AAFC – Multiple Sites - Saskatchewan
LiDAR Survey Report
March, 2012
TABLE OF CONTENTS
1. SUMMARY……...……………………………………………………………………....1
2. MATRIX LiDAR SYSTEM……………………………………………………….........2
2.1 MATRIX Installation…………………...…………………………………............2
2.2 IMU-GPS Antenna Offset Survey………………………………………………...4
2.3 IMU-Laser Misalignment…………………………………………………………4
3. GPS SURVEY CONTROL…...………………………………………………………....5
3.1 GPS Control Points……………....……………………………………………….5
4. DATA COLLECTION…………………………………………………………………8
5. GROUND CHECKPOINTS……………………….…………………………………. 10
6. DATA PROCESSING……………………………..…………………………….……..13
6.1 LiDAR Point Clouds……………………………………………………….……13
6.1.1 LiDAR Tiles……………………………………………………………….13
6.1.2 Grounds Points……………………………………………………………..13
6.1.3 DTM Key Points…………………………………………………………...14
6.1.4 Vegetation………………………………………………………………….15
6.2 Grid Points……………………………………………………………………….15
6.3 Hillshades……………………………………………………………………...…16
6.4 Orthorectified Imagery…………………………………………………………..16
6.5 LiDAR Contours…………………………………………………………….…..17
APPENDIX A – GPS NETWORKS…………………………………………………..19
APPENDIX B – CONTROL PHOTOS…………………………………………….....23
1
1. SUMMARY
LiDAR Services International (LSI), a Calgary-based LiDAR company completed an airborne
LiDAR survey for Agriculture and Agri-Foods Canada (AAFC) in October-November 2011. The
Fall 2011 portion of the project involved collection of LiDAR data for the Pheasant Creek,
Roughbark, Moosomin, Braddock, Maple Creek, Eastend and Altawan project sites in Southern
Saskatchewan. LiDAR data was successfully collected, processed and delivered with the following
conditions:
LiDAR system installed in a Cessna 185 airplane owned and operated by CanWest
Corporate Air Charters of Slave Lake, Alberta
Airborne LiDAR collection occurred October 25-November 20th, 2011.
LiDAR data was collected at a flying height of 600 m above ground level and an air
speed of 240 km/h.
Riegl LMS-Q560 laser used pulsed at an approximate rate of 137 kHz resulting in a
computed average of ground spacing equal to 0.70 m
Horizontal Datum: NAD83 (CSRS)
Vertical Datum: CGVD28 orthometric heights (HTv2.0 height transformation model)
Map projection: UTM Zone 13 (Central meridian = 105 degrees west longitude)
Deliverable included:
o 1m bare earth and full feature grids in 1 km x 1 km tiles (ASCII XYZ format)
o 1m bare earth and full feature greyscale hillshades for each project area
(GeofTiff)
o Classified LiDAR point clouds and ASCII extractor program in 1km x1km
tiles (LAS v1.2 format)
o Orthorectified imagery with 0.2m pixel size in1 km x 1 km tiles (GeoTiff and
ECW format) and 1 MrSID image for each project area
o LiDAR contours 0.5m intervals (DWG and Shp format) in 1 km x 1 km tiles
o LiDAR tile Index (ESRI shp format)
o LiDAR survey report
2
2. MATRIX LIDAR SYSTEM
2.1 MATRIX Installation
The MATRIX LiDAR system was installed in a Cessna 185 (C-GAYZ) airplane, as shown below in
Figure 1, owned and operated by CanWest Corporate Air Charters of Slave Lake, Alberta.
Figure 1: Cessna 185 with MATRIX LiDAR system
The Riegl LMS-Q560 scanning laser and inertial measurement unit were mounted on a plate
extending out of the rear baggage hold, as seen in Figure 2. The system computers and data storage
devices were mounted to the floor in the rear of the aircraft, as seen in Figure 3. The GPS antenna
was mounted with a clamp on the front of the right wing next to the fuselage and the operator
controlled the MATRIX system with a monitor and keyboard from the front passenger seat.
Transport Canada has approved the installation of the MATRIX LiDAR system into this survey
aircraft.
