Raster Dataset
Tags
LiDAR, airborne, point cloud, elevation, light detection and ranging, terrain, topography, DEM, DSM
The main goal of the project was to produce LiDAR classified point clouds, Digital Surface Models (DSM), and Digital Elevation Models (DEM) for flood forecasting models or the creation of flood risk maps. These data can support communities with development planning, emergency preparedness, flood response, surface water management and nutrient loading.
The project area covers approximately 7280 square kilometres in the area designated as Whitemud River Watershed, Manitoba, Canada. The LiDAR data acquisition was done in two years (2016- 2017) under favourable atmospheric conditions (low wind, no cloud, smoke or mist between the aircraft and the ground). Also, the ground conditions were free of snow and any unusual flooding or inundation.LiDAR LAS data was provided. In addition to LAS LiDAR data, a Digital Surface Model (DSM) was produced using all first return data from LiDAR mass points such as vegetation and buildings. The DSM has a 1m grid spacing and is tiled in 2km by 2km tiles. A Digital Elevation Model (DEM) was also produced using the Ground and Model key point classes. The DEM also has 1m grid spacing and is tiled in 2km by 2km tiles.
Groupe Info Consult
There are no access and use limitations for this item.
Extent
West | -99.9583733060255 | East | -98.5615280250647 |
North | 50.7668273618685 | South | 49.7239799583674 |
Maximum (zoomed in) | 1:5,000 |
Minimum (zoomed out) | 1:150,000,000 |
LiDAR acquisition parameters: The LiDAR systems used for collection of the point cloud were set to record a minimum of three outbound pulses per square meter in non-overlap areas. These systems recorded intensity and detected multiple discrete returns, with a minimum of four potential returns. The followings show more parameters concerning both systems used. LiDAR acquisition parameters for RIEGL LMS-Q1560: Altitude above ground level (AGL): 1790 m Speed of the plane: 142 knots (73m/sec) Effective Field of View angle: 20° to 26° each side Effective Corridor swath: 1305 m Distance between flight lines: 1044 m Lateral overlap: 20% to 25% Effective repetition rate: 532 kHz Width of overlap zone: 261 m Density in non-overlap zone (pts/m2): 3.62 LiDAR acquisition parameters for ALTM Galaxy Altitude above ground level (AGL): 1477 m Speed of the plane: 140 knots (72m/sec) Effective Field of View angle: 24 ° each side Effective Corridor swath: 1336 m Distance between flight lines: 1069 m Lateral overlap: 20% Effective repetition rate: 400 kHz Width of overlap zone: 267 m Density in non-overlap zone (pts/m2): 3.32 Many aspects of the point cloud were analysed and validated before the classification step. In addition to the density and the accuracy requirements, GPS time was validated to correspond to adjusted GPS time. Concerning the scan angle, although it was originally proposed to limit it at ± 20° from the nadir it was agreed with Manitoba Infrastructure to increase this limit to ± 26,25° (F0V +52.5°). It was also agreed to increase lateral overlap up to 25%. Calibration process: The vertical and horizontal absolute position (inherent) is validated using GPS static and kinematic precision measured in the calibration zone. During this process, ground controls points are measured and this allows to calibrate the LiDAR system. In order to have high precision on GPS positioning during aerial survey, we used GPS bases. Airborne operations used the existing geodetic network in the reference system NAD83-CSRs as GPS base for aerial surveys. A selection was made in order to ensure GPS bases at a distance necessary to obtain good results (about 30km). When a GPS point was not available, we installed our bases on a temporary point which was measured and linked to the Nad83-CSRs network. We also used Can-NET base station network and PPP correction from NRCan. Also, all flight lines lasted a maximum of 20 minutes. This avoid the drift of the inertial platform. Point cloud metadata: The point cloud collected is classified intthese eleven classes following. Class 0 = Unclassified Class 1 = Default: Class 2 = Ground Class 3 = Low vegetation Class 4 = Medium vegetation Class 5 = High vegetation Class 6 = Buildings, structures Class 7 = Low / high points Class 8 = Model key points Class 9 = Water Class 10 = Bridge To carry out cleaning and classification data we used an automatic process coupled with a verification step and manual classification with qualified technicians, based on references control points measured on the ground. Images of laser intensity help to identify many elements such as water, building and bridges. The level 1 of classification required for the project is normally only an automatic classification process realized with TerraScan macros. Although automatic classification was process, manual verification was done all over the project. For example, concerning bridges and water, it’s impossible to use automatic classification. A manual editing was processed to produce theses classes. For the water surface, only permanent water connected to hydrographic network are classified. The proposed minimum area for water classification is 1ha for lakes and more than 30m width for rivers. These standards are from Ministère de l’Énergie et des Ressources du Québec. Concerning classes 2 and 8, an automatic and manual classification were made. A special attention was brought to areas such as cliffs, urban areas, ditches, forest areas, etc. A final validation was also done on all LiDAR data by generating all the tiles color coded raster images and validating visually each of these images in order to detect all detectable abnormalities (absence of points, small density, inconsistency in Z, etc...).
Groupe Info Consult
20 cm. 95% confidence interval
Horizontal accuracy is validated with ground controls across road sections. These profiles across roads are made in different axes to validate accuracy 360°. These profiles are compared to the intensity and the elevation profiles of the point cloud in order to locate any horizontal difference. All the profiles match very well with the point cloud in all directions.
9 cm. 95% confidence interval
Validation of vertical accuracy: A total of 1533 ground control points were measured across the entire project. Most of the points are separated into sectors and they are located in order to make many profiles of 2 crossing roads. These points are measured using RTK and in post processing using 2 GPS receivers double frequency simultaneously. Base receiver are located on existing or densified, localized geodetic points located at less than 10 km from the work, while we move with the last receiver to perform necessary surveys. Some geodetic points located in the area are also used to calculate statistics of vertical accuracy. The vertical difference between these ground controls and laser points representing the ground is calculated. Statistics made out of these differences are showed in the following table. These results show that the vertical accuracy respects the needed precision of ±9 cm. 95% confidence interval. Number of sectors:13 Number of controls:585 Minimum:-0.087 Maximum:0.105 Mean:0.006 Absolute mean: 0.029 68th percentile:0.035 95th percentile:0.078 Rms:0.037 Standard deviation:0.037 95% confidence interval: 0.003