 759 363.2016 111.334179 11.1334179 10607.29934 952.744201 180 no 4388659.266 5362076.662 351 268.0734 94.664453 9.4664453 8995.387264 950.2391847 181 no 4388679.553 5362096.147 207 236.6274 72.02207 7.202207 6842.604406 950.0705 182 no 4388714.376 5361851.383 395 759C 313.22112 84.919141 8.4919141 8061.126955 949.2709017 183 no 4388589.272 5361857.183 10 759X 247.58676 89.991602 8.9991602 8507.004288 945.3109067 184 no 4388408.713 5361560.859 36 759D 329.54064 100.398633 10.0398633 9482.508111 944.4857791 185 no 4388885.081 5361891.193 759 247.45548 121.337305 12.1337305 11454.53305 944.0240204 186 no 4388309.505 5361737.56 283 759D 202.8258 32.278125 3.2278125 3026.327081 937.5783385 187 no 4388906.88 5361727.72 73 759 303.96564 114.928516 11.4928516 10759.62005 936.2010773 188 no 4388726.011 5361735.128 76 759F 288.92808 126.207812 12.6207812 11811.5137 935.8781767 189 no 4388543.825 5361965.236 171 759X 519.7758 184.591601 18.4591601 17218.50476 932.7891771 190 no 4388385.965 5361817.694 299 759D 582.1476 196.457422 19.6457422 18319.02492 932.4679484 191 no 4388643.654 5361977.384 234 759X 49.76016 22.049414 2.2049414 2046.735882 928.2495588 192 no 4388690.028 5362124.133 352 233.1774 91.693164 9.1693164 8499.419997 926.9415108 193 no 4388708.622 5362071.456 204 759C 527.24292 391.183203 39.1183203 36141.49185 923.9019357 194 no 4388587.691 5361606.506 249 759F 283.81404 82.966211 8.2966211 7638.70617 920.7008586 195 no 4388829.011 5361832.288 105 759C 336.17112 120.3125 12.03125 11067.84558 919.9248273 196 no 4388552.518 5362242.807 217 89.83824 51.347656 5.1347656 4721.573563 919.5304967 197 no 4388667.403 5361596.651 247 759F 481.50936 170.360352 17.0360352 15573.0669 914.1250721 198 no 4388265.737 5361568.442 37 759D 539.55 188.791797 18.8791797 17248.65609 913.633768 199 no

Appendix D : PV potential and expansion pathways 87

4388990.497 5361953.785 167 759 517.39452 157.892969 15.7892969 14401.64857 912.1146215 200 no 4388836.569 5362073.421 202 759 451.60308 138.226172 13.8226172 12592.31967 910.9938799 201 no 4388504.779 5361975.844 178 759X 186.41484 34.26543 3.426543 3111.42866 908.0372432 202 no 4388540.455 5362240.965 216 320.10468 114.470898 11.4470898 10347.05906 903.9030217 203 no 4388387.464 5361641.911 383 759D 418.68048 131.581055 13.1581055 11875.4464 902.5194702 204 no 4388771.796 5361723.237 70 35.84064 12.950781 1.2950781 1161.578655 896.9178421 205 no 4388386.986 5361704.871 19 759D 258.77112 81.045508 8.1045508 7253.459355 894.9859818 206 no 4388616.427 5362315.322 371 586.91952 196.710937 19.6710937 17547.59009 892.0495402 207 no 4388657.29 5362121.75 353 130.28676 49.0875 4.90875 4350.595571 886.2939794 208 no 4388583.603 5361704.488 406 117.65856 120.905469 12.0905469 10660.78784 881.7457079 209 no 4388267.656 5362109.385 236 126.61176 39.535547 3.9535547 3482.15537 880.7656992 210 no 4388472.806 5362257.489 219 179.1258 45.74668 4.574668 4025.705072 879.9993949 211 no 4388834.491 5361900.945 759 159.19224 58.899414 5.8899414 5172.439379 878.1818065 212 no 4388464.797 5361846.967 312 759X 130.03128 29.319727 2.9319727 2574.105589 877.943232 213 no 4388781.729 5361921.454 759 144.18756 67.65 6.765 5939.194152 877.9296603 214 no 4388797.758 5362037.018 196 759C 159.9258 70.866211 7.0866211 6205.22606 875.6254881 215 no 4388342.023 5361644.063 265 759D 159.14292 53.977344 5.3977344 4705.672009 871.7865053 216 no 4388824.467 5361978.76 176 759C 48.88128 17.986719 1.7986719 1566.644631 871.0007819 217 no 4388295.94 5361782.678 294 759D 55.53276 17.344336 1.7344336 1510.282903 870.7643247 218 no 4388499.929 5362273.439 220 222.17808 102.364453 10.2364453 8910.968901 870.5139958 219 no 4388637.512 5362183.084 363 160.53048 72.415234 7.2415234 6300.038523 869.9880088 220 no 4388597.821 5362191.579 404 127.5726 107.488477 10.7488477 9329.225676 867.9279805 221 no 4388622.684 5362157.531 362 182.36724 78.7875 7.87875 6836.137142 867.6677317 222 no 4388318.716 5361621.469 50 759D 232.16952 96.436914 9.6436914 8332.682814 864.0553154 223 no 4388911.931 5361901.33 759 358.38516 165.678906 16.5678906 14294.92973 862.809278 224 no 4388534.408 5361632.074 257 759D 239.34372 91.001367 9.1001367 7829.951747 860.4213327 225 no 4388929.78 5361804.97 98 759 289.83756 98.34043 9.834043 8451.094877 859.3713569 226 no 4389236.012 5361839.269 23 759E 206.13984 66.358789 6.6358789 5699.151898 858.8390452 227 no 4388637.226 5361679.118 271 759F 266.69292 32.14707 3.214707 2760.203017 858.6172914 228 no

