Open BaptisteVandecrux opened 2 years ago
zooming in it doesn't look bad, with a constant offset between the reported height and the photogrammetry estimation:
when we have >1 photo per site visit, it led us to find that for some photos, the top of the anemometer was difficult to locate. So, we may rule out those photos (that contribute to outliers). To still get a useful height, one can choose another target like the anemometer axle. Main point: ruling out some photos can increase confidence in the residual sample of heights.
I've plotted the instrument height along with the photogrammetry estimation and it looks rather noisy:
@maikenrevheim is the photogrammetry repo up to date? Can we trust these estimations?