From the Ground Up: Collecting Breaklines
Typical breaklines include ridges; tops and toes of slopes; double-line drains for larger streams and rivers; single-line drains for small streams; edges of roadways; closed polygons for ponds and lakes; open polygons for shorelines; headwalls and abutments; and any other feature that produces an abrupt change or discontinuity in the elevation surface. As an example, picture a gentle slope leading down to a lake. The smoothness of the bank is abruptly interrupted by the water where the elevation surface immediately becomes flat. The water’s edge would be represented as a breakline in the most accurate digital representation of the true elevation surface.
But breaklines do more than add accuracy to a terrain model or allow for smoother, more continuous models. Breaklines are also used in automated mapping methods to remove unwanted elevation points from the elevation model collected by LiDAR. For example, deep, clear water will absorb all of the energy from the laser, thereby eliminating water returns. By contrast, turbid water with considerable suspended sediment or water with a rough surface from wind or wave action will normally provide a weak return and resultant false elevation points in the water. If directionality (the direction in which breaklines are captured*) is carefully considered around water features, automated routines to remove all LiDAR elevation points inside the breaklines from the bare-earth model can be used in a batch process.
There are various methods for collecting breaklines. Here I will discuss a few of those methods, including their advantages and drawbacks.
Breakline Collection from ImageryMost projects allow for a number of options for breakline collection. Many mapping assignments include a requirement for the mapping of planimetric features in addition to the creation of an elevation model. Planimetric features often include building footprints; transportation features such as roads, bridges, railroads and trails; and utility features such as power poles, manholes and curb-box inlets. Traditional stereo imagery is required to compile these planimetric features. Moreover, digital orthophotography, or image-based mapping, is a requirement for many of the projects completed today. And controlled aerial imagery is necessary for both the production of digital orthophotos and the creation of stereo images for planimetric capture. Imagery captured for either of these project requirements will also be available for breakline collection.
Two options are often employed in breakline collection from stereo imagery. The most accurate and robust is to collect 3D breaklines along all abrupt changes in the elevation of the landscape. In this case, the photography must be captured at an altitude appropriate to meet the requirements of both the horizontal and vertical accuracies for the project. Mapping technicians create a breakline by collecting a series of 3D points that approximate the abrupt change in the elevation surface from their visual interpretation of the stereo imagery.
The second option is to collect 2D breaklines from the stereo imagery. Only the horizontal location of the breakline is collected in the stereo environment, again through visual interpretation, and the elevations along the breakline are subsequently populated from an independent data source, typically LiDAR. While this solution does not provide the same level of accuracy as the 3D collection, it does have its advantages. The main advantage is cost. The imagery can normally be acquired at a higher elevation when compared to imagery planned for a 3D collection, and only the horizontal accuracy of the breakline’s location must be met since the elevation will be populated from a different data source. And higher collection altitudes result in lower project costs.
In addition to a somewhat lower level of accuracy, there is another drawback to this 2D methodology: a loss of some QA/QC assessment. When 3D breaklines are collected from imagery, mapping professionals have elevation data from two independent sources: mass points from LiDAR capture and breaklines from the imagery. The two are controlled separately and, therefore, provide a good overall indication of the mapping accuracy when they fit closely. If there are problems bringing these two elevations sources together, additional project checks are required to resolve the discrepancies. This additional QA/QC step is lost with 2D breakline collection because there is only one elevation source.
Breakline Collection from LiDAR IntensityA new breakline collection methodology gaining popularity makes use of additional data commonly collected during the LiDAR mission. Most sensors collect and record intensity data at the same time as elevation data. During LiDAR collection, a laser is pulsed from the aircraft to the ground, and the time of travel is used to accurately determine the distance from the LiDAR sensor to the ground and back, much like an EDM or total station does on the ground. At the same time, the intensity of the returned pulse provides significant information about the ground surface illuminated by the laser. For example, asphalt has a very different intensity compared to concrete. The tar in the asphalt absorbs much more of the light from the laser and, therefore, is recorded as a much weaker intensity. Deciduous trees are different from coniferous; and paint striping along roadways, parking lots and runways is a very highly reflective surface and appears as a very high intensity.
Since we have an extremely accurate location of where the laser was pointed at all times during collection, individual intensities can be brought together to render an intensity surface that is orthometrically correct. The science of photogrammetry provides accurate mathematical models to derive 3D information from two controlled images taken from different perspectives in the air when they are viewed together. When the left eye views an area on the ground in one image, the right eye views the same area in another image or from a different perspective. This gives us the ability to see in stereo--to view the third dimension, or elevation, of objects on the ground. If we use these same mathematical models in reverse with the intensity surface, we have the ability to generate multiple overlapping images of the ground surface.
These images are created as if taken from different perspectives in the air and can be loaded in the same office stereoplotters that are used with more traditional mapping imagery. This allows mapping technicians to view the elevation surface in stereo, map the same abrupt changes in this surface and create breaklines just as if imagery was collected as part of the project. The process of generating these stereo pairs from the LiDAR intensity and the subsequent collection of features from this imagery is commonly referred to as LiDARgrammetry.
The obvious advantage to this approach is cost. Separate imagery flights are not necessary since the mapping images are “created” from the LiDAR intensity surface. The control for these intensity images is created from the geometry of the LiDAR collection. Therefore, additional photo-control efforts are not required in the field, which is effective in managing project costs. Project timing is also improved since the intensity images can be created at the same time that LiDAR processing of the elevation surface is taking place.
There are, however, disadvantages to this approach. First, any advantages in the QA/QC process that imagery provides us are completely lost. Any systematic errors in the LiDAR acquisition will apply equally to the intensity images since they were created directly from the LiDAR data. Also, the resolution of the imagery generated from the LiDAR collection is more coarse. It is common to have point densities of 1 to 3 points per square meter in today’s LiDAR collections used to generate 1- to 2-foot contours. This allows technicians to generate intensity images with a resolution of 1 to 3 pixels per square meter--significantly more coarse than imagery used in mapping. Photogrammetric imagery captured at a resolution of one-half foot is equivalent to 43 pixels per square meter; similar imagery captured at a resolution of one-quarter foot is equivalent to 172 pixels per square meter. Mapping imagery normally provides significantly more information in terms of ground resolution, and, in this case, the increased resolution provides better abilities to visualize and collect accurate breaklines.
Whether mapping or LiDAR is used, breaklines serve an important role in the digital representation of most elevation surfaces. They improve the accuracy of digital terrain models in areas with abrupt changes. They allow the generation of more aesthetically pleasing contour lines compared to contours generated solely from LiDAR mass points. And finally, they allow for the use of automated routines to remove unwanted points in an elevation model, such as points that fall within water features.
* Closed water bodies are collected in a clockwise fashion. Automated routines then clear out all points inside the closed polygon (as long as it was done in a clockwise manner). Roads and double-line drains are collected in a flow, or traffic, direction the same way a car would be traveling--up the right side and down the left. Automated routines then clear out everything between the two lines as long as they are consistently collected in that fashion.