PE and RS PUBLIC October 2011 : Page-965were returns from architectural remains. A custom filter was then developed using MARS ® that included all ground points and points up to 1.25 m above the ground and the point cloud was reclassified. A digital surface model (DSM) of the results was created at a resolution of 0.25 m (Figure 1). An overlay of the vector map files made during the 2009 and 2010 field seasons of architectural remains on a hillshade of the DSM was done in order to provide an initial check of what level of detail the processed lidar data was showing. The results were amazing as all of the architectural remains that had been mapped were clearly visible in the DSM, and there was clear evidence of small remains that had been missed in the field. These included foundations of domestic dwellings that were 1.5 m x 1.5 m in size and walls of buildings and other structures that were 0.5 m in width and possibly even smaller. Realizing what could be seen in the processed lidar it was decided to create printouts of the DSM hillshade of the whole area at a scale of 1:2500 for the team to take into the field. Since the researchers could clearly see most of the architectural remains on the DSM hillshade, as well as roads and paths that went through the remains, it was decided that the full coverage survey methodology could be modified to make use of the lidar-derived DSM product. Before going into the field, the researchers planned out what paths they would take into the site and what remains they would search for. They loaded the DSM hillshade onto the Trimble® handheld computers and GPS units they were using so that they could use the GPS to navigate through the DSM as they walked through the site. In the field they used the GPS to follow the path to the architectural remains evident on the DSM and verified the location, shape and condition of the architectural remains. They were also able to verify if they could find any architectural remains not evident on the DSM. This new methodology contrasted greatly with the traditional full coverage survey methods that had been used during the previous two field seasons. In each of the previous field seasons an average of 1,250 features were recorded over the course of a 12-week period. However, in 2011, using the lidar DSM hillshade map as a guide, over 3,800 features were mapped during an eight-week period. Both the speed and the accuracy of the mapping were improved using lidar-derived data. This accelerated mapping capacity means that the cost of fieldwork is decreased by a factor of four for the Legacies Project and the team estimates that what would have taken 10 years to map, given the rugged terrain and dense vegetation, can now be mapped in two years. To date, using a combination of archaeological survey, remote sensing, and other techniques, the team has documented, with sub-meter accuracy, over 6,000 architectural features. continued on page 966 Figure 1. Three images illustrating the different ways that the LiDAR data can be visualized from the site of Sacapu Angamuco, Michoacán, Mexico: Top, unfiltered LiDAR point cloud showing the forest canopy visualized as a 3-D image. Middle, Lidar point cloud filtered to show the ground surface (last returns) and prehistoric features, and visualized as a 3-D image. Bottom, hillshade displayed as a 3-D image of the DSM (25 cm post spacing) that was created from the filtered point cloud. Numerous ancient structures, walls, roads, pyramids, and other features are clearly displayed in the middle and bottom images. Photogrammetric engineering & remote SenSing Oct ober 2011 965 Publication List Using a screen reader? Click Here |
