A challenge faced by transportation agencies is the requirement of engineering quality topography during planning. The new design projects demand topographic information with a high vertical accuracy of 1.5 cm to 3 cm or even better and its generation is not only technically challenging but also necessitates additional logistic arrangements like a solid control layout, meticulous flight planning, etc. MA Engineering Consultants with its long history of investing in the best people and technology understands that surveying and mapping are critical elements to all road networks. Very recently, MA Engineering has performed multiple projects using helicopter-based aerial LiDAR supplemented with digital leveling and was able to achieve the vertical accuracy needed for such design level studies in a consistent manner.
A 1.164Mb PDF of this article as it appeared in the magazine is available HERE
Methodology
Understanding that the vertical accuracy requirements of the topographic data for the design level studies requires a new approach, MA Engineering has developed a comprehensive method for meeting the requirements. This multi-faceted approach was designed into seven (7) primary tasks to promote a logical progression of activities and ensure overall completeness. The flowchart shown in Figure 1 provides an overview of identified project tasks described in further detail in subsequent sections.
Control Point Layout
As with any other LiDAR project, the success of any highly accurate topographic project depends on selection, placement, distribution and measurement of accurate ground control points. For all three projects, the control points were painted in pairs to control the entire mapping polygon with the majority of the control set on hard surfaces in order to cover the mapping area. There a few exceptions at locations where hard surfaces did not exist. A permanent marker such as PK nail (painted) or a rebar (non-hard surface) was placed at the surveyed location of the panel points. All control were observed using Digital Levels Method and tied to the State Plane Grid Coordinate System using NAD83 (current adjustment) for horizontal datum and NAVD88 for vertical datum.
As part of the control process, Primary Survey Control (PSC) points were also established at the beginning and end of each project section as well as at intervals along the corridor sufficient for final construction. For each project area, a shape file was created showing all of the control points and check points in the study area which had been used during the accuracy investigations stage.
Data Collection
The low-altitude LiDAR surveys from helicopters provides a way of acquiring engineering grade elevation data. Based on the desired accuracy of the final product, the height, spacing, speed, and point density of the data acquisition were determined and a flight map was prepared. Understanding that the accuracy of the elevation is directly related to the elevation of the system during acquisition, an altitude of 600 ft. above ground level (AGL) was chosen as the flying altitude.
To maintain the required accuracy, the data collection was performed to stay within five miles of the base station at all times. Base stations were established in pairs so that if anything happens to a unit during the mission, the other station can be utilized and the mission does not have to be reflown. The satellite ephemeris are also consulted to determine the position dilution of precision (PDOP) and geometric dilution of precision (GDOP) of the satellite configuration at the project area. Aerial data acquisition was performed when PDOP no higher than 3.0. Two-hour missions were performed during periods of good PDOP.
After the data acquisition was complete, the trajectories were processed to a satisfactory level and the inertial measurement unit (IMU) data was then applied to the trajectory so that the attitude of the helicopter is known at each transmission of light from the LiDAR unit. Adjacent flight lines are flown in opposite directions to provide a quality control check of the point cloud. If there is a time or elevation bias, it can be seen by the comparison of opposing flight lines. Also, cross flight lines were used to provide quality control across all flight lines. For all three projects, the aerial LiDAR data collection flight plan was made with the intention of meeting the final vertical accuracy requirement of 1.5 to 3 cm. The data in all projects was collected with a Riegl Q560 laser scanner at an altitude of 600 ft. AGL by helicopter.
The flight parameters for all projects are given in Table 1.
Calibration and Spatial Constrain of LiDAR Data
After the data was collected, a two-stage data calibration process was performed:
1. Assigning of scanner and trajectory parameters to each scan record
2. “Tie plane” least squares adjustment of the dataset in order to fit all lines to each other.
In the first step, laser data is imported into the Riegl processing software package RiProcess. RiProcess reads the timestamp recorded for every laser pulse and matches it to a precise location along the flight trajectory. The end product is a set of scan lines in the correct location and orientation, but due to small misalignments between the IMU and the laser scanner, as well as other sources of minor interference, further calibration is needed to “tighten up” the dataset before export.
The second step performed with RiProcess is a tie plane adjustment. In this step, an automated process finds planes of at least 10 to 15 points within each scan line and these matching planes are then adjusted by the least squares method to fit each other via roll, pitch, yaw, northing, easting, and height adjustments.
