We captured thousands of images using a quadcopter, Survey 3 camera and then created the model with Agisoft Photoscan. Calibration targets on the ground geo-referenced the model. This fly-through is a screen capture from their viewer. The model-making process took around five days on a quad-core 4.4GHz PC. Models will be produced each year to monitor changes but they also help visualise moraine formation. The images captured were about 7mm/pixel resolution.
Here is a quick snapshot of some data from the new Fjalls site:
It is already approaching 2m distance from its original location – also the jitter on the positions is very small.
This is temperature from Rover4’s Peli case, which is the heighest on Breida and is showing good variations (X axis starts in August).
In preparation for deploying more systems in Iceland this summer – I updated our spare Piksi Multis and did a garden test. It performed very well!
I set them up to fix at 10Hz but report every 5 readings – so its closer to our slow system in Iceland. The fix hopped around within about 1.5cm – which is good for my garden as the sky box is not that wide.
These new dGPS units seem to be accurate to around 2cm as shown in our test. This is for a close baseline (and at the moment doesn’t use GLONASS).