I want to estimate glacier surface lowering using UAV data acquired on glacier frontal areas where it is too difficult and dangerous to measure GCPs.
Yes Hayder Dibs. But what I am concerned about, is how to coregister properly (horizontally and vertically) the DEMs, without very accurate GCPs, so that we can profit of the high resolution DEMs (30 cm in this case) that can be generated with the pictures we acquired. I simple solution I found was to generated a mosaic and a DEM with data from the first survey, using the GPS information of the drone for positioning, and the coastline to give a relative altitudinal reference. Then, I used the resulting mosaic and DEM to get GCPs for processing aerial pictures of a second survey. The altitudinal differences between both DEMs were very low in non-changing areas (< 0.4 m) compared with the changes on the glacier surface for the period (> 3.0 m). But anyway, I was wondering if the photogrammetrists have other ideas about how to proceed in this case...
Hi, if you want to simply compare two surveys with high accuracy you can coregister the point-clouds with manually picked control points. You need to find and mark 4 or more points on both clouds where you are sure they are not moving, then move and scale them together. This can be done either in SfM-software (Metashape or Pix4D) or in external freeware (e.g. CloudCompare, but here you need to export your model as a .las in UTM coordinates). This way you align the models very accurately and can even get an RMS from the point picking. If you publish like this you need to mention that the absolute geopositioning for both models can be rough (I assume geotagged by the UAV GPS), but relative to each other they are accurate and thus flow speed or volume change measurements are possible.
Jorge, you may be interested in this article by Cook and Dietze (2019)
https://www.earth-surf-dynam-discuss.net/esurf-2019-27/esurf-2019-27.pdf
They propose a system to compare UAV derived point clouds without ground control essentially by using the sparse point cloud from Survey 1 as control for Survey 2. As photogrammetric datasets can be subject to distortion, this essential would give the same distortion to both datasets, so change detection results are more representative of process than reconstruction error.
As mentioned by Edgar, automated or pick-point registration can be a nice approach too, but since you are working with a glacier where surfaces could potentially retreat at a similar rate, you might overfit your data and estimate lower retreats than actually happened. To get around that, you could focus on bounding rocky sections of the valley to get your translation matrix (the geometric matrix used to move your data to the best fit in CloudCompare) and apply the best translation to your entire dataset for change detection.
Thank you Edgar and Joshua for your inputs!
My experience so far was using mosaic and DEM generated with data from survey 1 to get GCPs for a relative coregistering when generating the DEM of the survey 2. Of course, selecting GCPs in rocky areas outside the glacier.
But I have not compared yet point clouds for estimating vertical changes on the glacier surface. The method proposed in the paper is very simple and seems to be effective for what I want.
Will keep you posted on the progress.
Cheers!
Flight design is critical. Make sure proper flight geometry (altitude, overlap) and use combined nadir and oblique images to reduce systematic errors during primary image alignment. Set the first high-resolution DEM and assume it as a control surface (ground-truth) of the subsequent "medium" or desired resolution DEMs for your purposes.
Hi Jairo Escobar Villanueva , do you mean using the first high-resolution DEM and mosaic as a reference for the second survey? Or as a reference for a second image alignment of the same dataset?
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing