A 328Kb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
The data you capture is only as good as the applied control
Mobile LiDAR data, on its own, is not what we would classify as "survey accurate"–meaning accurate enough to defend (sign and seal) for engineering design or similar services. Now that I’ve got your attention, that same statement can be said for practically any survey instrument, because the result depends entirely on the standard of care applied during collection and subsequent processing.
For Mobile LiDAR, as with traditional surveying techniques, that standard of care includes referencing captured data to some type of control network to generate the desired results. Individual project specifications typically dictate the level of accuracy that needs to be achieved (whether relative or absolute), but it’s our responsibility to determine and deploy the appropriate techniques, design, and apply that control to produce repeatable and defensible results.
Control Plan: Layout/Design
Detailed "Mission Planning" efforts are a key component of the success of any field survey. Mission Planning efforts for Mobile LiDAR surveys also include a concerted focus on required ground control. Project requirements ultimately drive the implementation of control, but there are a myriad of factors and options that must also be considered before the actual scanning is performed; from desired accuracy to project location, and the numerous factors in between.
The application of Mobile LiDAR technology, and the products and services provided, are incredibly varied. In addition, there are a number of commercial off-the-shelf (COTS) systems available should you have the necessary capital for such an investment. With each application and system come specific needs for control that must be customized by the provider.
The design and layout of a proper control plan needs to address: client requirements or standards, driving speeds, site access restrictions/limitations, sensor orientation, staff safety, localized obstructions (traffic, vegetation, elevated or sunken roadways, etc), style/type of target, terrain, control point spacing, and processing software limitations; all of which can have a dramatic impact on spatial accuracies of the end product.
We like to think that the proper implementation of ground control, especially when you consider all factors (including cost), is more of an art than just a necessity. Sure, anyone that has a basic understanding of the technology and general requirements could develop a control plan, but it’s likely to be either over-engineered (resulting in significantly higher cost), or insufficient to meet the true project requirements; resulting in either additional field visits, or increased processing costs to counteract the deficient ground control. A quality control plan should create efficiencies across the board.
A control plan specialist is an often overlooked and seldom-discussed (sometimes for good reason), integral part of the survey team. Perhaps we’re over glorifying their role; but then again, maybe it’s time to let them take a bow (albeit quickly…we don’t want it to go to their heads). Although no survey activities are occurring at the time of the plans’ conception, the control plan specialist can be thought of as the architect of a carefully orchestrated ballet that hasn’t even started.
Their decisions and interpretation of local conditions could dictate the overall success or failure of the mission. Place too many control points along the route, and you waste critical time (and money) physically establishing the control and applying it to constrain the point-cloud. Carelessly select target locations on a high-volume traffic area and you run the risk of creating unnecessarily dangerous work hazards for the conventional survey team. Neglect the existence of terrain that could infringe on the laser’s line-ofsight to the target, and you might end up with a deliverable that doesn’t meet the clients’ accuracy requirement(s). The number of scenarios and pitfalls are endless. The key is not to overlook this valuable component of Mission Planning. Maybe we’ve gone a little far by propping it up on a pedestal. And perhaps you don’t even have to put it front and center in the spotlight. But you definitely need to pay close attention, and devote the necessary time to ensure a successful mission and satisfied client.
Ground Control Methodology
The approach to establishing ground control is just as varied as the Mobile LiDAR systems on the road. Each project presents accuracy needs which are not always met by simply applying the postprocessed vehicle trajectory to create a point-cloud and resultant product(s). The specific project requirements will almost certainly dictate which survey methodology is applied for establishing horizontal coordinates, as well as an elevation on the targets–if only indirectly.
Real-Time Kinematic GPS
One of the most efficient methods for establishing control is the use of Real-time Kinematic (RTK) GPS. RTK utilizes a base station placed on a known position to broadcast corrections to a roving unit. In the past, the process was a little more complicated than today’s standards due to the range (distance) of the radios used to broadcast the correction. Now, with cellular modems, the baseline distance between base and rover can be greatly extended–further increasing productivity. However, with increased baseline lengths, comes the inherent degradation in the accuracy of RTK GPS, which you’ll witness during post-processing routines of the vehicle’s trajectory utilizing a single base solution.
With the proliferation of Virtual Reference Stations (VRS), the use of a single base scenario is eliminated by using a network of stations. The corrections are modeled over the entire network, providing a greater range without the degradation in accuracy. A noticeable increase in productivity is also achieved as you eliminate the need to set up a base on a daily basis. And in some areas of the country, you’ll actually achieve even greater cost savings by eliminating the "babysitter" whose sole , responsibility is to protect your $20,000+ investment from walking away.
