Northland LiDAR Index Tiles (2018-2020)

Data Aquisition:

The aerial survey program was conpleted over a period between December 2018 and July 2019, with some post QA flights undertaken in February 2020 to fill in data voids.



Survey Specifications:

 Scanner: Trimble AX60i

 Flying Height: 820 m AGL

 Scan Angle: ±60 degrees

 Scan Frequency: 200 Hz
 Pulse Rate: 300 kHz

 Swath Overlap: 60%

 Swath Points Per M2: 4 in nominated forestry areas and 3 across the rest of the dataset

RPS used the Trimble AX60i scanner in order to carry out the LiDAR survey. The Trimble AX60i is a high performance, versatile, and fully integrated airborne LiDAR solution. It integrates a powerful long-range laser operating at 400 kHz pulse return rate with a maximum scan frequency of 200 Hz. Beam deflection is achieved through a rotating mirror instead of the common oscillating mirror, resulting in parallel scan lines on the ground with uniform point-spacing and high accuracy within a maximum of 60 degree field-of-view. With the capability of detecting theoretically unlimited returns for each outbound laser pulse, the AX60i is ideal for surveying dense vegetation areas and complex terrain.

Data Processing:

For each LiDAR swath the point cloud was generated from the full wave form instrument data and post processed GPS/IMU logs. Each sortie was validated against check points to quantify residual errors.

The cross strips and LiDAR swaths form a network of overlapping data that was analysed and levelled to establish the internal integrity of the dataset using TopoSys software. The survey swaths were adjusted to the cross strips in increments of 1cm, while viewed in profile. Several profiles were measured in each intersecting block and adjustments were averaged along the swath. No automatic adjustments were applied as this can introduce errors (by polynomial warping) and obscure systematic errors.

The vertical accuracy of the LiDAR was tested against control survey. If necessary a vertical adjustment of the LiDAR can be applied at this point, however, the vertical accuracy assessment result was within the specifications of 0.05m therefore no adjustment to control was applied.

The tiles were automatically classified to ground and non-ground (classes 2 and 1, respectively) using Terrascan software. Low and high noise points were automatically classified and allocated to class 7 prior to review, where low noise was identified below the ground level, high noise identified as points greater than 100m above ground.

The classified LiDAR ground points were improved to ICSM Level 2 specification by manual editing in TerraScan software. This involved reclassifying points from or to the Ground class, so that confirmed ground points were in the Ground class. This was done with the point cloud displayed in profile view and the terrain surface (TIN) in plan view. The ‘above ground’ points were then automatically classified as per the ICSM specification Level 1 using classes specified by Northland Regional Council.

The LiDAR point cloud for each survey strip was generated from the full waveform LiDAR signal and translated from the temporal/angular units to geographic coordinates by reference to the calculated flight trajectories.

Residual errors were measured using the “boot control” points at the start and end of each sortie. This process provided an initial quality check and an opportunity to enhance the accuracy of the LiDAR data.

The LiDAR strips were levelled and combined, and then adjusted to ground control to produce a set of LAS files with the required spatial accuracy. The process provided a high level of confidence in the internal integrity of LiDAR heights, and a sound assessment of the quality of GPS/IMU data.

All data was processed without any major issues.

Raw LiDAR strips are checked visually to ensure there is complete coverage of the areas surveyed and there are no obvious discrepancies. More thorough visual assessments are undertaken during processes such as levelling, classification and quality assurance of classified tiles. During processing any voids that are not due to water or non-reflect surfaces is investigated. In our experience the only situation in which such voids occur is where there is no overlap between adjacent strips, in which case the missing data will be reflown.

Once the LiDAR for an AOI has been internally levelled, TerraScan software is used to calculate residual error reports for the complete set of ground control points. The vertical accuracy of the point cloud dataset is tested using a TIN surface constructed from bare-earth LiDAR points, which is compared against ground survey locations.

RPS obtained ground survey from Opus surveyors to evaluate the vertical accuracy of the LiDAR data. The comparison of these heights to the ellipsoidal heights will be used to give a measure of the quality of the LiDAR ellipsoidal heights.

The residual error reports list the location of each check point and the differences between the measure z values and the interpolated LiDAR ground surface.

Data for the entire project area was processed as a whole to ensure vertical consistency.

Once classification was complete, the data was regenerated to 1km x 1km tiles in LAS v1.2.

Classification of the point cloud followed the classification scheme below:
1 - Unclassified
2 - Ground
3 - Low Vegetation
4 - Medium Vegetation
5 - High Vegetation
6 - Buildings
7 - Low Noise
9 - Water
10 - Rail
13 - Wire - Guard (Shield)
17 - Bridge Deck
18 - High Noise

Ground survey data was collected by Opus (now WSP NZ Ltd) surveyors to collect information for measuring the fundamental accuracy of the LiDAR. The standard deviation statistic is 0.056m, a RMS of 0.062 and the average difference is 0.049.

Residual errors (dz) were calculated as the difference between the height of fundamental check points (Known Z) and the interpolated LiDAR ground surfaces using Terrascan software. Ellipsoidal Heights were used for all the comparisons. The LiDAR point cloud and derived products were not adjusted to the control.

The Fundamental Vertical and Horizontal Accuracy Reports were generated from the Terrascan residual error report for the entire AOI and is supplied as a Microsoft Excel spreadsheet and a text file.

This dataset is comprised of tiles from multiple data supplies from RPS to build a best fit product. As a result, there are some inconsistencies in the point cloud classification across tiles. LINZ has also fixed the Z scale factor for a portion of the data in eight pointcloud tiles:

- AW29_1000_4903
- AW29_1000_4904
- AW29_1000_4905
- AW29_1000_5003
- AW29_1000_5004
- AW29_1000_5005
- AY30_1000_2444
- AY30_1000_2544

The deliverables to LINZ were:

1m gridded bare earth digital elevation model (DEM)
1m gridded digital surface model (DSM)
Classified point cloud

All product deliverables supplied in terms of NZTM map projection and NZVD2016 vertical datum.