Estimating the rooftop PV potential of a large-scale geographical region

Background

In many studies, scientists have determined methods of obtaining of Building-Integrated Photovoltaic (BIPV) potential as a function of the available roof area. However, the roof area is typically left as a variable, as there is no direct roof area data available in most regions of the world. Various modelling techniques can determine available roof area for a sample consisting of a few buildings, or perhaps a university campus. However, to estimate BIPV potential for a large region, these techniques are too labour-intensive. It is thought, however, that this information can be approximated on a large scale using data on urban area, building and population density, in combination with analysis of air photos or satellite images of urban regions.

This has been completed by Izquierdo, Rodrigues and Fueyo at the University of Zaragoza and LITEC in Spain (article available below). The purpose of this page is to attempt to customize their methodology to be applied to the South Eastern Region of Ontario, Canada.

This work is being done in conjunction with the Queeen's Institute for Energy and Environment Policy.

Literature Review

Part I: Determining Rooftop Area (GIS)

Izquierdo, S., Rodrigues, M., & Fueyo, N., "A method for estimating the geographical distribution of the available roof surface area for large-scale photovoltaic energy-potential evaluations" Solar Energy, 82, 929-939, 2008.

Summary:This paper presents methodology for determining roof area available for solar photovoltaic applications on a large geographic scale (the country of Spain). The method involves stratifying geographic units based on population density and building density. They consider only the urban classified areas of the country. Then, a statistically representative sample of roof area is taken for each category using Geographic Information Science (GIS) techniques and a specific plug-in for Google Earth. Researchers were also able to quantify error. Finally, various reduction coefficients were applied to the roof area to account for shading, void spaces and alternative uses. The final estimate of available roof area for solar photovoltaic applications in Spain is 14.0 +/- 4.5 m2/capita.

Castro, M. et al. 2005. “Grid-connected PV buildings: analysis of future scenarios with an example of Southern Spain.” Solar Energy 79:86-95.

This paper also focuses on the solar photovoltaic potential in Spain, but in a future context. The authors have developed a methodology to experiment with future scenarios to determine overall progress in grid-connected building-integrated photovoltaics (BIPV), called the “ScenariosPV” software. The model encorporates the global projection of markets for solar cells, as well as the associated price reductions and emissions savings. Thus, it considers the economic and social issues surrounding BIPV. However, as with most solar potential analyses, it leaves roof area as a variable.

Kraines, S. B., & Wallace, D. R. (2003). "Urban sustainability technology evaluation in a distributed object-based modeling environment." Computers, Environment and Urban Systems, 27(2), 143-161.

This paper describes an internet-based infrastructure called DOME (distributed object-based modeling environment) for simulating integrated sustainable technologies in a large system, ie. an urban region. This infrastructure can serve to integrate Geographic Information Systems (GIS) with other environmental analysis tools such as life-cycle analysis, input-output analysis, and technology process modeling.
It is proposed that this infrastructure can be used for a wide variety of decision-making. Most importantly, however, it outlines an example of using DOME to determine the feasibility of using rooftop PV modules for electric power generation in Tokyo (Section 5). Using information from the 1991 Tokyo GIS database, the researchers simulated PV deployment on all available roof space as well as partial implementation scenarios.

Kraines, S. B., Wallace, D. R., Iwafune, Y., Yoshida, Y., Aramaki, T., Kato, K., et al. (2001). An Integrated Computational Infrastructure for a Virtual Tokyo: Concepts and Examples. Journal of Industrial Ecology, 5(1), 35-54. doi: 10.1162/108819801753358490.

This paper is related to the paper by Kraines et. al. which describes the DOME integrated software for coordinating several computational processes in order to generate a thorough analysis of a project to reduce emissions. This paper explains the DOME software, and most importantly, demonstrates how it can be used with the example of evaluating large-scale deployment of solar photovoltaics in Tokyo.

The analysis includes finding roof area with a GIS-based Land-Use Aggregator. It draws on a database which is constructed using total planar area of each of 30 land use types (15 of which are building types) in Tokyo. The Aggregator is implemented as a DOME plug-in with a Java-based graphic user interface (GUI). Some interesting and potentially useful assumptions were made in the process. Other relevant sections are the Rooftop Irradiation Profile Model and the PV Generation Model.

follow references for pt II

Aramaki T, Sugimoto R, Hanaki K, & Matsuo T. (2001, March 1). Evaluation of appropriate system for reclaimed wastewater reuse in eacharea of Tokyo using GIS-based water balance model. Water Science & Technology 43 (5), 301–308.

This article is referenced by Kraines et. al. with respect to how to construct the database for the DOME land-use applications from the GIS data. This construction was originally done in Tokyo for this paper, which spatially evaluates different wastewater reuse systems. They use raster GIS data in the form of floor area, rooftop area, and rainfall data. Unfortunately, this paper reveals that the Bureau of Urban Planning in Tokyo possesses this data, and thus they did not develop an innovative way to extract this data.

Rylatt, M., Gadsden, S., & Lomas, K. (2001). GIS-based decision support for solar energy planning in urban environments. Computers,Environment and Urban Systems, 25(6), 579-603. doi: doi: DOI: 10.1016/S0198-9715(00)00032-6.

This paper discusses a solar energy planning system which has been developed for use by planners and energy advisers. The system is scalable from individual property analyses up to entire cities. Importantly, the model is linked to a customized geographical information system (GIS). The paper is specifically focused on solar thermal generation; however it is acknowledged that the method could be shifted to both solar photovoltaics and passive solar applications. The paper uses the Solar Energy Planning (SEP) model. It also describes other related projects, including Energy and Environment Prediction (EEP) and Potential for Renewable Energy in Cities (PRECis). These models are generally set up to determine energy demand and whether the proposed technology (solar thermal) can meet the demand. However, the methods may still be applicable to the needs of the RER project. To briefly summarize,

SEP
- fully uses digital urban maps to visualize at different levels and to derive information from GIS
- conducts analysis at the level of the individual dwelling unit but permits coarser models
- uses building age and type to classify buildings; discriminates based on income
- gives feedback on level of confidence

EEP
- statistical clustering method; divided at the level of postal code areas
- estimates the Standard Assessment Procedure ratings shared among members of a cluster
- uses 4 variables related to built form while all others are approximated based on global statistics

PRECis
- most coarse-grained approach to broad urban areas
- uses novel image processing techniques using digital elevation models (DEM) and Lighting and Thermal (LT) data
- does not relate to georeferenced objects in GIS
- was conducted for only non-domestic buildings

It is felt that EEP and PRECis, not SEP, should be further investigated for the RER project.

Jones, P.J., Vaughan, N.D., Cooke, P., Sutcliffe, A. An energy and environmental prediction model for cities. In Brandon, P.S., Lombardi, P.L. and Bentivegna, V (Ed.), Evaluation of the Built Environment for Sustainability. Accessed at: http://books.google.ca/books?hl=en&lr=&id=iHGIlMD08n0C&oi=fnd&pg=PA53&dq=Energy+and+Environmental+prediction+model&ots=BKWWDXgI3u&sig=-WeJAUVACAw-GAqRrzx_dMG1V0U#PPP1,M1 June 10 2009.

This article describes the EEP model which is used to predict energy use and emissions for different scenarios of urban built form. While it does use GIS-related built form analyses, they are focused around predicting flows of people and energy and do not relate to roof area.

Grosso, M. (1998). Urban form and renewable energy potential. Renewable Energy, 15(1-4), 331-336. doi: 10.1016/S0960-1481(98)00182-7.

This paper outlines the objectives, work content and expected outcomes of the PRECis initative. It also describes the people and institutions responsible for the content. It also describes the basics of the relationship between built form and climate/microclimate relating to solar and wind considerations.

Ratti, C. & Richens, P. (1999) Urban Texture Analysis with Image Processing Techniques. Computers in Building: Proceedings of the CAADFutures '99 Conference, Atlanta, 7-8 June 1999. Accessed at: http://senseable.mit.edu/papers/

Partly funded by PRECis, this conference paper describes an image processing method which was developed involving the use of NIH Image (Mac software, the Windows equivalent is Image/J) and the Image Processing Toolbox in Matlab for urban analysis. It describes various operations that can be done using digital elevation models (DEMs) of urban form, which are a form of raster data. Important techniques discussed include:
- LUT (look-up table) modification, which can enable thresholding and density slicing
- Applying filters, which can remove noise from an image, smooth or blur an image, or even detect edges of buildings

These researchers were able to find the built area using Image without the use of Matlab. They also developed a shadow casting algorithm and sky view factors (amount of time exposed to the sun). More high level applications were possible with the use of Matlab.

follow references for pt II

Guindon, B., Ying Zhang, & Dillabaugh, C. (2004). Landsat urban mapping based on a combined spectral-spatial methodology. In Geoscience and Remote Sensing Symposium, 2004. IGARSS '04. Proceedings. 2004 IEEE International (Vol. 2, pp. 1080-1083 vol.2).

This paper describes methodology for monitoring urban growth in Canada using Landsat images (satellite imaging). The outcome of the analysis is to classify land into 5 classes: forest, water, herbaceous, residential and commercial/industrial. This is particularly important in Canada due to the recent explosion of low-density residential area (ie. suburbs) that is difficult to classify.

Once having obtained classifications, the researchers used six test sites sized 7.5 x 7.5 km to determine the accuracy of their results. They determined building density through manual image counts and population density from census data. They then compared residential classification components obtained with these data. Poor correlation was generally found, at best, the relationship was very complex. This suggests that Landsat classification methods are not directly applicable to the RER project; however, the methods used to test the image processing results do have relevance.