Overall, there are few challenges with RTK, and most can be effectively mitigated, but there are a couple prominent items that need to be considered. The infrequent "bad initialization" can be a killer to field operations. Although bad initializations are not an everyday occurrence, the only way to identify and counteract them is through redundant observations–a common practice we employ to provide a measure of certainty. A second concern, whose effects can additionally be counteracted by employing multiple bases or a larger network, is that RTK does not provide a true measure of accuracy relative to other observations. Each measurement is an independent observation and not part of a closed network. As the GPS constellation changes, the baseline lengths increase/decrease and begin to introduce additional factors into the observations; shifting the relative accuracy of your control with the tide.
Although the use of RTK to establish horizontal control is a valid approach for virtually every scenario (and in many cases for the vertical component as well), roadway design and survey grade solutions require tighter vertical accuracies to yield defensible results. By using differential leveling techniques to establish the vertical component, you eliminate the uncertainties in the Geoid model, as well as the fluctuations in the observed RTK accuracies. Some State Departments of Transportation have gone in so far as to specifically require the combination of RTK and digital leveling for the establishment of control.
The use of digital leveling has advantages over three-wire leveling. First and foremost, like a data collector connected to a total station, you effectively eliminate errors by instantly recording the measurements from the level rod bar code. The instrument operator can effectively balance the foresights and backsights in the field without needing to read multiple stadia lines. Furthermore, the office processing effort occurs much more rapidly following data download than attempting to adjust based on handwritten field notes.
The methodology for establishing Mobile LiDAR control should also take into consideration the re-use of the targets as secondary control for other surveying activities that may need to occur to overcome the few limitations of Mobile LiDAR surveys. Since Mobile LiDAR operates on the basic principle of line-of-sight, there may be localized permanent obstructions where the laser(s) are unable to fix on the desired target, providing the opportunity to utilize the control points to fill in obscured areas, tie in underground utilities, or perform accuracy assessments of the collection.
Application of Control
The term "Application of Control" can be interpreted or defined differently depending on with whom you’re speaking. To a manager or Quality Control/Quality Assurance (QA/QC) specialist, it will likely mean the necessary checks and balances that occur at each stage of the project. For this discussion, we’re referring to how we adjust our Mobile LiDAR data relative to the survey ground control. There are a number of methods for applying control or performing adjustments.
It’s All Relative
Sometimes high absolute accuracy does not matter, but relative accuracy does. The determination of relative underheight clearances for bridge overpasses (i.e. the measured distance from the roadway surface to the low-hanging steel or concrete), only requires the data to have a high relative accuracy between points. In these scenarios, and assuming the laser being utilized possesses sufficient precision between readings deemed acceptable, the "Application of Control" also takes on a relative significance, whereas it’s primarily important to constrain the point-cloud (or collection strips) within itself. Given that there will be errors in the vehicle’s trajectory, areas within the point-cloud that contain overlap from multiple scanning passes collected at different times, will not be coincident. Therefore, common points in each laser "strip" are identified and used to perform spatial adjustments to eliminate "ghosts" (visually observed shifts in the data that present a duplication of the target object). The adjustments can be very minor (fractions of an inch) where there is unlimited visibility to the sky, but occasionally there can be major inaccuracies (inches or feet) caused by prolonged GPS outages.
As mentioned, even with great visibility to the sky above and short baseline lengths during processing, there are errors in the data. The dominate error, as with any system whose positional foundation is GPS, will be in the vertical component–the general rule of thumb is a factor of 3 times the horizontal error. Therefore, spot elevations on large flat surfaces can be utilized to translate the LAS strips up or down to counter the difference, regardless of whether the variance originated from the on-board GPS, or from an equally common (if not more so) localized datum shift.
If survey grade measurements are your goal, more robust control adjustments are required. The processing requires development of tie lines (linking ground control to the point cloud) and ground lines (linking unconstrained data to constrained data). Using an iterative process to apply control and constrain overlapping flight lines, the full adjustment produces a solid framework of points from which comprehensive survey products can be developed.
No matter what we would like to believe, surveying (Mobile LiDAR) is not absolute. There are errors, whether systematic or random, in each measurement. Where the differentiation lies, is in our ability to minimize or effectively eliminate those errors through mission planning, ground control, field collection best practices, processing algorithms and a general appreciation of the work being performed.
Stephen Clancy is a Florida licensed Professional Surveyor and Mapper as well as a Certified GIS Professional. Most recently has been charged with the technical management of Baker’s Mobile LiDAR system.
A 328Kb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE