We will use the LiDAR data imported in the previous recipe as our dataset of choice. We named that table chp07.lidar. To perform a nearest-neighbor search, we will require an index created on the dataset. Spatial indexes, much like ordinary database table indexes, are similar to book indexes insofar as they help us find what we are looking for faster. Ordinarily, such an index-creation step would look like the following (which we won't run this time):
CREATE INDEX chp07_lidar_the_geom_idx ON chp07.lidar USING gist(the_geom);
A 3D index does not perform as quickly as a 2D index for 2D queries, so a CREATE INDEX query defaults to creating a 2D index. In our case, we want to force the gist to apply to all three dimensions, so ...