I construct a spatial model of Portland’s urban forest using high-resolution lidar data. I then use several spatial analysis techniques to search for patterns in the distribution of the urban forest throughout the city, the object of which is to describe Portland’s urban forest through multiple metrics.
Analysis began with two lidar rasters, which were used to construct an elevation model of the city. Lidar functions by firing a laser from an aerial sensor, and recording the surfaces it encounters as it progresses toward the ground. An individual lidar beam will return several times, as it comes into contact with surfaces which partially arrest the beam, but are also partially permeable. In this way, lidar detects an array of surfaces starting from the highest-elevation surface and progressing down to the first impermeable surface. This is typically the ground, or another impermeable object such as a building. By subtracting the last-hit values from the highest-hit values, I control for the topography of the city and effectively flatten it, yielding a model with values which only vary as a result of the differential height of surface objects.
I originally intended for this process to remove buildings from the elevation model, as they appear to be impermeable surfaces which would be subtracted with the last-hit raster. Unfortunately, the last-hit raster which I employed did not include built structures, and I was forced to employ other means to distinguish trees from buildings in my analysis. Several methods were attempted, but ultimately the most useful one was to distinguish trees from buildings by their height in the area. The weakness of this method was that it missed trees which were shorter than buildings, but this nonetheless rendered the data useful.