The collected LiDAR data was constrained to highly accurate control points collected using ground-based surveying. The LiDAR data was geometrically corrected in order to account for the potential errors that are inherent in the LiDAR geocoded data either due to calibration, GPS/IMU anomalies, or any other associated issues. Before performing the filtering/classification of LiDAR data, geometric correction of LiDAR data was completed by comparing the LiDAR data with known control points. During this process, the ground control was intersected with the triangular irregular networks (TIN) model of the calibrated point cloud. The Z values were checked against one another, and the difference was calculated. These results were then used to adjust the LiDAR data. A calibrated and constrained LiDAR is
shown in Figure 2.
Classification of LiDAR Data
LiDAR data processing using a three stage process:
1. Generation of bare earth involving the filtering of ground points using an algorithm that considers the geometry of adjacent points as well as parameters set by the user.
2. Interactive editing of the resulting ground surface involving surface visualization tools to reclassify points that were incorrectly classified during automatic processing.
3. Creation of intensity image and extraction of planimetric features using intensity images.
The ground classification process involves building an iterative surface model. This surface model is generated using three main parameters: building size, iteration angle, and iteration distance. The initial model is based on low points being selected by a roaming window with the assumption is that they are ground points. The size of this roaming window is determined by the building size parameter. The low points are triangulated and the remaining points are evaluated and subsequently added to the model if they meet the iteration angle and distance constraints. This process is repeated until no additional points are added within iterations. A second critical parameter is the maximum terrain angle constraint which determines the maximum terrain angle allowed within the classification model. Once the automated classification has finished, the second step is to rely on manual editing of the data to “clean up” artifacts still remaining in the data set which were left by the automated process. A grid generated from ground points is shown in Figure 3.
Accuracy Analysis
After the data processing is completed, the Non Vegetated Vertical Accuracy was computed using check points collected separately for each project site. The vertical accuracy assessment of LiDAR data involves comparison of measured survey checkpoint elevations with those of the corresponding LiDAR point. The determination of LiDAR x, y, z, was performed using TIN based approach. In the TIN based approach, the X/Y locations of the survey checkpoints are overlaid on the TIN and the interpolated Z values of the LiDAR are recorded. These interpolated Z values are then compared with the survey checkpoint Z values and this difference represents the amount of error between the measurements.
After calculating the LiDAR Z values corresponding to check point Z values, the Root Mean Square Error (RMSE) was calculated and the vertical Figure 3: Grid generated from ground points accuracy scores are interpolated from the RMSE value. The RMSE equals the square root of the terrain mapping for use in engineering average of the set of squared differences and planning work. This project was between the dataset coordinate values proof that aerial LiDAR can be more and the coordinate values from the economical and easier to collect detailed survey checkpoints.
Conclusion
The accuracy results conclusively prove that high resolution aerial LiDAR can be used effectively to develop survey grade terrain mapping for use in engineering and planning work. This project wasproof that aerial LiDAR can be more economical and easier to collect detailed data than standard ground-based surveys. Additionally, LiDAR data can be processed quickly and accurately into engineering quality mapping products. Vegetation canopy was not a major factor in this project, but brush vegetation was penetrated well due to the high resolution nature of this collection leading to more accurate topographic mapping. Finally, LiDAR serves multiple uses: providing not only topographic data, but intensity imagery that is very useful for delineating ground features, water bodies, and vegetation.
Dr. Srini Dharmapuri, CP, PMP, GISP is with MA Engineering Consultants (MAEC) in Dulles, VA as Director– Geospatial. Dr. Dharmapuri has Master of Science (Physics), Master of Technology (Remote Sensing), and Doctorate (Satellite Photogrammetry). Dr. Dharmapuri has over 30 years of extensive, wide-ranging experience within the Geospatial industry; most notably with LiDAR, Photogrammetry, and GIS. He has worked in both the private and public sectors, as well as internationally. In addition to his educational achievements, Dr. Dharmapuri is also an ASPRS Certified Photogrammetrist, Certified Mapping Scientist–LiDAR and licensed Photogrammetric Surveyor in South Carolina and Virginia, as well as a Certified GIS Professional and Project Management Professional. Dr. Dharmapuri is actively involved with ASPRS and ASPRS-EGLR. Mr.
Matthew Elious, CP is with MA Engineering Consultants (MAEC) as Photogrammetry Director. Mr. Elious’ professional career spans over 33 years and managed open-end photogrammetric and LiDAR projects for State DOT’s as well as FAA’s WAAS aeronautical and obstruction survey projects.
A 1.164Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE