MIT maps solar potential of Cambridge, Massachusetts with record accuracy
June 5, 2013
MIT researchers have developed a new technique that can be used to accurately predict the annual yield of a photovoltaic solar array located anywhere on the planet, taking into account local climate, panel orientation, and obstructions from nearby buildings. As a proof of concept, the scientists have mapped out the 17,000 rooftops of Cambridge, Massachusetts and created a user-friendly web interface that residents can use to look up their homes and get an accurate projection of the cost and return on investment of placing a PV panel over their heads.
Introducing ... the Solar System
The software, called "Solar System," is the brainchild of the Mapdwell Project, a collaborative effort of academics, engineers, designers and computer scientists that strives to collect research-based information on sustainable technology and present them to the public through a user-friendly interface.
Solar System takes data on photovoltaic panel potential and overlays it on top of Google satellite imagery. Selecting a building within the map will bring up a host of information on installing PV panels on that specific rooftop, including the system size in kilowatts, the projected cost to the owner (including deductions and local incentives), the carbon offset, and the projected yearly revenue generated by the array.
Users can also design their own system by either adjusting its size with a slider (in which case the software returns the best arrangement of the panels for the desired array size) or by drawing an area on the map (in which case the software will return information on the projected efficiency of the system).
Just how accurate is it?
To create the map, the researchers combined Google's satellite imagery with LiDAR (Light Detection and Ranging) data from a prior aerial survey, and in doing so they were able to create a 3D model of the entire city. Using the model, the researchers knew the precise shape of rooftops and the possible obstructions – trees, nearby buildings, and so on. They combined this information with local weather station data, and, crunching the numbers in computer simulation package Radiance, were able to map the solar potential of the rooftops throughout the year, and overlay the data on the aerial and satellite imagery.
Solar Systems is not the first solar map to have been created, but it is probably the most accurate. Analogous solar maps in other US cities usually make one of two mistakes: they either assume that every rooftop is completely flat, or that the ratio between direct and diffuse solar irradiation is fixed throughout the year. Unfortunately, both approximations can lead to inaccuracies.
MIT's map, by contrast, accounts for hourly solar radiation for a typical year, then translates that data into an actual sky condition for each hour of the year. The system, which admittedly required a lot more number-crunching to produce than the previous approaches, is however surprisingly accurate: when tested, its predictions have consistently fallen between 4 and 10 percent of real-world results.
In its current iteration, the estimates may be a bit on the cautious side. As of now, the data assumes that the solar panels are being placed parallel to the roof, but this is not always the case because flat-roofed buildings often have PV panels installed at a 45 degree tilt toward the south to improve efficiency. Future version of the map will take this into account, adding an additional data layer on top of flat-roofed buildings that will provide a more realistic estimate.
Even with its slightly pessimistic estimate, the tool shows that if photovoltaic panels were installed on all of the locations classified as either "good" or "excellent" on the map, the city of Cambridge could generate about one third of its energy needs by means of solar panels, for an investment of US$2.8 billion that would certainly pay off in the years to come.
The video below walks you through the features of the "Solar System" project.
Just enter your friends and your email address into the form below
For multiple addresses, separate each with a comma