The i-Tree Canopy Project is designed to use Google Maps, with a satellite image of Cambridge, to determine Cambridge’s tree canopy health. The i-Tree technology separated Cambridge into its 13 neighborhoods. Each of Cambridge's 13 neighborhoods was given a certain number of “points," which the volunteer who worked on that neighborhood had to classify. The number of points for each neighborhood correlated with that neighborhood's area. For instance, the volunteer who worked on Neighborhood 1, East Cambridge, was required by the technology to complete 902 points. These points were individually yet randomly selected by the i-Tree technology, and a volunteer working on a neighborhood had to classify each point as an “Impermeable Surface” (like turf or pavement), a “Permeable Surface” (like grass or dirt), a “Building,” or part of Cambridge's “Tree Canopy.” For each neighborhood, therefore, we were left with a percentage of the tree canopy. While the percentage is an estimate, a comparison between our data and other tree canopy data (like the UVM LiDAR study, or MIT’s TreePedia) will allow for a greater understanding of not only our own data but also the health of Cambridge’s trees. Using our data, Green Cambridge can reassess its goals for creating a healthy tree canopy because the data shows how much we have left to do.
Here is an image of the i-Tree technology. As described in About, the i-Tree technology selects a random point, and whoever is working on that neighborhood must classify that point given four options. In the image above, the volunteer working on this neighborhood (Neighborhood 10 - West Cambridge) determined that the portion selected by the point was part of a "tree canopy."
This data shows the number of points needed for each neighborhood to determine "completion." The number of points for each neighborhood depends on the area of the neighborhood, such that the neighborhood with the greatest area has the most points. Note that some areas were "overcompleted," meaning that more points were completed than necessary (which is why the percent completion is above 100%). "Overcompletion" is fine; it does not affect our data. In fact, the more points we complete, the more accurate our data becomes because completing more points lowers the standard error.
This image shows the percent cover of "permeable surfaces," "impermeable surfaces," "tree canopy," and "buildings" in each of the 13 Cambridge neighborhoods. The SE +/- values are the 'standard error' values. A greater number of points means a more accurate estimation, which decreases the SE value. The average values are just an average of the 13 data points, but the weighted average takes into account the area of the 13 neighborhoods.
If you're interested in anything like this project, MIT recently created their own tree canopy inventory. Here is the link to their website. The University of Vermont used LiDAR technology to map Cambridge's tree canopy. Here is a link to their report.