Appendix D : PV potential and expansion pathways 88

4388240.037 5361581.484 41 759D 204.69612 89.228906 8.9228906 7656.833724 858.1113528 229 no 4388631.458 5362198.327 365 44.17032 25.385938 2.5385938 2177.082476 857.5938678 230 no 4388248.751 5361939.781 232 92.03904 43.166406 4.3166406 3691.963164 855.2862065 231 no 4388418.164 5361927.432 334 759X 103.9266 11.61875 1.161875 993.7244579 855.2765641 232 no 4388780.378 5361640.985 261 759F 332.31564 131.707813 13.1707813 11250.70454 854.2169431 233 no 4388710.235 5362122.209 354 157.93128 29.266016 2.9266016 2493.259041 851.9297745 234 no 4388624.533 5361657.816 268 759F 291.36564 72.859961 7.2859961 6203.755859 851.4629672 235 no 4389015.551 5361907.67 327 759 135.5766 50.638672 5.0638672 4310.053624 851.138755 236 no 4388550.301 5362226.423 368 306.8016 143.195508 14.3195508 12166.40403 849.6358722 237 no 4388803.502 5361662.173 3 759F 300.84612 107.587305 10.7587305 9135.148689 849.0916925 238 no 4388722.411 5362135.167 358 182.21952 108.92793 10.892793 9238.000901 848.0837652 239 no 4388565.547 5361743.891 77 759D 325.7016 104.510742 10.4510742 8863.357114 848.080967 240 no 4388475.971 5361869.744 125 759X 379.88208 140.103906 14.0103906 11856.95072 846.296942 241 no 4388356.041 5361845.618 315 759X 181.1976 24.184961 2.4184961 2045.540847 845.7904261 242 no 4388484.749 5361559.846 240 759D 39.7734 18.027539 1.8027539 1524.724908 845.7754039 243 no 4388280.92 5361577.534 243 759D 182.5524 86.11582 8.611582 7261.548247 843.2304595 244 no 4388921.5 5362046.651 200 759 63.02112 13.303125 1.3303125 1120.373329 842.1880793 245 no 4388420.69 5361748.999 286 759D 146.87112 66.223438 6.6223438 5575.704373 841.9533237 246 no 4388293.443 5361755.449 287 759D 292.8234 141.603516 14.1603516 11918.85572 841.706199 247 no 4388482.819 5361977.084 179 759X 225.00936 102.714648 10.2714648 8642.900252 841.4476826 248 no 4388488.894 5361580.254 244 759D 300.62808 151.602344 15.1602344 12740.96656 840.4201562 249 no 4388470.198 5361821.418 304 759X 163.10628 50.842773 5.0842773 4262.621495 838.3928026 250 no 4388444.531 5361838.703 109 759X 147.18048 68.017383 6.8017383 5697.33992 837.6299806 251 no 4388556.12 5361828.774 307 759X 123.94452 19.96543 1.996543 1670.823259 836.8581388 252 no 4388520.289 5362240.602 369 309.0726 110.279297 11.0279297 9226.625037 836.6597619 253 no 4388497.079 5361892.164 323 759X 349.59372 76.82168 7.682168 6426.960703 836.6076741 254 no 4389017.762 5361869.928 127 759 316.43904 105.092968 10.5092968 8784.147566 835.845417 255 no 4388403.932 5361753.304 81 759D 285.83904 123.438477 12.3438477 10311.19916 835.331042 256 no 4388983.012 5361923.086 156 759 123.76404 54.209375 5.4209375 4528.124004 835.3027505 257 no