Key sensors utilized in the MATRIX installation for the LiDAR survey included:
Riegl LMS-Q560 200 kHz laser scanner and data recorder
NovAtel V-3 dual frequency GPS receiver
NovAtel SPAN LCI 200 Hz Inertial Measurement Unit (IMU)
Canon EOS 1D Mark III, 10 Mega Pixel Digital Camera
3
Figure 2: Laser, Camera and IMU mounted on Cessna 185F
Figure 3: Matrix computers and data storage devices
4
2.2 IMU - GPS Antenna Offset Survey
Several parameters unique to each aircraft LiDAR installation must be determined in order to
produce accurately positioned LiDAR point clouds. These parameters include the three dimensional
vector (lever-arm) between the GPS antenna phase center and the inertial body reference. Using a
total station and prisms at several points surrounding the aircraft, redundant distances and angles to
the IMU unit and GPS antenna were observed. The observations were then subjected to a least-
squares adjustment to compute the final lever arm values. As this particular aircraft had been used
by LSI for LiDAR surveys many times in the past, the GPS to IMU distance had been previously
calculated. A portion of a GPS-IMU offset survey on the aircraft is shown in Figure 4 below.
Figure 4: Cessna 185F lever-arm survey
2.3 IMU – Laser Misalignment
LiDAR calibration passes were made at the beginning and end of each flight to allow for the
determination and verification of the roll, pitch and heading misalignment angles between the IMU
measurement axis and the laser sensor. The calibration passes consisted of three to four flight lines
flown at orthogonal and parallel headings at the project flying height and speed. Features such as
buildings and roads were used to compute and verify the misalignment angles for the project install.
5
3. GPS SURVEY CONTROL
3.1 GPS Control Points
High-precision kinematic GPS solutions were obtained for the LIDAR data collection missions
using differential GPS (DGPS) survey techniques. DGPS requires a static GPS receiver
collecting data at a known ground control point in the vicinity (generally within 35 km) of the
airborne (remote) GPS receiver during LiDAR data collection. All of the collected and
processed LiDAR and imagery data is referenced to the 3D coordinates of the ground control
points.
For consistency and accuracy high order, Federal Geodetic Survey Division (GSD) GPS control
points were used for referencing of all the project sites. LSI surveyors monumented and
increased the density of the existing GSD control networks to allow GPS benchmarks to be
within 35km of the project areas and at local utilized airports. To determine the accurate
positions of the new GPS benchmarks four GPS survey networks were created; Pheasant Creek,
Moosomin, Roughbark and one for the areas surrounding Swift Current including Braddock,
Eastend, Maple Creek and Altawan. Figures of the four control point networks can be seen in
Appendix A.
Below, in Tables 1 and 2 are the control coordinates used for the 2011 AAFC LiDAR survey. Table
1 are the geodetic control coordinates in NAD83 (CSRS) with ellipsoid heights and Table 2 contains
the UTM Zone 13 positions with CGVD28 orthometric heights utilizing the HTv2.0 height
transformation model. Additionally photos of all of the GPS control points can be seen in Appendix
B.
6
Table 1: GPS Control Coordinates NAD83 (CSRS)
ID Monument
Type Latitude Longitude Ellipsoidal Height (m)
HT2.0 Geoid (m)
Moosomin 68s273 GSD 50 12 57.94860 -101 48 47.10640 560.030 21.677
MoosominGPS LSI spike 50 02 55.71764 -101 40 36.17673 520.544 21.543
VirdenAIR LSI spike 49 52 35.96509 -100 55 08.02237 419.151 22.456 Pheasant
Creek
84s275 GSD 50 50 46.3946 -103 49 53.5362 570.040 20.481
PCGPS LSI spike 50 44 13.32691 -103 20 00.71256 561.434 20.741
ReginaAIR LSI spike 50 25 55.85721 -104 39 11.67804 555.748 19.597
Roughbark 94v053 GSD 49 40 43.339 -102 58 38.20920 604.480 19.035
WeyburnAIR LSI spike 49 41 49.63949 -103 48 26.58886 566.354 18.872
WeyGPS LSI spike 49 31 22.10015 -103 44 52.62617 555.544 18.654
SW Sask 80s094 GSD 49 43 37.17512 -108 09 29.59387 854.737 16.531
94v050 GSD 49 59 23.10819 -109 28 05.92757 771.362 17.233
94v051 GSD 50 14 57.87090 -107 46 09.26740 798.870 17.893
A230581 GSD 49 58 39.18304 -110 45 46.26312 706.423 17.143
Admiral LSI spike 49 43 39.86000 -107 54 07.00811 777.775 16.810
Altawan LSI spike 49 13 26.49991 -109 49 03.84398 918.015 15.603
Braddock LSI spike 50 06 22.46381 -107 18 06.31430 744.920 18.142
Cadillac LSI spike 49 46 07.88578 -107 35 08.94909 740.030 17.243
Eastend LSI spike 49 30 34.13721 -108 45 48.76528 896.127 16.016
MapleAIR LSI spike 49 53 49.20037 -109 28 50.21885 750.623 16.926
Medicine Hat LSI spike 50 01 22.50574 -110 43 27.34653 699.783 17.254
Russell LSI spike 49 54 13.19900 -107 30 22.64803 756.087 17.545
Shaunovan LSI spike 49 39 18.09667 -108 24 25.94074 904.853 16.261
USBorder LSI spike 49 00 01.64843 -109 43 57.33403 826.921 15.388
7
Table 2: GPS Control Coordinates UTM Zone 13
ID Easting (m) Northing (m) CGVD28
Elevation (m)
Moosomin 68s273 727350.326 5567518.961 581.706
MoosominGPS 737905.533 5549349.536 542.087
VirdenAIR 793186.537 5532907.521 441.607 Pheasant
Creek
84s275 582260.100 5633374.760 590.520
PCGPS 617591.903 5621907.092 561.434
ReginaAIR 524627.481 5586741.982 575.344
Roughbark
94v053 645921.284 5504872.873 623.515
WeyburnAIR 586005.920 5505638.923 585.226
WeyGPS 590614.525 5486329.411 574.197
SW Sask
80s094 272399.306 5513065.499 871.268
94v050 179757.204 5547064.648 788.594
94v051 302579.154 5570031.036 816.762
A230581 86921.816 5552067.817 723.565
Admiral 290868.468 5512402.603 794.584
Altawan 149262.559 5463539.272 933.618
Braddock 335411.637 5552980.174 763.062
Cadillac 313805.352 5516140.380 757.272
Eastend 227574.018 5490904.144 912.142
MapleAIR 178257.547 5536810.399 767.550
MedicineHat 90073.167 5556894.640 717.036
Russell 320032.520 5530931.881 773.632
Shaunovan 254096.096 5505851.789 921.112
USBorder 153904.113 5438305.786 842.309
For future surveys in the project areas it is important that at least one of the control points listed in
the above tables are used for geo-referencing in order to obtain positions in agreement with the
LiDAR data collected in 2011.
8
4. DATA COLLECTION
The LiDAR survey was completed over 13 flights between October 25th and November 20
th, 2011.
As the project extended late into the fall there were some delays in collection due to snow and as a
result only the areas of Pheasant Creek, Roughbark, Moosomin, Braddock, Maple Creek, Eastend
and Altawan were collected and Admiral, Russell Creek, Cadillac-Gouveneur and Lafleche areas
were not. In total there were7 stand by days with no flight missions due to snow or high winds.
Below in Table 3 is a complete listing of the days each project area was collected.
