- 144 km2 surveyed using only one ground reference survey point
- This Kompsat-3A satellite elevation survey is accurate to 50 cm RMSE, determined by comparison to 6,294 elevation check points from a highly accurate LiDAR survey
- July 2016 Kompsat-3A stereo satellite photos processed by PhotoSat in August 2016
Gerry Mitchell, P. Geo, Geophysicist, President PhotoSat
Summary report below:
A 1m grid of elevation values, covering an area of 144 square kilometres, was produced over a PhotoSat test area in California. The elevation grid was made using geophysical processing of stereo satellite photos taken by the Kompsat-3A satellite. The stereo satellite elevation processing was referenced to one ground survey point. The elevation surveying accuracy was determined by direct comparison to 6,294 elevation check points extracted from a highly accurate LiDAR survey.
Kompsat-3A colour image with 50cm contours from the PhotoSat Kompsat-3A survey of the California test area.
The Garlock Fault was surveyed with a LiDAR in April 2008 by OpenTopography. www.opentopography.org. We believe this LiDAR survey to be accurate to 5cm RMSE.
An elevation image showing the portion of the Opentopography.org LiDAR survey used in this accuracy study. The dimensions of the area are 14.5km east–west by 9.9km north–south. Lower elevations are blue and higher elevations are red. The oval is the Honda California Test Track.
PhotoSat satellite surveying uses high quality stereo satellite photos. These photos are taken by the satellite as it passes over the survey area along a north to south satellite orbit.
The satellite photographs the same ground area within a minute or two. The ground conditions are close to identical on the two photos. The difference in appearance of ground features on the photos is due to the different look directions of the satellite camera.
PhotoSat Geophysical Stereo Satellite Processing System:
Survey coordinates of ground features are determined by measuring the apparent shift in location of the features between the two satellite photos. PhotoSat uses a proprietary geophysical processing system to generate survey coordinates from stereo satellite photos. This system is described in a PhotoSat white paper published at a 2010 ASPRS conference.
Stereo Kompsat-3A elevation image covering the area of the LiDAR survey. This elevation grid has an elevation point every meter. Low elevations are blue and high elevations are red. The rectangular feature south of the track and the smaller area to the west are fields of solar panels installed since 2014.
Elevation checkpoints for areas with slopes of less than 20% grade.
California elevation grid:
PhotoSat produced a 1m grid of elevations over the entire project area. We processed the stereo Kompsat-3A photos with our geophysical stereo satellite processing system in August 2016. An image of the 1m elevation grid is shown above.
With this accuracy study we are demonstrating that we can produce accurate Kompsat-3A surveys with only one ground reference point. Having as few as one ground reference survey point for a 144 km2 satellite surveying project is not uncommon in mining exploration projects.
Accuracy evaluation check points:
The accuracy of the PhotoSat 1m survey grid was evaluated with 6,294 elevation checkpoints derived from the LiDAR elevation grid.
Elevation survey accuracy statistics:
The Guidelines for Digital Elevation Data of the US National Digital Elevation Program (NDEP) recommends that elevation checkpoints should be chosen in areas with slopes less than 20% grade. The 6,294 elevation checkpoints with slopes less than 20% grade have an RMSE of 50 cm and LE90 of 85cm.
Histogram of the elevation differences between the Kompsat-3A stereo satellite elevations for the 14.5 km by 9.9 km area and the 6,294 elevation checkpoints with slopes less than 20% grade. The Guidelines for Digital Elevation Data of the US National Digital Elevation Program (NDEP) recommends that elevation checkpoints should be chosen in areas with slopes less than 20% grade. RMSE 50 cm, LE90 85 cm.