Appendix D : PV potential and expansion pathways 89

4389037.449 5361863.433 121 759 201.81564 91.559961 9.1559961 7644.333595 834.8991756 258 no 4388759.982 5361619.729 254 759F 302.65548 128.979297 12.8979297 10750.58597 833.5125267 259 no 4388785.063 5361974.718 759 52.7976 22.103125 2.2103125 1839.497329 832.2340526 260 no 4388810.828 5362032.871 193 759C 102.06792 42.960156 4.2960156 3572.466764 831.5767671 261 no 4388535.306 5361845.524 314 759X 227.26644 100.600586 10.0600586 8364.664318 831.472723 262 no 4388552.091 5361657.105 382 759D 214.63824 144.379297 14.4379297 11978.60675 829.6623544 263 no 4388438.778 5361530.563 239 759D 430.69452 177.692969 17.7692969 14734.78998 829.2275187 264 no 4388798.634 5362004.256 185 759C 164.76096 72.013477 7.2013477 5969.061918 828.8812271 265 no 4388618.815 5362109.219 210 294.77808 106.294662 10.6294662 8795.943404 827.5056563 266 no 4388321.955 5361708.184 397 759D 548.76792 194.336914 19.4336914 16068.37095 826.8306118 267 no 4388620.046 5361695.262 273 759F 272.7 113.953125 11.3953125 9415.390367 826.2511772 268 no 4388574.105 5361685.476 272 152.56404 63.284375 6.3284375 5224.249143 825.5195919 269 no 4388386.787 5361380.683 222 74.77032 43.42207 4.342207 3578.970354 824.2284059 270 no 4388825.701 5362033.962 198 759C 267.94692 57.992773 5.7992773 4778.357601 823.9574268 271 no 4388581.841 5361697.396 274 46.96872 14.512695 1.4512695 1192.778474 821.8862687 272 no 4388812.163 5361950.232 759 190.98048 73.478711 7.3478711 6021.826471 819.5334934 273 no 4388743.78 5361595.209 248 759F 309.81324 115.833008 11.5833008 9486.825334 819.008804 274 no 4388764.449 5362133.824 211 230.0742 14.579297 1.4579297 1193.690986 818.7575755 275 no 4388893.707 5361805.444 759 310.91256 85.207031 8.5207031 6947.195012 815.3311916 276 no 4388802.186 5361917.848 759 277.6242 29.955664 2.9955664 2437.836115 813.814748 277 no 4388732.311 5362060.69 386 759C 281.73048 55.610156 5.5610156 4524.234346 813.5626065 278 no 4388855.346 5361929.805 759 209.48436 81.507422 8.1507422 6616.314917 811.7438577 279 no 4388967.188 5361921.577 157 759 194.15388 85.149023 8.5149023 6911.863299 811.7372409 280 no 4388628.808 5362084.455 205 202.56324 91.663086 9.1663086 7433.741703 810.9853189 281 no 4388667.56 5361886.978 141 759C 284.48676 89.147266 8.9147266 7213.57026 809.1745921 282 no 4388257.499 5361685.062 58 759D 105.09612 22.176172 2.2176172 1793.342896 808.6800987 283 no 4389028.339 5361815.39 300 759 100.2726 28.221875 2.8221875 2277.042015 806.8358376 284 no 4388759.933 5361935.598 759 249.06096 102.867188 10.2867188 8270.816134 804.0286018 285 no 4388871.334 5361921.044 759 197.75856 64.835547 6.4835547 5206.273139 802.996717 286 no

Appendix D : PV potential and expansion pathways 90

4388434.589 5361815.127 101 759X 234.56016 112.161328 11.2161328 8995.314127 801.9978265 287 no 4388407.49 5361939.804 163 759X 66.53436 11.86582 1.186582 950.4286607 800.9801773 288 no 4388928.291 5361835.92 107 759 176.21016 73.227344 7.3227344 5852.065726 799.1640016 289 no 4388991.876 5361882.723 320 759 177.98436 81.681445 8.1681445 6506.918331 796.621354 290 no 4388961.266 5361964.041 169 759 215.05308 70.844726 7.0844726 5521.036394 779.3150889 291 no 4389022.053 5361925.32 158 759 278.88516 121.176172 12.1176172 9294.527595 767.0260119 292 no 4388842.37 5361933.607 759 235.5492 83.105859 8.3105859 6342.028327 763.1264995 293 no

Appendix E: Maps 91

Appendix E: Maps 92

Appendix E: Maps 93

Appendix E: Maps 94

Appendix E: Maps 95

Appendix E: Maps 96

Appendix E: Maps 97

Appendix E: Maps 98

Appendix E: Maps 99