Table 3: AAFC Fall 2011Flight Summary
Project Area Date Julian Day
Flight Time (GMT) GPS
Pheasant Creek Oct-25 JD298 19:17-21:50 Regina Air, PheasantCreekGPS
Roughbark Oct-27 JD300 16:29-19:14 WeyburnAIR, WeyGPS
Moosomin Oct-28 JD301 19:45-22:41 68s273, MoosominGPS
Moosomin Oct-29 JD302 15:02-18:21 MoosominGPS, VirdenAIR
Braddock Oct-30 JD303 21:04-23:14 94v051, Braddock
Braddock Oct-31 JD304 21:23-23:54 94v051, Braddock
Braddock Nov-01 JD305 19:28-23:13 94v051, Braddock, Russell
Maple Creek Nov-08 JD312 16:59-20:38 94v051, 94v050, MapleAIR
Maple Creek Nov-08 JD312 21:12-23:31 94v051, 94v050, MapleAIR
Eastend Nov-09 JD313 16:15-19:54 Eastend, Shaunovan, MapleAIR
Maple Creek Nov-16 JD320 15:15-19:07 Medicine Hat, A230581, MapleAIR
Maple Creek/ Eastend Nov-16 JD320 20:39-23:31 MapleAIR, Eastend
Altawan Nov-19 JD323 17:42-19:56 MapleAIR, Altawan, US border
Altawan Nov-20 JD324 16:53-21:10 MapleAIR, Altawan
Several airports were used to base collection missions out of, depending on proximity to the project
area. Airports utilized included; Regina International Airport, Weyburn Airport, Virden Airport,
Swift Current Regional Airport, Maple Creek Airport and Medicine Hat Municipal Airport.
9
All flight lines were planned and flown at 600 m above ground level and an approximate speed of
240 km/h. The parallel flight lines were collected with 400 m separation resulting in approximately
40% side overlap of the LiDAR and imagery data. In addition, for each area one or more
perpendicular cross or tie lines were flown for quality control. Below in Figure 5 is an illustration of
the planned and collected flight lines for the Maple Creek project area.
Figure 5: Maple Creek flight lines collected
During data collection the Q560 laser pulsed at 137 kHz with full waveform multi-return capability
resulting in an average point spacing of 0.70 m, or 2.0 points per square metre. The airborne GPS
receiver logged at 1-second intervals simultaneously with the IMU recording the orientation and
accelerations of the sensor plate every 0.005 seconds. Also, downward photos were collected with
the Canon EOS-1D Mark III digital camera every 2.2 seconds for an average of 60% forward
overlap between consecutive photos.
10
5. GROUND CHECK POINTS
To ensure data accuracy and quality assurance of the LiDAR data, multiple ground check point data
verification tests were performed. Independent, high accuracy GPS ground check points were
collected on foot with a pole mounted GPS receiver and antenna as recommended in the ASPRS
Guidelines – Vertical Accuracy Reporting for LiDAR Data V1.0.
Two check point surveys were conducted, one at the Swift Current and one at the Maple Creek
airports, where the majority of flights were based out of. The check points overlaid with the grey-
scale intensity images at the Maple Creek and Swift Current Airports are shown below in Figures 6
and 7. Several calibration passes were performed at the start and end of each flight at these airports.
Ground points were classified from each individual calibration pass, and the resulting triangulated
surface model was compared to the independently-observed ground check points. The resulting
height residuals and statistics for each calibration pass are shown below in Tables 4 and 5.
Figure 6: Check points at Swift Current Airport
11
Figure 6: Check points at Maple Creek Airport
Table 4: Check Point Residuals for Swift Current Airport Checkpoints
Flight Line Average Dz (m) Standard Dev
(m)
RMSE (m)
Accuracy @ 95%
Confidence Interval
(m)
JD305flt1_172 -0.023 0.021 0.031 0.061
JD305flt1_173 -0.027 0.019 0.034 0.065
JD305flt1_174 -0.022 0.021 0.030 0.059
JD305flt1_175 -0.044 0.021 0.049 0.096
JD305flt1_198 0.014 0.023 0.027 0.053
JD305flt1_199 -0.059 0.025 0.065 0.127
JD305flt1_200 -0.007 0.021 0.022 0.042
JD305flt1_201 -0.056 0.021 0.059 0.117
Average -0.0283 0.0213 0.039 0.078
12
Table 5: Check Point Residuals for Maple Creek Airport
Flight Line Average Dz
(m)
Standard Dev (m) RMSE (m)
