Hello Everyone,
I am having a difficult time getting my Check Points to line up in the Z axis post processing during accuracy assessment of both DSM and DTM.
I have imagery collected from a Parrot ANAFI Ai at an altitude of 65m (~1cm GSD) with a 75%/75% overlap and side lap. The imagery is of intertidal oyster reefs during low tide so the entire reef is out of the water. I have placed 10-20 GCPs across each of the reefs depending on the are covered. Giving a minimum coverage of ~2 GCPs per 1 hectar. The GCPs XYZ is collected using RTK Rover Receiver with NAD83 Transverse Mercator Zone 17N and a Datum Geoid18. The pattern of the GCPs on the reef create many quincunx patterns to increase links between GCPs and the imagery. I set up the project coordinates with NAD83 (XY) (D2M changes it to Projected Coordinates from Geographic Coordinates resulting in WGS84 UTM Zone 17N) and NAD83 or NAVD88 (Z) there is no Geoid18 option. I then import all GCPs and link 4 images initially then after initial process I link as many images as possible for final processing. I set to Ultra Point Density, 1x image, and high for point matching. When the imagery has completed processing and I export to ArcGIS Pro to conduct a vertical accuracy assessment on both the DSM and DTM using points randomly collected during the same time as the GCPs are collected. I import CPs as a .CSV then Table to Point using the Raster Geographic Coordinates and vertical datum of the imported imagery. Then I extract to point DTM and DSM based on the CPs. The resulting table is then calculated to see what the imagery vertical accuracy is. My results are 40-70cm in the positive no matter what technique or coordinates I use (the above is just one of many setting I have attempted and many reefs (3 reefs in different location on different days) and flight patterns (2D flight pattern and 3D grid flight pattern)). I conducted a vertical check of the GCPs used in the D2M processing in ArcGIS Pro using the same process as the CPs and get ~7cm positive vertical error. Why is there so much vertical error in the imagery and how can I resolve this problem?
I do have the 3D LAS point cloud of each processed data set, but have not tested that yet. Nor do I know the steps to set up the LAS to check vertical accuracy of the model.
The XYZ is very important to this project as we are attempting to measure reef footprint change overtime, volume change overtime, and the roughness of the surface.
Any guidance on finding a solution to the vertical accuracy error would be much appreciated.
Best,
Allix
