From The Ground Up: LiDAR and photogrammetry
Together, LiDAR and traditional photogrammetry play a role in most large mapping projects today. LiDAR has evolved into an accurate, cost-effective method of generating the mass points used to create a digital elevation model. The increases in the accuracy of LiDAR sensors have been impressive, if not astonishing, over the last couple of years. Today’s sensor technology allows us to place points on the ground at very dense postings and achieve extremely accurate elevations for these points.
Photography is still necessary, however, to generate breaklines at abrupt changes in the terrain. These breaklines are normally collected from photography at the edges of roadways and water bodies; at the tops and toes of slopes; along headwalls and retaining walls; and other places in the landscape with sudden changes in elevation. Stereo photography is also used for planimetric feature capture common in large-scale mapping. Typical planimetric features collected in mapping include the edges of roadways, building footprints, driveways, fences, utility poles, group vegetation, sidewalks and drainage features that are easily seen in project photography.
Photography also serves as the base for digital orthophotography. Digital orthophotos are generated for most mapping projects today to supplement topographic and planimetric mapping. They can be a significant source of information during planning and design, and serve as a useful tool for communicating with a lay audience that likely doesn’t fully understand more traditional maps (such as at public hearings).
But are breaklines always required for topographic mapping? As data postings from LiDAR become denser, can mass points without breaklines provide an accurate model of the ground? Data postings from the LiDAR collection used for large-scale projects collected from fixed wing aircraft (airplanes) normally range from one to four points per square meter. Rotary wing aircraft (helicopters) can collect significantly more dense data, but unit costs compared to fixed wing are considerably higher due to the slower speeds and narrower swath widths common with this collection. Point densities of 80 or more points per square meter are common with helicopter-based units. At these densities, breaklines provide very little additional information to a digital elevation model. But helicopter-based applications are normally constrained to specialized applications because of the higher unit costs and not employed on typical large block projects like city/county or statewide mapping.
We will continue to collect a higher density of data as the LiDAR sensors continue to improve. We may get to the point where breaklines are not as necessary due to the increased density of the mass points available from fixed wing collection, but for most projects, we are not quite there yet.
Lidargrammetry is an interesting development for breakline collection and the compilation of some planimetric features without the use of photography. This new application of LiDAR technology refers to the creation of stereopairs from LiDAR intensity models that can subsequently be used in the soft copy photogrammetric collection. It works well in some areas of the country, particularly in areas with significant terrain relief that is covered by dense post spacings within the LiDAR acquisition. But it does not provide the same accuracy available from low-altitude photo flights (mainly due to the relative coarseness of the LiDAR coverage) and, therefore, its application is somewhat limited depending on the scope of the project.
There are some reasonably good methods and software for doing automated feature extraction from LiDAR data, but the accuracy and detail expected for most large-scale planimetric maps in terms of common features contained in the maps (roads, driveways, building footprints, sidewalks, manholes, curb inlets, utility poles, fences, group vegetation, etc.) require traditional mapping approaches from stereophotography.
Finally, LiDAR does not provide imagery, although the intensity of the returns or the ranges of elevation from LiDAR are often rendered to simulate imagery. Most units in use today have the ability to measure the amount of energy returned from each outgoing laser pulse, which is commonly referred to as the intensity, when the laser pulses are reflected off the ground. The intensity of a return off asphalt is very different from a return off concrete. Concrete is a highly reflective surface while the tar in asphalt absorbs quite a bit of the energy from the laser. Similarly, vegetation has a different reflectance when compared to bare ground. Water typically absorbs all of the energy from the laser. But intensity images don’t provide nearly the same information due to the limited range of data and the relative coarseness of the data postings.
Simultaneous Collection?I am asked quite often why LiDAR and imagery are not collected together on most projects. That too is a great question. If they were collected together it would seem to save considerable money during the collection. Only one aerial platform would be mobilized to the site and total flying time over the project would be reduced. Flying has certainly become more expensive with today’s fuel prices and the significant increases in insurance premiums for most aircraft after Sept. 11. Flying with two sensors requires two holes in the floor of the aircraft. Several platforms flown by aerial acquisition firms today have that capability.
But the challenges or issues with simultaneous collection generally preclude its application on most projects. Aerial photography depends on the sun to provide illumination of the ground for the imagery collection. Therefore, aerial cameras are known as passive sensors. Conversely, LiDAR is an active sensor. The laser is used to illuminate the ground. Because of this, LiDAR can be flown 24 hours a day. In fact, the majority of our LiDAR acquisition takes place outside the normal window for aerial photo collection. For a couple of our recent projects flown in the vicinity of Atlanta and Newark airports, collection at night when commercial air traffic was at a minimum was the only real option for data collection. In general terms, trying to collect the imagery and LiDAR together would result in a loss of anywhere from 21 to 16 hours (depending on the time of the season and the span of appropriate sun angles for imagery collection) of otherwise perfect collection time for LiDAR. When flying sensors that generally cost $1 million or more, it is important to maximize their use to keep unit costs down.
Secondly, the flying heights necessary for imagery and LiDAR collection are often different. Most LiDAR collection for 2-foot contour mapping is flown around 5,000 to 5,500 feet above ground. Two-foot contours are most often paired with 100' scale planimetric mapping and digital orthophotos at a 0.5-foot resolution or ground sample distance (GSD). Imagery collection to provide this resolution and planimetric accuracy is generally flown around 4,200 feet above ground. Moreover, the line spacing between parallel flight lines flown back and forth over a project is very different for LiDAR and imagery collection. At 4,200 feet above ground, the line spacing for imagery collection (assuming a typical large format camera) is 4,410 feet. At 5,000 feet above ground with a 30-degree field of view, the line spacing for LiDAR collection is about 1,900 feet. Similar arguments could be made for 1-foot contours paired with 50' scale mapping and 0.25-foot GSD orthophotos. The significant differences in flying height and line spacing provide a real challenge for simultaneous collection.