Accuracy @ 95%
Confidence Interval
(m)
JD312flt1_301 -0.042 0.020 0.046 0.091
JD312flt1_302 0.022 0.032 0.039 0.076
JD312flt1_304 -0.059 0.025 0.064 0.125
JD312flt1_336 0.023 0.015 0.028 0.054
JD312flt1_337 -0.039 0.032 0.051 0.099
JD320flt1_428 0.002 0.017 0.017 0.033
JD320flt1_429 -0.041 0.015 0.043 0.085
JD320flt1_430 0.016 0.014 0.021 0.042
JD320flt1_431 0.029 0.013 0.033 0.064
JD320flt2_432 0.033 0.016 0.036 0.071
JD320flt2_433 -0.006 0.012 0.014 0.026
JD320flt2_434 0.024 0.012 0.027 0.053
JD320flt2_449 -0.003 0.016 0.016 0.031
JD320flt2_450 0.020 0.013 0.024 0.048
JD320flt2_451 -0.009 0.013 0.016 0.031
JD320flt2_452 0.017 0.015 0.023 0.045
JD323flt1_453 0.008 0.017 0.019 0.037
JD323flt1_454 -0.038 0.018 0.042 0.081
JD323flt1_455 -0.098 0.017 0.099 0.194
JD323flt1_456 -0.043 0.019 0.047 0.093
JD323flt1_464 -0.013 0.161 0.021 0.040
JD323flt1_465 0.018 0.017 0.025 0.048
JD323flt1_466 0.036 0.019 0.041 0.079
JD323flt1_467 -0.004 0.015 0.016 0.031
JD324flt1_468 -0.011 0.018 0.021 0.042
JD324flt1_469 0.007 0.019 0.019 0.039
JD324flt1_470 0.084 0.016 0.086 0.168
JD324flt1_471 -0.007 0.021 0.022 0.043
JD324flt1_492 0.009 0.018 0.020 0.039
JD324flt1_493 -0.015 0.019 0.025 0.048
JD324flt1_494 -0.026 0.016 0.030 0.059
JD324flt1_495 0.009 0.021 0.022 0.044
Average -0.003 0.022 0.032 0.064
13
6. DATA PROCESSING AND DELIVERABLES
6.1 LiDAR Point Clouds
6.1.1 LiDAR Tiles
Unclassified point clouds were generated for each individual flight line from the raw laser data, the
GPS-IMU post-processed solutions and the measured system calibration parameters. The point
clouds were then imported into 1 km x 1 km tiles using TerraSolid software so that the average
quantity of points per tile was around 4 million. The name for each tile was derived from the
coordinate of the southwest corner of the tile. The tile naming structure is as follows:
Southwest corner coordinate of tile = (East, North) = (484000, 5269000)
Tile name = EEENNNN = 4845269
A total of 1226 LiDAR tiles were created to cover the seven project areas. A 50 m buffer was added
to the project boundaries and any excess points outside of the buffer were clipped from the project.
The LiDAR tiles were delivered in LAS 1.2 format along with an ASCII extractor program.
6.1.2 Ground Points
An initial automatic ground classification was applied to the tiles. The automatic ground macro
classified ground points using a sequence of steps that identifies the lowest LiDAR point in an area
and then finds neighbouring ground points based on user-specified iteration angles and tolerances.
After the automatic ground classification, trained technicians inspected each tile and either added or
removed points from the Ground class that were incorrectly classified by the automatic ground
macro. This was done using the TerraSolid suite of LiDAR editing tools in the MicroStation
environment, as displayed in Figure 8 below.
14
Figure 8: LiDAR ground editing using TerraSolid software
6.1.3 DTM Key Points
After completion of the manual ground editing, DTM Key Points were classified from the Ground
point class. The automatic DTM Key Point classification selects key points from the Ground class
and chooses neighbouring Ground points using a horizontal tolerance of 10 m and a vertical
tolerance of 10 cm. That is, the maximum horizontal distance between DTM Key Points is 10 m and
the maximum vertical distance is 10 cm.
The DTM Key Points are a subset of the Ground points taken directly from the Ground class. The
DTM Key Point class typically has 40-80% less points than the original Ground class, depending on
the terrain. Because the DTM Key Points are taken from the Ground class, it is important that the
Ground class never be used by itself. Either the DTM Key Point class can be used alone, or the
DTM Key Point and Ground classes can be used together. The DTM Key Point and Ground
classes together will produce the maximum possible terrain detail, with the largest number of points.
