The point of interest interesting point is in the visual demonstration: aerial imagery is a mixed bag of varying age and quality, hard to trust without proper assessment. (2.7 m is actually pretty bad in a rather flat, densely-surveyed (sub)urban area with abundant landmarks at hand - many a national mapping agency would deem it unacceptable even for remote mountain areas, whereby they reach sub-meter accuracy.)
Yeah, the question that starts off the blog post is
I am often asked by clients why their field data collected with a highly accurate GNSS receiver doesn’t line-up the way they expect with their aerial imagery?
Clearly it’s geared toward people using equipment and data with sub-meter accuracy, which is probably not the case with most Google Earth imagery, and certainly not with most iNatters’ observation equipment. This issue seems like a different order of magnitude.
Certainly! Google Maps or OSM with all their flaws is better than no geolocation, or a misused/misconfigured device. However, one would be wrong to assume that aerial imagery equals ground truth or has any defined accuracy, until proven so with e.g. a modern smartphone or by cross-checking with hi-quality imagery. Both able to reach meter-level (HH1) accuracy nowadays.
yup. there are distinct ridges or elevation changes to the west, south and east of the area of interest here. it looks like things line up fine once you get to these ridges. so it seems like maybe they didn’t have an accurate / granular enough elevation model to map the aerial images against.
I agree with @jdmore - I’ve also seen instance in Google Maps and Earth where roads and other items are offset due to datum shifts. Somebody traced/digitized maps from images and maps using one datum, and then it gets projected in Googler mostly using WGS1984. If you also work in GIS is this is one of the core issues impacting data digitization and mapping when combining various sources of data…
And depending on your GPS, there could also be introduced errors - e.g. Garmin and other devices often have “remain on road” constraints that slightly change your displayed location to “fit” on their internal map. But most modern GPS apps and devices allow for saving waypoints and markers in the “native” format and not “correct” for device maps, so trusting your app based GPS is often the most accurate location. And are iNaturalist observations really so much worse if locations are off by a few meters/feet (even 50) - it’s not as if most consumer devices are accurate to within less than 10ft/3m in any case :-)
And orthorectifying aerial photos are also affected by flight height and elevation data availability - incorrect DEM might have been used, or none at all, impacting the accuracy. There’s a whole different set of science behind this that all impacts how data is displayed. If you want to track sub meter accuracy of observations then iNaturalist or any solution that rely on Google maps or OSM data is going to be problematic.