15
6.1.4 Vegetation
The points remaining after the ground classification were classified into the Low Vegetation class (0
to 0.25m above ground), and the High Vegetation class (greater than 0.25 m above ground). The
vegetation classes include all objects and structures above the ground, including buildings,
transmission lines, bridges, fences, vehicles and piles of non-earth materials (garbage, wood, etc.).
Because of the large quantity of High Vegetation points, an automatic thinning classification was
performed to reduce the number of points in the High Vegetation class while maintaining the outline
of the forest canopy. The quantity of High Vegetation points was reduced by up to 50% and the
points removed from the High Vegetation class were saved in the Thinned Vegetation class.
6.2 Grid Points
Bare earth grid points were created at a 1 m interval and delivered in ASCII XYZ format using the
same tile structure as the LiDAR tiles. The bare earth grid point elevations were derived from a TIN
surface model of the combined DTM Key Point and Ground classes in the LiDAR point cloud tiles.
It should be noted that the grid point elevations have been interpolated from the LiDAR points and
may contain greater uncertainty depending on the amount of interpolation performed.
Full feature grid points were also created at a 1 m interval and delivered in ASCII XYZ format. The
full feature grid point elevations were derived from the highest point in the High Vegetation class.
At coordinates with no High Vegetation points the elevation of the corresponding bare earth grid
point was applied.
16
6.3 Hillshades
Georeferenced grayscale raster images with a 1 m pixel size were delivered in GeoTIF format. The
bare earth hillshade images were derived from the bare earth grid points and the full feature hillshade
images were derived from the full feature grid points. The hillshades were created using a 315
degree sun azimuth and 45 degree sun angle. A total of 1 bare earth and 1 full feature hillshade in
GeoTiF format were created for each project area. An example of a portion of a bare earth hillshade
is shown below in Figure 9.
Figure 9: Bare earth hillshade
6.4 Orthorectified Imagery
Georeferenced color digital orthophoto mosaics with 20 cm pixel size were delivered in GeoTIF and
ECW formats. The TIF and ECW mosaics were delivered in the same1 km x 1 km tiles as the
LiDAR data, and complete mosaics for each area in MrSID format were also provided.
17
The digital photos were orthorectified using the ground model created from the DTM Key Points.
With orthorectification, only features on the surface of the ground are correctly positioned in the
orthophotos. Objects above the surface of the ground, such as building rooftops and trees, may
contain horizontal displacement due to image parallax experienced when the photos were captured.
This is sometimes apparent along the cut lines between photos. For positioning of above-ground
structures it is recommended to use the LiDAR point clouds for accurate horizontal placement.
6.5 LiDAR Contours
LiDAR contours with an interval of 0.5m were delivered in DWG and ESRI Shape format and the
files were provided in the 1km x 1km tile structure.
The contours were derived using the contour keypoints modeling function in the TerraSolid software
package. Contour keypoints were modeled from the ground class at a maximum vertical distance of
0.5m and a horizontal distance of 20 m. Breaklines were not used around water features therefore a
uniform height of water bodies is not necessarily present if overlapping data was collected on
different days. Major contours were defined every 5m and minor contours every 0.5 m. An
illustration of the contours is shown below in Figure 10.
Figure 10: LiDAR Contours
18
LSI greatly appreciates the opportunity to have performed the LiDAR survey for Agriculture and
Agri-Foods Canada and is available for any questions or comments regarding the survey or the
contents of this report.
LiDAR Services International Inc. Phone: (403) 517-3130
400, 3115 – 12 St. N.E. Fax: (403) 291-5390
Calgary, Alberta T2E 7J2 Website: www.lidarservices.ca
19
APPENDIX A
GPS Networks
Moosomin GPS Network
20
Pheasant Creek GPS Network
21
Roughbark GPS Network
22
South West Saskatchewan GPS Network
23
APPENDIX B
Control Point Photos
68s273
MoosominGPS
24
VirdenAIR
84s275
25
Pheasant Creek GPS
Regina AIR
26
94v053
Weyburn AIR
27
Weyburn GPS
80s094
28
94v050
94v051
29
A230581
Admiral GPS
30
Altawan GPS
Braddock GPS
31
Cadillac GPS
Eastend GPS
32
MapleAIR GPS
MedicineHat GPS
33
Russell GPS
Shaunovan GPS